Page 1
TYPE-2 FUZZY LOGIC CONTROLLER BAESD MAXIMUM
POWER POINT TRACKING ALGORITHM FOR SOLAR
PHOTOVOLTAIC APPLICATION
A Thesis
Submitted in partial fulfilment of the requirement for the
Degree of
Master of Technology in Intelligent Automation and Robotics
Jadavpur University
May 2014
By
Abhishek Pandit
Registration No: 117000 of 2011-12 Examination Roll No: M6IAR14-14
Under the Guidance of
Prof. Amit Konar
Department of Electronics & Telecommunication Engineering Jadavpur University, Kolkata-700032
India
Page 2
2
FACULTY OF ENGINEERING AND TECHNOLOGY
JADAVPUR UNIVERSITY
CERTIFICATE
This is to certify that the dissertation entitled “Type-2 Fuzzy Controller Based Maximum
Power Point Tracking Algorithm for Solar Photovoltaic Application” has been carried
out by ABHISHEK PANDIT (University Registration No. : 117000 of 2011-12) under my
guidance and supervision and be accepted in partial fulfilment of the requirement for the
degree of Master of Technology in Intelligent Automation and Robotics. The research results
presented in the thesis have not been included in any other paper submitted for the award of
any other University or Institute.
-------------------------------------------------
Prof. Amit Konar Supervisor Dept. of Electronics and Telecommunication Engineering, Jadavpur University -------------------------------------------------
Prof. Iti Saha Misra Head of the Department Dept. of Electronics and Telecommunication Engineering, Jadavpur University
-------------------------------------------------
Prof. Sivaji Bandyopadhaya Dean Faculty of Engineering and Technology Jadavpur University
Page 3
3
FACULTY OF ENGINEERING AND TECHNOLOGY
JADAVPUR UNIVERSITY
CERTIFICATE OF APPROVAL*
The foregoing thesis is here by approved as a creditable study of an engineering subject and
presented in a manner satisfactory to warrant acceptance as pre-requisite to the degree for
which it has been submitted. It is understood that by this approval the undersigned do not
necessary endorse or approve any statement made, opinion expressed or conclusion drawn
there in but approve the thesis only for which it is submitted.
Committee on final examination
For the evaluation of the Thesis
------------------------------------------
Signature of the Examiner
------------------------------------------
Signature of the Supervisor
* Only in the case the thesis is approved.
Page 4
4
FACULTY OF ENGINEERING AND TECHNOLOGY
JADAVPUR UNIVERSITY
DECLARATION OF ORIGINALITY AND COMPLIANCE OF ACADEMIC THESIS
I here declare that this thesis titled “Type-2 Fuzzy Controller Based Maximum Power Point Tracking Algorithm for Solar Photovoltaic Application” contains literature survey and original research work by the undersigned candidate, as part of her Degree of Master of Technology in Intelligent Automation and Robotics.
All in information have been obtained and presented in accordance with academic rules and ethical conduct.
I also declare that, as required by these rules and conduct, I have fully cited and reference all materials and results that are not original to this work.
Name: Abhishek Pandit
Examination Roll No: M6IAR14-14
Thesis Title: Type-2 Fuzzy Controlled MPPT Algorithm for Solar Photovoltaic Application
------------------------------------------
Signature of the Candidate
Page 5
5
ACKNOWLEDGEMENTS
This thesis has been conducted at the Intelligent Automation and Robotics unit in the
Department of Electronics and Tele-communication of the Faculty of Engineering and
Technology of Jadavpur University. I would like to acknowledge all the people and institutions
that have contributed directly and indirectly in this work.
First of all, I would like to express my gratitude to my supervisor, Prof. (Dr.) Amit Konar Head
Intelligent Automation and Robotics unit for giving me the opportunity of working under his
supervision. To Mr. Sumantra Chakraborty, instructor of this thesis: it would have not been
possible to complete this work without his invaluable guidance, advice and support. I have
learnt a lot from him during the realization of this thesis.
I am also grateful to Dr. Swati Purakayastha, Managing Director of Optimal Power Synergy
India Pvt. Ltd. and Mr. Ashok Prakash, Chief Technology Officer of Optimal Power Solutions.
Without their active support and encouragement, I could not able to complete my thesis.
Also I would like to express thank to all my fellow colleges in Optimal Power Synergy
India and my University departmental mates.
Finally I would like to dedicate this thesis to my parents and wife, who have encouraged,
motivated and supported me in all odds and to achieve my desired goal in life.
Page 6
6
ABSTRACT
Solar panels have a nonlinear voltage-current characteristic, with a distinct maximum power
point (MPP), which depends on the environmental factors, such as temperature and
irradiation. In order to continuously harvest maximum power from the solar panels, they have
to operate at their MPP despite the inevitable changes in the environment. This is why the
controllers of all solar power electronic converters employ some method for maximum power
point tracking (MPPT). Over the past decades many MPPT techniques have been published.
The three algorithms that where found most suitable for large and medium size photovoltaic
(PV) applications are perturb and observe (P&O), incremental conductance (InCond) and
fuzzy logic control (FLC) as on today. The first objective of this thesis is to study and analyze
them and later come up with a new type of MPPT technique using Interval Type-2 Fuzzy
control system. These were compared and tested dynamically according a recently issued
standard. This new technique is better in terms of noise reduction over Type-1 Fussy
Controlled MPPT, which overcomes their poor performance when the irradiation changes
continuously.
The dynamic MPPT efficiency tests require long simulations and if detailed models of
the power converter are used they can take a lot of memory and computation time. To
overcome this challenge a simplified model of the PV system was developed. This model was
validated with simulations.
Page 7
7
CONTENTS
CERTIFICATE .................................................................................................................................................... 2
CERTIFICATE OF APPROVAL* ................................................................................................................... 3
DECLARATION OF ORIGINALITY AND COMPLIANCE OF ACADEMIC THESIS ....................... 4
ACKNOWLEDGEMENTS ................................................................................................................................ 5
ABSTRACT .......................................................................................................................................................... 6
CONTENTS .......................................................................................................................................................... 7
TABLE OF FIGURES ........................................................................................................................................10
CHAPTER 1: INTRODUCTION.......................................................................................................................12
MOTIVATION .................................................................................................................................................14
PROBLEM STATEMENT ...............................................................................................................................15
CONTRIBUTIONS ...........................................................................................................................................15
THESIS OUTLINE ...........................................................................................................................................15
CHAPTER 2: SOLAR CELL AND PHOTOVOLTAIC SYSTEM CONFIGURATION ............................16
SOLAR CELL ...................................................................................................................................................16
OPERATING PRINCIPLE ...................................................................................................................................16
EQUIVALENT CIRCUIT OF A SOLAR CELL ..................................................................................................18
OPEN CIRCUIT VOLTAGE, SHORT CIRCUIT CURRENT AND MAXIMUM POWER POINT ..................19
FILL FACTOR ......................................................................................................................................................20
TEMPERATURE AND IRRADIANCE EFFECTS .............................................................................................20
TYPES OF SOLAR CELLS .............................................................................................................................23
MONO-CRYSTALLINE SILICON ......................................................................................................................23
POLYCRYSTALLINE SILICON .........................................................................................................................23
AMORPHOUS AND THIN-FILM SILICON.......................................................................................................24
OTHER CELLS AND MATERIALS ...................................................................................................................24
PHOTOVOLTAIC MODULES ........................................................................................................................26
PHOTOVOLTAIC SYSTEM CONFIGURATION ..........................................................................................27
CENTRAL INVERTER ........................................................................................................................................28
STRING INVERTER ............................................................................................................................................28
Page 8
8
MULTI-STRING INVERTER ..............................................................................................................................29
MODULE INTEGRATED INVERTER ...............................................................................................................29
CHAPTER 3: MAXIMUM POWER POINT TRACKING.............................................................................31
HILL-CLIMBING TECHNIQUES ...................................................................................................................31
PERTURB AND OBSERVE ................................................................................................................................31
INCREMENTAL CONDUCTANCE....................................................................................................................33
FUZZY LOGIC CONTROL .............................................................................................................................37
MAXIMUM POWER POINT TRACKING SUMMARY................................................................................40
CHAPTER 4: DC-DC CONVERTERS .............................................................................................................41
IDENTIFICATION OF SUITABLE CONVERTER FOR MPPT ....................................................................42
BUCK CONVERTER ...........................................................................................................................................43
BOOST CONVERTER .........................................................................................................................................44
BUCK-BOOST CONVERTER .............................................................................................................................44
CONCLUSION .....................................................................................................................................................45
CHAPTER 5: FUNDAMENTAL OF TYPE- 1 FUZZY SETS AND SYSTEMS AND APPLICATION IN MPPT CONTROL ...............................................................................................................................................46
TYPE-1 FUZZY SETS AND MEMBERSHIP FUNCTION ............................................................................47
FUZZY RULES AND INFERENCE ....................................................................................................................48
FUZZY CONTROLLER STRUCTURE ...............................................................................................................48
APPLICATION OF TYPE-1 FUZZY LOGIC IN MPPT CONTROL .................................................................49
DESIGN OF FUZZY LOGIC CONTROLLER PARAMETERS .....................................................................51
CONTROLLER STRUCTURE .............................................................................................................................51
MEMBERSHIP FUNCTIONS ..............................................................................................................................52
SCALING FACTORS ...........................................................................................................................................53
DERIVATION OF CONTROL RULES ...............................................................................................................53
DECISION MAKING ...........................................................................................................................................56
DEFUZZIFICATION ............................................................................................................................................58
SIMULATION OF TYPE-1 FUZZY CONTROL FOR MPPT ............................................................................60
CHAPTER 6: TYPE-2 FUZZY LOGIC AND DESIGN OF CONTROLLER FOR MPPT ALGORITHM ...............................................................................................................................................................................62
TYPE-2 FUZZY SETS ..........................................................................................................................................63
Page 9
9
INTERVAL TYPE-2 FUZZY SETS .....................................................................................................................64
TYPE-2 FUZZY LOGIC SYSTEM BLOCK DIAGRAM ....................................................................................65
TYPE2 FUZZY LOGIC CONTROLLER.........................................................................................................66
FUZZIFICATION AND THE RULES .................................................................................................................68
INFERENCE .........................................................................................................................................................68
TYPE-REDUCER AND DEFUZZIFICATION....................................................................................................69
SIMULATION OF TYPE-2 MPPT FUZZY CONTROLLER ..............................................................................71
CHAPTER7: CONCLUSION AND FUTURE WORK ...................................................................................73
CONCLUSION .....................................................................................................................................................73
FUTURE WORK ..................................................................................................................................................74
REFERENCE ......................................................................................................................................................75
APPENDICES ......................................................................................................................................................77
A-1 MATLAB CODE FOR TYPE-1 FUZZY MPPT SYSTEMS ........................................................................77
A-2 MATLAB CODE FOR TYPE-2 FUZZY MPPT SYSTEMS ........................................................................84
Page 10
10
TABLE OF FIGURES
Figure 1-Structure of Solar Cell ............................................................................................................ 17
Figure 2-Equivalent Circuit of Solar Cell ............................................................................................. 18
Figure 3-P-V-I Curve of Solar Cell ....................................................................................................... 19
Figure 4-V-I and V-P curves at constant temperature (25°C) and three different insolation values .... 21
Figure 5-V-I and V-P curves at constant irradiation (1 kW/m2) and three different temperatures ...... 22
Figure 6-Typical construction of PV Module ....................................................................................... 26
Figure 7-MPPT Block Diagram ............................................................................................................ 27
Figure 8-Arrangement for Central Inverter ........................................................................................... 28
Figure 9-Arrangement for String Inverter ............................................................................................. 29
Figure 10-Arrangement for Multi String Inverter ................................................................................. 29
Figure 11-Arrangement for Module Inverter ........................................................................................ 30
Figure 12-V-I-P of Solar Array ............................................................................................................. 32
Figure 13-Flow Chart of P&O algorithm .............................................................................................. 33
Figure 14-Flow Chart of InCond Algorithm ......................................................................................... 34
Figure 15-MPP at different radiation level ........................................................................................... 35
Figure 16-Membership function for MPPT .......................................................................................... 38
Figure 17-Flow Chart of FLC based MPPT algorithm ......................................................................... 39
Figure 18-Block Diagram of DC-DC converter ................................................................................... 41
Figure 19-Duty Ration of DC-DC converter ........................................................................................ 42
Figure 20-Behavioural curve of MPP for different converter operation ............................................... 43
Figure 21-Circuit of a buck-boost converter ......................................................................................... 45
Figure 22-A Gaussian type-1 fuzzy membership function ................................................................... 48
Figure 23-Basic configuration of a fuzzy logic controller .................................................................... 49
Figure 24-Fuzzy control scheme for a maximum power point tracker ................................................. 51
Figure 25-Functional block of the fuzzy controller .............................................................................. 52
Figure 26-Membership functions for (a) change in power (b) change in duty cycle ............................ 53
Page 11
11
Figure 27-Quantization effect during maximum power search ............................................................ 56
Figure 28-Fuzzy inference and defuzzification using Mamdani method .............................................. 59
Figure 29-Fuzzy Membership function for Power ................................................................................ 60
Figure 30-Fuzzy Membership function of output ................................................................................. 60
Figure 31- Final output of Duty Cycle Signal ....................................................................................... 61
Figure 32-A Gaussian type-2 fuzzy membership function (FOU) ........................................................ 64
Figure 33-T2FLS block diagram........................................................................................................... 66
Figure 34-Structure of a type2 FLC ...................................................................................................... 67
Figure 35-Membership function of input .............................................................................................. 67
Figure 36-Membership function of output ............................................................................................ 68
Figure 37- Fuzzy Membership function of Power ................................................................................ 71
Figure 38-Fuzzy Membership function of Voltage ............................................................................... 71
Figure 39-Fuzzy Membership function of Duty cycle .......................................................................... 72
Figure 40-Final output of Duty Cycle Signal in Type-2 Fuzzy Controller ........................................... 72
Page 12
12
CHAPTER 1: INTRODUCTION
Energy has been one of the most important driving forces in our fast growing world. Nations
are in an endless search for new energy sources. This search passed through many
revolutions, from the use of fire and coal to the discovery of fossil fuels such as oil and
natural gas. However, this search was faced by many challenges, since many of these
resources are expensive, destructive to the environment, or ceasing to exist in the near future.
Significant progress has been made over the last few years in the research and development of
renewable energy systems such as wind, sea wave and solar energy systems. Among these
resources, solar energy is considered nowadays as one of the most reliable, daily available,
and environment friendly renewable energy source.
The amount of energy in the sunlight reaching the earth’s surface is equivalent to around
10,000 times the world’s energy requirements. Consequently, only 0.01 percent of the energy
in sunlight would need to be harnessed to cover mankind total energy needs. Given that the
sun shines on this part of the globe nearly the entire year, many new projects are intended to
make use of solar energy as a backup source to the existing power system.
Another incentive of the spread of such projects is the fact that the cost trend of photovoltaic
(PV) systems is descending, while the fuel price is ascending. The reason behind this trend is
the growing mass production and market of PV systems. The advancement in solar systems is
based on the new technological advances in the industries of photovoltaic cells, power
electronics switches, microcontrollers, and computer-based simulation packages. Solar
systems not only can serve as a backup to existing energy systems, but they can also be easily
integrated with large grid systems, or can be used as standalone systems serving as
independent energy sources.
India is a country that has tremendous solar energy potential. As the nation is facing an
increasing demand - supply gap in energy, it is important to tap the solar potential to meet the
energy needs. This article analyzes the Indian Solar industry, its major growth drivers, the
challenges it faces and the various policy initiatives taken by the government. The article also
tries to identify the various actions required to promote the growth and development of the
industry, enabling India to meet the rising energy demands of the future.
India is in a state of perennial energy shortage with a demand-supply gap of almost 12% of
the total energy demand. This trend is significant in the electricity segment that is heavily
Page 13
13
dependent on coal and other non-renewable sources of energy. Renewable energy (RE)
sources contribute only 7.7% of the total installed power capacity of 167,077 MW in India.
Among the RE sources, wind power is the dominating component while solar energy
currently contributes to less than 0.1% (on-grid+ off-grid) of the total installed capacity.
The solar energy potential in India is immense due to its convenient location near the
Equator. India receives nearly 3000 hours of sunshine every year, which is equivalent to 5000
trillion kWh of energy. India can generate over 1,900 billion units of solar power annually,
which is enough to service the entire annual power demand even in 2030 (estimates).
Rajasthan and Gujarat are the regions with maximum solar energy potential. This, coupled
with the availability of barren land, increases the feasibility of solar energy systems in these
regions. Considering India’s solar potential, the government has rolled out various policies
and subsidy schemes to encourage growth of the Solar Industry, which is expected to
experience exponential growth in the coming years.
There are three government bodies established to promote solar energy in India. The first is
the Ministry of New and Renewable Energy (MNRE), which is the nodal unit for all matters
relating to RE. The second, India Renewable Energy Development Agency (IREDA), is a
public limited company established in 1987 to promote, develop and extend financial
assistance for RE and energy efficiency/conservation projects. Finally, Solar Energy Centre
(SEC) is a dedicated unit of the MNRE and the Government for the development of solar
energy technologies and promotion of its applications through product development. Besides
this, government has also rolled out various policies and subsidies to promote this sector.
India’s National Action Plan on climate change (NAPCC) identifies eight critical missions to
promote climate mitigation and adaptation. National Solar Mission, which has the specific
goal of increasing the usage of solar thermal technologies in urban areas, industry, and
commercial establishments, is one of the core components of this policy. The government
also offers capital subsidies to semiconductor manufacturing plants in Special Economic
Zone (SEZs) and outside SEZs through semiconductor policy launched in 2007. In 2009,
MNRE launched “Jawaharlal Nehru National Solar Mission (JNNSM)” with the ambitious
goal of making India a global leader in solar energy. JNNSM plans a three-phase approach
with specific targets for each phase. The other targets of this mission include achieving grid
parity (same production cost as current electricity source) by 2022 and parity with coal based
power generation in 2030.
Page 14
14
One of the recent applications of such intelligent controllers is the efficiency optimization of
photovoltaic energy systems. These techniques are used to maximize the power output of solar
systems as well as improve their robustness against external disturbances.
MOTIVATION
If the parameters of a system can be obtained precisely, then its control would be a relatively
straightforward problem and model-based approaches such as PID and pole placement could
be used. However, in real-time industrial systems, it is often the case that there exist
considerable difficulties in obtaining an accurate model. Even when the model is sufficiently
accurate, there are many other uncertainties for example due to the precision of the sensors,
noise produced by the sensors, environmental conditions of the sensors, and nonlinear
characteristics of the actuators. Then, not only does the performance of the model-based
approaches drastically decrease, but the complexity of the controller design also increases. In
such cases, model-free approaches are generally preferred both for modelling and control
purposes. The most common model-free approaches are the use of fuzzy logic systems
(FLSs).
The concept of a type-2 fuzzy set was introduced by Zadeh as an extension of the concept of
an ordinary fuzzy set (henceforth called a type-1 fuzzy set). Such sets are fuzzy sets whose
membership grades themselves are type-1 fuzzy sets; they are very useful in circumstances
where it is difficult to determine an exact membership function for a fuzzy set; hence, they
are useful for incorporating linguistic uncertainties, e.g., the words that are used in linguistic
knowledge can mean different things to different people. A fuzzy relation of higher type (e.g.,
type-2) has been regarded as one way to increase the fuzziness of a relation, and, according to
Hisdal, “increased fuzziness in a description means increased ability to handle inexact
information in a logically correct manner. According to John, “Type-2 fuzzy sets allow for
Linguistic grades of membership, thus assisting in knowledge representation, and they also
offer improvement on inference with type-1 sets.
Type-2 sets can be used to convey the uncertainties in membership functions of type-1 sets, due to
the dependence of the membership functions on available linguistic and numerical information.
Linguistic information (e.g., rules from experts), in general, does not give any information about
the shapes of the membership functions. When membership functions are determined or tuned
based on numerical data, the uncertainty in the numerical data, e.g., noise, translates into
uncertainty in the membership functions.
Page 15
15
Out the many MPPT algorithm Type-1 Fuzzy one of the proven algorithm and there are many
papers have been published. But it will be the first approach to apply Type-2 Fuzzy logic in MPPT
algorithm. Here we will the Interval Type-2 Fuzzy controller to design the MPPT.
PROBLEM STATEMENT
In line with the global initiative toward the design and development of PV systems as alternative
energy sources, this work will investigate the potential of using solar energy systems in the India.
One of the drawbacks of PV systems is their low efficiencies compared to their cost. In order to
overcome these drawbacks, maximum power should be extracted from these systems. The aim is
to develop an efficient standalone photovoltaic system. This system employs a new digital control
scheme using fuzzy-logic and a dual maximum power point tracking (MPPT) controller. The
MPPT algorithm controls the power converter between the PV panel and the load and implements
a new Type-2 fuzzy-logic (FLC) based MPPT control scheme to keep the system power operating
point at its maximum. Such systems are exceedingly demanded in remote areas where it is difficult
to connect to the grid system.
CONTRIBUTIONS
This work will explore the effectiveness of intelligent and digital control techniques for PV system
efficiency optimization. These techniques combine both physical as well as Type-2 fuzzy-based
MPPT tracking techniques. Furthermore, this work will offer a stability analysis for the
Type-2 FLC based MPPT controller. Though literature is rich with PV systems implementations,
no work was based on actual outdoor testing in this part of the world. This work will use
experimental data to investigate the potential of solar energy in India and the effects of the harsh
environment on PV systems efficiencies.
THESIS OUTLINE
In this thesis, a detailed literature review about tracking techniques will be presented in Chapter 3.
Chapter 2 will focus on PV systems modelling and simulation featuring a MATLAB/Simulink full
simulation of the overall standalone PV system. Chapter 5 will discuss the design and
implementation of type-1 fuzzy logic controller systems. The proposed Type-2 FLC controller
will be fully investigated in chapter 6 highlighting a simulation comparison between the
proposed controller and the conventional ones. Chapter 6 will verify the proposed technique
effectiveness through presenting the PV system hardware implementation and trial results. Finally,
chapter 7 will conclude the work discussing the work limitations along with the future work.
Page 16
16
CHAPTER 2: SOLAR CELL AND PHOTOVOLTAIC SYSTEM
CONFIGURATION
SOLAR CELL
OPERATING PRINCIPLE
Solar cells are the basic components of photovoltaic panels. Most are made from silicon even
though other materials are also used.
Solar cells take advantage of the photoelectric effect: the ability of some
semiconductors to convert electromagnetic radiation directly into electrical current. The
charged particles generated by the incident radiation are separated conveniently to
create an electrical current by an appropriate design of the structure of the solar cell, as
will be explained in brief below.
A solar cell is basically a p-n junction which is made from two different layers of
silicon doped with a small quantity of impurity atoms: in the case of the n-layer, atoms
with one more valence electron, called donors, and in the case of the p-layer, with one
less valence electron, known as acceptors. When the two layers are joined together, near
the interface the free electrons of the n-layer are diffused in the p-side, leaving behind
an area positively charged by the donors. Similarly, the free holes in the p-layer are
diffused in the n-side, leaving behind a region negatively charged by the acceptors. This
creates an electrical field between the two sides that is a potential barrier to further flow.
The equilibrium is reached in the junction when the electrons and holes cannot surpass
that potential barrier and consequently they cannot move. This electric field pulls the
electrons and holes in opposite directions so the current can flow in one way only:
electrons can move from the p-side to the n-side and the holes in the opposite direction.
A diagram of the p-n junction showing the effect of the mentioned electric field is
illustrated in Figure 1.
Page 17
17
Figure 1-Structure of Solar Cell
Metallic contacts are added at both sides to collect the electrons and holes so the current
can flow. In the case of the n-layer, which is facing the solar irradiance, the contacts are
several metallic strips, as they must allow the light to pass to the solar cell, called
fingers.
The structure of the solar cell has been described so far and the operating principle is
next. The photons of the solar radiation shine on the cell. Three different cases can
happen: some of the photons are reflected from the top surface of the cell and metal
fingers. Those that are not reflected penetrate in the substrate. Some of them, usually the
ones with less energy, pass through the cell without causing any effect. Only those with
energy level above the band gap of the silicon can create an electron-hole pair. These
pairs are generated at both sides of the p-n junction. The minority charges (electrons in
the p-side, holes in the n-side) are diffused to the junction and swept away in opposite
directions (electrons towards the n-side, holes towards the p-side) by the electric field,
generating a current in the cell, which is collected by the metal contacts at both sides.
This can be seen in the figure above, Figure-1. This is the light-generated current which
depends directly on the irradiation: if it is higher, then it contains more photons with enough
energy to create more electron-hole pairs and consequently more current is generated by
the solar cell.
Page 18
18
EQUIVALENT CIRCUIT OF A SOLAR CELL
The solar cell can be represented by the electrical model shown in Figure-2. Its current-
voltage characteristic is expressed by the following equation (1):
(1)
where I and V are the solar cell output current and voltage respectively, I0 is the dark
saturation current, q is the charge of an electron, A is the diode quality factor, k is the
Boltzmann constant, T is the absolute temperature and RS and RSH are the series and shunt
resistances of the solar cell. RS is the resistance offered by the contacts and the bulk
semiconductor material of the solar cell. The origin of the shunt resistance RSH is more
difficult to explain. It is related to the non ideal nature of the p-n junction and the presence of
impurities near the edges of the cell that provide a short-circuit path around the junction. In
an ideal case RS would be zero and RSH infinite. However, this ideal scenario is not possible
and manufacturers try to minimize the effect of both resistances to improve their products.
Figure 2-Equivalent Circuit of Solar Cell
Sometimes, to simplify the model, the effect of the shunt resistance is not considered, i.e.
RSH is infinite, so the last term is neglected.
A PV panel is composed of many solar cells, which are connected in series and parallel
so the output current and voltage of the PV panel are high enough to the requirements of
the grid or equipment. Taking into account the simplification mentioned above, the
output current-voltage characteristic of a PV panel is expressed by equation (2), where
np and ns are the number of solar cells in parallel and series respectively.
Page 19
19
(2)
OPEN CIRCUIT VOLTAGE, SHORT CIRCUIT CURRENT AND MAXIMUM POWER
POINT
Two important points of the current-voltage characteristic must be pointed out: the open
circuit voltage VOC and the short circuit current ISC. At both points the power generated
is zero. VOC can be approximated from (1) when the output current of the cell is zero, i.e. I=0
and the shunt resistance RSH is neglected. It is represented by equation (3). The short circuit
current ISC is the current at V = 0 and is approximately equal to the light generated current IL
as shown in equation (4).
(3)
(4)
The maximum power is generated by the solar cell at a point of the current-voltage
characteristic where the product of VI is Maximum. This point is known as the MPP and is
unique, as can be seen in Figure 3, where the previous points are represented.
Figure 3-P-V-I Curve of Solar Cell
Page 20
20
FILL FACTOR
Using the MPP current and voltage, IMPP and VMPP, the open circuit voltage (VOC) and the
short circuit current (ISC), the fill factor (FF) can be defined as:
(5)
It is a widely used measure of the solar cell overall quality [4]. It is the ratio of the
actual maximum power (IMPPVMPP) to the theoretical one (ISCVOC), which is actually not
obtainable. The reason for that is that the MPP voltage and current are always below the
open circuit voltage and the short circuit current respectively, because of the series and
shunt resistances and the diode depicted in Figure 2. The typical fill factor for commercial solar
cells is usually over 0.70.
TEMPERATURE AND IRRADIANCE EFFECTS
Two important factors that have to be taken into account are the irradiation and the
temperature. They strongly affect the characteristics of solar modules. As a result, the MPP
varies during the day and that is the main reason why the MPP must constantly be tracked and
ensure that the maximum available power is obtained from the panel. The effect of the
irradiance on the voltage-current (V-I) and voltage-power (V-P) characteristics is depicted in
Figure 4, where the curves are shown in per unit, i.e. the voltage and current are normalized
using the VOC and the ISC respectively, in order to illustrate better the effects of the
irradiance on the V-I and V-P curves. As was previously mentioned, the photo-generated
current is directly proportional to the irradiance level, so an increment in the irradiation leads
to a higher photo-generated current. Moreover, the short circuit current is directly
proportional to the photo-generated current; therefore it is directly proportional to the
irradiance. When the operating point is not the short circuit, in which no power is generated,
the photo-generated current is also the main factor in the PV current, as is expressed by
equations (1) and (2). For this reason the voltage-current characteristic varies with the
irradiation. In contrast, the effect in the open circuit voltage is relatively small, as the
dependence of the light generated current is logarithmic, as is shown in equation (4).
Page 21
21
Figure 4-V-I and V-P curves at constant temperature (25°C) and three different insolation values
Figure-4 shows that the change in the current is greater than in the voltage. In practice,
the voltage dependency on the irradiation is often neglected. As the effect on both the current
and voltage is positive, i.e. both increase when the irradiation rises, the effect on the
power is also positive: the more irradiation, the more power is generated.
The temperature, on the other hand, affects mostly the voltage. The open circuit voltage is
linearly dependent on the temperature, as shown in the following equation:
(6)
According to (6), the effect of the temperature on VOC is negative, because Kv is
negative, i.e. when the temperature raises, the voltage decreases. The current increases with
the temperature but very little and it does not compensate the decrease in the voltage
caused by a given temperature rise. That is why the power also decreases. PV panel
Page 22
22
manufacturers provide in their data sheets the temperature coefficients, which are the parameters
that specify how the open circuit voltage, the short circuit current and the maximum power vary
when the temperature changes. As the effect of the temperature on the current is really small,
it is usually neglected. Figure 5 shows how the voltage-current and the voltage-power
characteristics change with temperature. The curves are again in per unit, as in the previous
case.
As was mentioned before, the temperature and the irradiation depend on the
atmospheric conditions, which are not constant during the year and not even during a single
day; they can vary rapidly due to fast changing conditions such as clouds. This causes the
MPP to move constantly, depending on the irradiation and temperature conditions. If the
operating point is not close to the MPP, great power losses occur. Hence it is essential to
track the MPP in any conditions to assure that the maximum available power is obtained
from the PV panel. In a modern solar power converter, this task is entrusted to the MPPT
algorithms.
Figure 5-V-I and V-P curves at constant irradiation (1 kW/m2) and three different temperatures
Page 23
23
TYPES OF SOLAR CELLS
Over the past decades, silicon has been almost the only material used for manufacturing solar
cells. Although other materials and techniques have been developed, silicon is used in more than
the 80% of the production. Silicon is so popular because it is one of the most abundant materials
in the Earth’s crust, in the form of silicon dioxide, and it is not toxic. Mono-crystalline and
polycrystalline silicon solar cells are the two major types of silicon solar cells. There is a third
type, amorphous silicon, but the efficiency is worse than with the previous types so it is less
used. Other new solar cells are made of copper indium gallium (di)selenide (CIGS) or cadmium
telluride (CdTe). Much research and development (R&D) effort is being made to develop new
materials, but nowadays there are no commercial substitutes to the above types of solar cells. In
this section these different solar cells are reviewed.
One of the most important characteristics of solar cells is the efficiency, which is the percentage
of solar radiation that is transformed into electricity. It is measured under Standard Test
Conditions (STC), irradiance of 1000 W/m², air mass coefficient (it characterizes the solar
spectrum after the solar radiation has travelled through the atmosphere) A.M 1.5, and a cell
junction temperature of 25°C. The higher efficiency, the smaller surface is needed for a given
power. This is important because in some applications the space is limited and other costs and
parameters of the installation depend on the installed PV surface.
MONO-CRYSTALLINE SILICON
Mono-crystalline silicon solar cells are the most efficient ones. They are made from wafers (very
thin slices) of single crystals obtained from pure molten silicon. These single crystal wafers have
uniform and predictable properties as the structure of the crystal is highly ordered. However the
manufacturing process must be really careful and occurs at high temperatures, which is
expensive. The efficiency of these cells is around 15-18% and the surface needed to get 1 kW in
STC is about 7 m2.
POLYCRYSTALLINE SILICON
These cells are also made from wafers of pure molten silicon. However, the crystal structure is
random: as the silicon cools, it crystallizes simultaneously in many different points producing an
irregular structure: crystals of random sizes, shapes and orientation. These structures are not as
ideal as in the mono-crystalline cells so the efficiency is lower, around 11-15%. However the
manufacturing process is less expensive, so the lower efficiency is compensated in some way.
Page 24
24
The surface needed to obtain 1 kW in STC is about 8m2.
AMORPHOUS AND THIN-FILM SILICON
Amorphous silicon is the non-crystalline form of the silicon and it can be deposited as thin-films
onto different substrates. The deposition can be made at low temperatures. The manufacturing
process is simpler, easier and cheaper than in the crystalline cells. The weak point of these cells
is their lower efficiency, around 6-8%. This efficiency is measured under STC. However, the
performance under weaker or diffuse irradiation, such as that in cloudy days, can be higher than
in crystalline cells and their temperature coefficient is smaller. Amorphous silicon is also a
better light absorber than crystalline, so despite having low efficiency, the thin film is a
competitive and promising technology. The first solar cells were of thin-film technology. They
have been used since the 1980s in consumer electronics applications, such as calculators. In
recent years it has also begun to be used in high power applications due to the characteristics
mentioned above. One common use nowadays is as building cladding, for example in facades, as
its price is competitive compared with other high quality cladding materials and it offer the
advantage of electricity generation. The main advantages of thin film technologies are the ease of
manufacturing at low temperatures using inexpensive substrates and continuous production
methods, avoiding the need for mounting individual wafers and the potential for lightweight and
flexible solar cells. These advantages are common to most of the thin-film solar cells, not only
the ones made from amorphous silicon.
Over recent years, one more type of silicon has been developed, microcrystalline silicon. It can
also be deposited as thin-films onto different substrates, minimizing the quantities of crystalline
silicon needed and improving the efficiency of amorphous silicon. However, the light absorption
of micro-crystalline silicon compared to amorphous silicon is poor. The solution can be an
effective light trapping to keep the incident light within the film. This type of silicon is not a
commercial technology yet and more R&D is needed.
OTHER CELLS AND MATERIALS
As was mention in the introduction of this chapter, there are other materials apart from silicon
that can be used for manufacturing solar cells. These compounds are also thin-film deposited, so
they have the same advantages as the silicon thin film solar cells but with a better efficiency.
Among these compounds, two are already used in commercial solar cells. They are CIGS and
CdTe. The efficiency is around 10-13% [3] and it will rise in the following years as the
technologies are improved. It is commonly said that thin film technology is the way to achieve
Page 25
25
the grid parity, i.e. the point at which the cost of generating electricity is equal, or cheaper than
grid power.
The main disadvantages of these technologies are the toxicity of some of the compounds
and the shortage of some of the elements used. In the case of the CIGS, indium is used. This
element is not as abundant as silicon in the Earth’s crust and it is in high demand for other
electronics products such as liquid-crystal display (LCD) monitors, which has generated a
shortage and consequently a high price rise in the recent years. Moreover, to create the p-n
junction, CIGS is interfaced with cadmium sulphide (CdS), another semiconductor. The problem
is that cadmium is a heavy metal which is cumulatively poisonous. In the case of CdTe, the other
compound used in commercial thin film solar cells, it is not as toxic as its individual
components, but some precautions must be taken during the manufacturing process. Gallium
Arsenide (GaAs) has been used for space applications mainly for two reasons: firstly, it is less
susceptible to suffer damage from the space radiation than silicon, and secondly, due to its direct
bandgap of 1.42 eV, it can take advantage of a greater part of the solar spectrum. Despite being a
more expensive material, space projects can afford it as cost is not the most important factor to
decide the components. Nowadays it is being investigated to be used in terrestrial PV
applications using light concentrators (mirror or lenses) to focus the light onto small cells,
reducing the price as less material is required.
Triple junction GaAs cells have already passed 40% efficiency in the laboratory using light
concentrators. The main handicap at present for this technology is that concentration systems are
expensive as they have to track the Sun along the day. One other technology that is being
actively researched is dye-sensitized cells. These cells are made from artificial organic materials
and are seen as part of the “third generation” of solar cells. The efficiency of these cells is above
that of amorphous silicon and within the thin-film ones. The main advantage is that they work
well under low and diffuse light and their temperature coefficients are lower. The materials used
are non-toxic and abundant and their manufacturing processes are relatively simple. Flexible
modules can easily be made using flexible substrates and they can be used for building integrated
PV: roofs, windows, as they can be manufactured in many shapes, sizes and design criteria.
These last two paragraphs illustrate technologies that are being currently investigated. They are
non commercial technologies yet, but it is expected that in the following years they will become
competitive and will be also used, increasing the possibilities of PV power generation. The
silicon and thin film solar cells described before are currently the technologies used in
commercial PV applications.
Page 26
26
Nevertheless, what is important for this work is that all the different solar cells presented
above have similar non-linear voltage-current characteristics and are affected by irradiation and
temperature in a similar way. The only difference is that different type of cells have different
levels of sensitivity, nevertheless the same algorithms can be used to track the MPP.
PHOTOVOLTAIC MODULES
PV modules are made from solar cells connected in series and parallel to obtain the desired
current and voltage levels. Solar cells are encapsulated as they have to be weatherproofed and
electric connections also have to be robust and corrosion free. The typical construction of a PV
module can be seen in Figure-6.
Figure 6-Typical construction of PV Module
As the cells are brittle, they are encapsulated in an airtight layer of ethylene vinyl acetate
(EVA), a polymer, so the cells are cushioned and in that way are protected during transport
and handling. The top cover is a tempered glass treated with an anti-reflection coating so the
maximum light is transmitted to the cell. The underneath is a sheet of polyvinyl fluoride
(PVF), also known as Tedlar, a synthetic polymer (CH2CHF)n that constitutes a barrier
to moisture and prevents the cell from chemical attack. An aluminium frame is used to
simplify mounting and handling and to give extra protection. Frameless modules are
sometimes used in facades for aesthetic reasons. This typical construction is used because the
PV module has to “survive” outdoors for at least 20-25 years under different weather
conditions, sometimes extreme. This construction assures at least the lifetime of the PV
Page 27
27
modules. In fact, PV panel manufacturers provide a guarantee of at least 20 years, for
example BP Solar assures 85 % of minimum warranted power output after 25 years of
service, 93 % of the minimum warranted power output at 12 years and a five-year warranty
of materials and workmanship. Such a long guarantee is extremely long compared to most
products and is due to the exceptional construction of PV modules.
PHOTOVOLTAIC SYSTEM CONFIGURATION
PV modules generate DC current and voltage. However, to feed the electricity to the
grid, AC current and voltage are needed. Inverters are the equipment used to convert
DC to AC. In addition, they can be in charge of keeping the operating point of the PV
array at the MPP. This is usually done with computational MPP tracking algorithms.
There are different inverter configurations depending on how the PV modules are
connected to the inverter. The main types are described in this chapter. The decision
on what configuration should be used has to be made for each case depending on the
environmental and financial requirements. If the modules are not identical or do not
work under the same conditions, the MPP is different in each panel and the resulting
voltage-power characteristic has multiple maxima, which constitutes a problem, because
most MPPT algorithms converge to a local maximum depending on the starting point. If
the operating point is not the MPP, not all the possible power is being fed to the grid.
For these reasons each case has to be carefully studied to optimize the plant and obtain
the maximum performance.
The below image describes the typical arrangement of the Solar PV with MPPT controller with
DC-DC converter Buck, Boost or Cuk converter.
Figure 7-MPPT Block Diagram
Page 28
28
The different configurations are described shortly in this chapter because they are not the
focus of this thesis.
CENTRAL INVERTER
It is the simpler configuration: PV strings, consisting of series connected PV panels, are
connected in parallel to obtain the desired output power. The resulting PV array is
connected to a single inverter, as is shown in Figure 7. In this configuration all PV
strings operate at the same voltage, which may not be the MPP voltage for all of them.
The problem of this configuration is the possible mismatches among the different PV
modules. If they are receiving different irradiation (shading or other problems), the true MPP
is difficult to find and consequently there are power losses and the PV modules are
underutilized.
Figure 8-Arrangement for Central Inverter
STRING INVERTER
In this configuration, every string of PV panels connected in series is connected to a
different inverter, as can be seen in Figure-8. This can improve the MPP tracking in case of
Page 29
29
mismatches or shading, because each string can operate at a different MPP, if
necessary, whereas in the central inverter there is only one operating point which may not be
the MPP for each string, thus leading to power losses. On the other hand, the number of
components of the system increases as well as the installation cost, as an inverter is used for
each string.
Figure 9-Arrangement for String Inverter
MULTI-STRING INVERTER
In this case each string is connected to a different DC-DC converter, which is in charge
of the MPP tracking of the string, and the converters are connected to a single inverter,
as depicted in Figure 9. The advantages related to MPP tracking are the same as in the
string configuration; each string can have a different MPP. The disadvantages, an
increase in the price compared to the central inverter, as a converter is used for each
string.
Figure 10-Arrangement for Multi String Inverter
MODULE INTEGRATED INVERTER
In this configuration, as shown in Figure-10, each PV module is connected to a different
inverter and consequently the maximum power is obtained from each panel as the individual
Page 30
30
MPP is tracked by each inverter. This configuration can be used when the differences in the
operating point of the different modules are large. However, it is more expensive because each
panel has its own inverter.
Figure 11-Arrangement for Module Inverter
Page 31
31
CHAPTER 3: MAXIMUM POWER POINT TRACKING
As was previously explained, MPPT algorithms are necessary in PV applications because the
MPP of a solar panel varies with the irradiation and temperature, so the use of MPPT
algorithms is required in order to obtain the maximum power from a solar array.
Over the past decades many methods to find the MPP have been developed and
published. These techniques differ in many aspects such as required sensors,
complexity, cost, range of effectiveness, convergence speed, correct tracking when
irradiation and/or temperature change, hardware needed for the implementation or
popularity, among others.
Among these techniques, the P&O and the InCond algorithms are the most common. These
techniques have the advantage of an easy implementation but they also have drawbacks,
as will be shown later. Other techniques based on different principles are fuzzy logic
control, neural network, fractional open circuit voltage or short circuit current, current
sweep, etc. Most of these methods yield a local maximum and some, like the fractional
open circuit voltage or short circuit current, give an approximated MPP, not the exact one.
In normal conditions the V-P curve has only one maximum, so it is not a problem. However,
if the PV array is partially shaded, there are multiple maxima in these curves. In order to
relieve this problem, some algorithms have been implemented. There are many methods
for MPPT, in the next section three most popular MPPT techniques are discussed.
HILL-CLIMBING TECHNIQUES
Both P&O and InCond algorithms are based on the “hill-climbing” principle, which consists of
moving the operation point of the PV array in the direction in which power increases. Hill-
climbing techniques are the most popular MPPT methods due to their ease of implementation
and good performance when the irradiation is constant. The advantages of both methods are the
simplicity and low computational power they need. The shortcomings are also well-known:
oscillations around the MPP and they can get lost and track the MPP in the wrong direction
during rapidly changing atmospheric conditions.
PERTURB AND OBSERVE
The P&O algorithm is also called “hill-climbing”, but both names refer to the same algorithm
depending on how it is implemented. Hill-climbing involves a perturbation on the duty cycle of
Page 32
32
the power converter and P&O a perturbation in the operating voltage of the DC link between
the PV array and the power converter. In the case of the Hill-climbing, perturbing the duty cycle
of the power converter implies modifying the voltage of the DC link between the PV array and
the power converter, so both names refer to the same technique.
In this method, the sign of the last perturbation and the sign of the last increment in the power
are used to decide what the next perturbation should be. As can be seen in Figure 11, on the left
of the MPP incrementing the voltage increases the power whereas on the right decrementing the
voltage increases the power.
Figure 12-V-I-P of Solar Array
If there is an increment in the power, the perturbation should be kept in the same
direction and if the power decreases, then the next perturbation should be in the opposite
direction. Based on these facts, the algorithm is implemented. The process is repeated until the
MPP is reached. Then the operating point oscillates around the MPP. This problem is common
also to the InCond method, as was mention earlier. A scheme of the algorithm is shown in
Figure 13.
Page 33
33
Figure 13-Flow Chart of P&O algorithm
INCREMENTAL CONDUCTANCE
The incremental conductance algorithm is based on the fact that the slope of the curve power vs.
voltage (current) of the PV module is zero at the MPP, positive (negative) on the left of it and
negative (positive) on the right, as can be seen in Figure 11:
• ∆V/∆P = 0 ( ∆I/∆P = 0 ) at the MPP
• ∆V/∆P > 0 ( ∆I/∆P < 0 ) on the Left
• ∆V/∆P < 0 ( ∆I/∆P > 0 ) at the Right
By comparing the increment of the power vs. the increment of the voltage (current)
between two consecutives samples, the change in the MPP voltage can be determined. A
scheme of the algorithm is shown in Figure 13.
Page 34
34
Figure 14-Flow Chart of InCond Algorithm
In both P&O and InCond schemes, how fast the MPP is reached depends on the size of the
increment of the reference voltage.
The drawbacks of these techniques are mainly two. The first and main one is that they can
easily lose track of the MPP if the irradiation changes rapidly. In case of step changes they
track the MPP very well, because the change is instantaneous and the curve does not keep on
changing. However, when the irradiation changes following a slope, the curve in which the
algorithms are based changes continuously with the irradiation, as can be seen in Figure 14,
so the changes in the voltage and current are not only due to the perturbation of the voltage. As
a consequence it is not possible for the algorithms to determine whether the change in the
power is due to its own voltage increment or due to the change in the irradiation.
Page 35
35
Figure 15-MPP at different radiation level
The other handicap of both methods is the oscillations of the voltage and current around the
MPP in the steady state. This is due to the fact that the control is discrete and the voltage and
current are not constantly at the MPP but oscillating around it. The size of the oscillations
depends on the size of the rate of change of the reference voltage. The greater it is, the higher
is the amplitude of the oscillations. However, how fast the MPP is reached also depends on this
rate of change and this dependence is inversely proportional to the size of the voltage
increments. The traditional solution is a trade off: if the increment is small so that the
oscillations decrease, then the MPP is reached slowly and vice versa, so a compromise solution
has to be found.
To overcome these drawbacks some solutions have been published in recent years. Regarding
the rapid change of the irradiation conditions, Sera et al. published in an improved P&O
method, called “dP-P&O”, in which an additional measurement is performed without
perturbation in the voltage and current. In this way, every three consecutive samples the effect
of the perturbation in the voltage (current) and the effect of the change in the atmospheric
conditions can be evaluated so that the increment in the power used in the algorithm only
Page 36
36
contains the effect caused purely by the MPPT algorithm. Then the correct decision about the
direction of the next perturbation can be taken. The efficiency of the tracking is improved.
Although the method was tested using irradiation slopes, they were not the ones proposed in
the new European Standard EN 50530.
A different solution is suggested, which considers the traditional P&O algorithm, in which the
perturbation amplitude is tuned constantly taking into account the previous changes in the
power. It also includes a stage in which the latest increment in the power is compared with the
latest perturbation amplitude to determine if the power increment was due to a change in the
irradiation. If this is the case, then the voltage perturbation is set to the same direction as the
change in the power condition. The steady state error and the tracking speed are improved, but
the algorithm has only been tested with irradiation step changes and not with the irradiation
slopes proposed.
In relationship with the oscillations around the MPP in steady state, Zhang et al. proposed a
variable perturbation step for the P&O algorithm to reduce the oscillation around it. This
modified P&O method determines also if the operating point is near to or far from the MPP
and adjusts the size of the perturbation according to that: if the operating point is near to the
MPP, the perturbation size is reduced and if the point is far, then it is increased. This technique
improves the convergence speed and reduces
the oscillation around the MPP. A similar technique is found: a variation of the traditional
P&O algorithm in which the amplitude of the voltage perturbation is adapted to the actual
operating conditions: large perturbation amplitudes are chosen far from the maximum whereas
small ones are used near the MPP. The proposed algorithm requires initial panel identification
and has to be tuned for each plant. With this technique the dynamic response and the steady
state stability are improved. Unfortunately, the last
two algorithms do not improve the tracking under changing irradiance conditions. Although
the authors claim the performance is better, the algorithms have only been tested with
irradiation step changes but not with irradiation ramps as proposed in the European Standard
mentioned above.
Many papers have been published about optimizing the parameters of these algorithms for
different hardware configurations. The sample frequency for P&O is optimized and it is shown
how the P&O MPPT parameters must be customized to the dynamic behaviour of the specific
converter adopted. It has been traditionally said that the performance of InCond algorithm is
Page 37
37
better than the P&O. However, according to the performance is similar if the parameters of the
P&O method are optimized. In any case, both algorithms are based on the same principle and
have the same problem so they have been analyzed together.
The amount of literature presenting slight modifications of the existing methods or adapting
them to different hardware configurations is so extensive that it is not possible to present it in
this thesis.
In any case, none of the solutions reviewed before solves the problems satisfactorily and none
has been tested under the slopes proposed to test the dynamic efficiency of the MPPT
algorithms. These profiles simulate rapid environmental changes such as clouds. It is very
important to track the MPP during these situations to obtain the maximum power from the PV
module. As will be shown in the next chapter, this thesis proposes some modification to both
P&O and InCond methods so that the tracking under irradiation profiles containing slopes is
very good.
FUZZY LOGIC CONTROL
The use of fuzzy logic control has become popular over the last decade because it can deal
with imprecise inputs, does not need an accurate mathematical model and can handle
nonlinearity. Microcontrollers have also helped in the popularization of fuzzy logic control.
The fuzzy logic consists of three stages: fuzzification, inference system and defuzzification.
Fuzzification comprises the process of transforming numerical crisp inputs into linguistic
variables based on the degree of membership to certain sets.
Membership functions, like the ones in Figure 16, are used to associate a grade to each
linguistic term. The number of membership functions used depends on the accuracy of the
controller, but it usually varies. In Figure 16 seven fuzzy levels are used: NB (Negative Big),
NM (Negative Medium), NS (Negative Small), ZE (Zero), PS (Positive Small), PM
(Positive Medium) and PB (Positive Big). The values a, b and c are based on the range values
of the numerical variable. In some cases the membership functions are chosen less
symmetric or even optimized for the application for better accuracy.
Page 38
38
Figure 16-Membership function for MPPT
The inputs of the fuzzy controller are usually an error, E, and the change in the error, ∆∆∆∆E.
The error can be chosen by the designer, but usually it is chosen as ∆∆∆∆P/∆∆∆∆V because it is zero
at the MPP. Then E and ∆∆∆∆E are defined as follows:
(7)
(8)
In other cases ∆P/∆I is used as error or other inputs are considered, where ∆V and ∆P are
used. The output of the fuzzy logic converter is usually a change in the duty ratio of the
power converter, ∆D, or a change in the reference voltage of the DC-link, ∆V. The rule base,
also known as rule base lookup table or fuzzy rule algorithm, associates the fuzzy output to
the fuzzy inputs based on the power converter used and on the knowledge of the user. Table I
shows the rules for a three phase inverter, where the inputs are E and ∆E, as defined in (7)
and (8), and the output is a change in the DC-link voltage, ∆V. For example, if the operating
point is far to the right of the MPP, E is NB, and ∆E is zero, then to reach the MPP the
reference voltage should decrease, so ∆V should be NB (Negative) to move the operating
point towards the MPP.
Page 39
39
Table 3-1 Fuzzy controller rule base for MPPT
The last stage of the fuzzy logic control is the defuzzification. In this stage the output is
converted from a linguistic variable to a numerical crisp one again using membership
functions as those in Figure 15. There are different methods to transform the linguistic
variables into crisp values. It can be said that the most popular is the centre of gravity
method. However the analysis of these methods is beyond the scope of this thesis.
Figure 17-Flow Chart of FLC based MPPT algorithm
Page 40
40
The advantages of these controllers, besides dealing with imprecise inputs, not needing
an accurate mathematical model and handling nonlinearity, are fast convergence and
minimal oscillations around the MPP. Furthermore, they have been shown to perform
well under step changes in the irradiation. However, no evidence was found that they
perform well under irradiation ramps. Therefore, their performance under the conditions
specified in for testing the dynamic MPPT efficiency is unknown. Another
disadvantage is that their effectiveness depends a lot on the skills of the designer; not
only on choosing the right error computation, but also in coming up with an appropriate
rule base.
MAXIMUM POWER POINT TRACKING SUMMARY
Most of the MPPT algorithms developed over the past years have been reviewed in the
previous sections. Some of them are very similar and use the same principle but expressed in
different ways, like the last three algorithms listed in the hill-climbing techniques. The most
popular MPPT algorithms according to the number of publications are P&O, InCond and
Fuzzy Logic. It makes sense because they are the simplest algorithms capable of finding the
real MPP. However, they have some disadvantages, as discussed earlier. In the following
chapter, the performance of these three algorithms is analyzed. They were selected because of
their simplicity and popularity. In the case of P&O and InCond some modifications are
proposed, which overcome the limitations of the original methods in tracking the MPP under
irradiation slopes. The FLC is designed according to the references and its dynamic efficiency
is tested and compared to the hill-climbing MPPT methods.
Page 41
41
CHAPTER 4: DC-DC CONVERTERS
In this chapter the characteristics of the basic dc-dc converter topologies; buck, boost, and
buck-boost, are analyzed to determine the best topology for performing PV module maximum
power point tracking. A model of the identified topology is then formulated. The model is
used to carry out simulations to determine the effect of component non-idealises on converter
efficiency and output voltage. The simulation results are used as the basis for developing
control strategies and selecting converter components. The converter model forms the main
part in the complete MPPT model used for tuning the fuzzy logic controller.
The schematic diagram of a dc-dc converter is shown in Figure 18. It converts a dc input
voltage Vg (t), to a dc output voltage VO (t), at a different voltage level from the input. It is
desirable that the conversion be made with low losses in the converter. Therefore, the
transistor is operated as a switch using the control signal δ (t), which is held high for a time
ton, and low for a time toff as shown in Figure 19.
Figure 18-Block Diagram of DC-DC converter
While the transistor is on, the voltage across it is low which means that the power loss in the
transistor is low. While the transistor is off, the current through it is low and the power loss is
also low. The average output voltage is controlled by changing the width of the pulses while
the switching period Ts is held constant. The duty cycle, d (t) is a real value in the interval 0 to
1 and it is equal to the ratio of the width of a pulse to the switching period i.e. d(t)= ton/Ts.
Page 42
42
Figure 19-Duty Ration of DC-DC converter
To obtain low losses, resistors are avoided in dc-dc converters. Capacitors and inductors are
used instead since ideally they have no losses. The electrical components can be combined
and connected to each other in different ways, called topologies, each one having different
properties. The buck, boost, and buck-boost converters are three basic converter topologies.
The buck converter has an output voltage that is lower than the input voltage; the boost
converter has an output voltage that is higher than the input voltage, and the buck-boost
converter is able to produce an output voltage magnitude that is higher or lower than the input
voltage magnitude.
IDENTIFICATION OF SUITABLE CONVERTER FOR MPPT
The different converter topologies are analyzed in this section in order to ascertain their
performance and identify the most suitable topology for maximum power point tracking.
The power produced from a photovoltaic module depends strongly on the operating voltage
of the load to which it is connected, as well as to the solar radiation level and cell
temperature. If a variable load resistance R is connected across the module’s terminals, the
operating point is determined by the intersection of module I-V curve and the load I-V
characteristic. Figure 20 illustrates the operating characteristic of a PV module. It consists of
two regions:
Zone I is the current source region, and Zone II is the voltage source region. In Zone I, the
internal impedance of the module is high, while in Zone II the internal impedance is low. The
maximum power point Pmp, is located at the knee of the power curve. Increase in solar
radiation at constant temperature causes a decrease in internal impedance as it causes an
increase in short-circuit current. An increase in temperature at constant solar radiation causes
a decrease in internal impedance since it causes a decrease in open circuit voltage.
Page 43
43
According to the maximum power transfer theory, the power delivered to the load is
Maximum when the source internal impedance matches the load impedance. The load
characteristic is a straight line with a slope of I /V = 1/ R. If R is small, the module operates in
the region AB only and behaves like a constant current source at a value close to Isc. If R is
large, the module operates in the region CD behaving like a constant voltage source, at a
value almost equal to Voc.
Figure 20-Behavioural curve of MPP for different converter operation
BUCK CONVERTER
For an ideal buck converter, averaged input voltage Vg , output voltage Vo , input current Ig ,
and output current Io are related as follows:
VO = Vg D (9)
Io = Ig / D (10)
Where, D is the equilibrium duty cycle of the converter. The dc load R connected to the
converter can be expressed using Ohm’s law as:
R = VO / Io (11)
Page 44
44
The load resistance ' R referred to the input terminals of the converter can be derived from Equations (9) and (10) as:
R’ = RL / D2 (12)
Since 0 < D <1, varying D can only increase the load seen by the source. A buck converter is
therefore, only able to extract maximum power if the original load draws a higher current
than the maximum power point current Imp of the PV module (Zone-1 in Figure-).
BOOST CONVERTER
For an ideal boost converter, the averaged input and output values of current and voltage are
related as follows:
VO =Vg / (1 - D) (13)
Io =Ig (1 – D) (14)
The load resistance R’ referred to the input side is given by:
R’ = R (1 – D) 2 (15)
Since 0 < D <1, varying D can only decrease the load seen by the source. It is therefore noted
that a tracker based on the boost converter is only able to extract maximum power if the
original load draws lower current than maximum power point current Imp, of the PV module
(Zone-II in Figure-).
BUCK-BOOST CONVERTER
For an ideal buck-boost converter, the averaged input and output values of current and
voltage are related as follows:
VO =Vg (D / 1 - D) (16)
Io =Ig (1 – D / D) (17)
The load resistance R’ referred to the input side is given by:
R’ = R (D / 1 – D) 2 (18)
Since 0 < D <1, varying D can increase or decrease the load seen by the source. The buck-
boost converter is therefore able to operate in both Zone-I and Zone-II in Figure-.
Page 45
45
Figure 21-Circuit of a buck-boost converter
CONCLUSION
It is noted from the analysis carried out in the preceding sub-sections that the buck-boost
converter has the best performance since it is able to perform maximum power tracking in
both zones I and II of Figure-20.
Page 46
46
CHAPTER 5: FUNDAMENTAL OF TYPE- 1 FUZZY SETS AND
SYSTEMS AND APPLICATION IN MPPT CONTROL
The fuzzy theory was first introduced into the scientific literature in 1965 by Professor Lotfi
A. Zadeh at the University of California at Berkeley who proposed a set theory that operated
over the range [0; 1]. While Boolean logic results are restricted to 0 and 1, fuzzy logic results
are between 0 and 1. In other words, fuzzy logic defines some intermediate values between
sharp evaluations like absolute true and absolute false. This means that fuzzy sets can handle
some concepts that we commonly meet in daily life, like ”very old”, ”old”, ”young”, ”very
young”. Fuzzy logic is more like human thinking because it is based on degrees of truth and
uses linguistic variables.
Fuzzy logic was not an acceptable theory for the scientists at that time because it contained
vagueness in the engineering field. However, since 1970s, this approach to set theory has
been widely applied to control systems. The principles of fuzzy logic were used to control a
steam engine by Ebraham Mamdani of University of London in 1974. It was a milestone for
fuzzy logic. The first industrial application was a cement kiln built in Denmark in 1975. In
the 1980s, Fuji Electric applied fuzzy logic theory to the control of a water purification
process. As a challenging engineering project, in 1987, Sendai Railway system that had
automatic train operation control was built with fuzzy logic principles in Japan. Fuzzy
control techniques were used in all the critical operations in the control of the train, such as
accelerating, breaking, and stopping operations. In 1987, Takeshi Yamakawa used fuzzy
control in an inverted pendulum experiment which is a classical control problem. After these
successful applications, not only the engineers but also the social scientists applied fuzzy
logic into different areas. In today’s technology, many companies use fuzzy logic in their
engineering projects like for example air conditioners, video cameras, televisions, washing
machines, bus time tables, medical diagnoses, antilock braking system, etc.
Classical control theory, typically PID controller, uses a mathematical model to define the
relationship between the inputs and the outputs of a system. The most serious disadvantage
of these controllers is that they usually assume the system to be linear or at least that it
behaves as a linear system in some range. If an accurate mathematical model of a system is
available, a conventional PID controller can make the performance of the system quite
acceptable. However, in real life, an accurate mathematical model of a control process is not
generally available, even it may not exist. The real world is nonlinear, uncertain and contains
Page 47
47
always incomplete data. If the mathematical model is not known by the designer, there is no
way to come up with a proper PID controller design. Even in those cases, when the
mathematical model is known to be relatively accurate, the parameters of the system are
likely to change by some external factors, like heat or pressure, etc. In such cases, a model-
free approach is preferable. Fortunately, fuzzy logic controllers (FLCs) have the ability to
control a system using some limited expert knowledge. In most cases, the design procedure of
a FLC tries to imitate an expert or a skilled human operator. Besides, FLCs are low-cost
implementations based on cheap sensors.
In general, fuzzy logic is a nonlinear mapping of an input data vector into a scalar output. The
main approaches to design of a Type-1 FLC in literature those Type-1 fuzzy sets:
Membership functions are totally certain.
TYPE-1 FUZZY SETS AND MEMBERSHIP FUNCTION
A type-1 fuzzy set, A, which is in terms of a single variable, x ∈ X, may be represented as:
A = {(x, µA(x))| ∀∀∀∀x ∈∈∈∈ X} (19)
A can also be defined as:
(20)
Where ∫∫∫∫ denotes union over all admissible x.
As can be seen from Figure 2.1, a type-1 Gaussian membership function, µA(x), is
constrained to be between 0 and 1 for all x ∈ X, and is a two-dimensional function. This type
of membership function does not contain any uncertainty. In the other word, there is a clear
membership value for every input data point.
Page 48
48
Figure 22-A Gaussian type-1 fuzzy membership function
FUZZY RULES AND INFERENCE
The use of fuzzy sets allows the characterization of the system behaviour through fuzzy rules
between linguistic variables. A fuzzy rule is a conditional statement Ri based on expert
knowledge expressed in the form:
Ri : IF x is small THEN y is large (21)
Where x and y are fuzzy variables and small and large are labels of the fuzzy sets. If there are
n rules, the rule set is represented by the union of these rules i.e.
R = R1 else R2 else ….. Rn . (22)
A fuzzy controller is based on a collection R, of control rules. The execution of these rules is
governed by the compositional rule of inference.
FUZZY CONTROLLER STRUCTURE
The general structure of a fuzzy logic controller is presented in Figure- and comprises of four
principal components:
• Fuzzification interface: - It converts input data into suitable linguistic values using a
membership function.
Page 49
49
• Knowledge base: - Consists of a database with the necessary linguistic definitions
and the control rule set.
• Inference engine: - It simulates a human decision making process in order to infer the
fuzzy control action from the knowledge of the control rules and the linguistic
variable definitions.
• Defuzzification interface: - Converts an inferred fuzzy controller output into a non-
fuzzy control action.
Figure 23-Basic configuration of a fuzzy logic controller
APPLICATION OF TYPE-1 FUZZY LOGIC IN MPPT CONTROL
DC-DC converter systems are becoming strong candidates for modern fuzzy control
techniques due to their complex, nonlinear behaviour, particularly for large load and line
variations. The highly nonlinear behaviour of these power circuits is caused by the presence
of a switch, which can be any electronic switch such as a transistor, a Mosfet, or any other
switching device. Depending on the state of the switch (ON/OFF) the plant structure exhibits
very different functioning modes, resulting in a severe nonlinearity. PV modules also have
nonlinear current-voltage (I-V) characteristics that are dependent on solar radiation,
temperature, and degradation due to environmental effects. Therefore, their operating point
that corresponds to the maximum output power varies with the environmental and load
conditions.
MPPT control is therefore an intriguing subject from the control point of view, due to the
intrinsic nonlinearity of dc-dc converters and PV modules. This is because an accurate model
of the plant and the controller is necessary while formulating the control algorithm. There are
Page 50
50
two possible ways of overcoming this. One method is to develop more accurate nonlinear
models for controllers, but the discouraging fact about taking this route is that complex
mathematical derivations are involved. Even when developed, the complicated control
algorithms may not be suitable for practical implementations. The other method is to employ
heuristic reasoning based on human experience of the plant. Such experience is usually
collected in the form of linguistic statements and rules. In this case, no modeling is required,
and the whole business of controller design reduces to the "conversion" of a set of linguistic
rules into an automatic control algorithm. Here, fuzzy logic comes into play as it provides the
essential machinery for performing the said conversion. Such a completely different approach
is offered by fuzzy logic, which does not require a precise mathematical modeling of the
system nor complex computations. This control technique relies on the human capability to
understand the system's behaviour and is based on qualitative control rules. Thus, controller
design is simple, since it is only based on linguistic rules of the type: “IF the change in output
power is positive AND the change in duty cycle is negative THEN reduce slightly the duty
cycle" and so on.
Fuzzy logic control relies on basic physical properties of the system, and it is potentially able
to extend control capability even to those operating conditions where linear control
techniques fail, i.e., large signal dynamics and large parameter variations. As fuzzy logic
control is based on heuristic rules, application of nonlinear control laws to overcome the
nonlinear nature of dc-dc converters is easy. Fuzzy logic offers several unique features that
make it a particularly good choice for these types of control problems because:
• It is inherently robust as it does not require precise, noise-free inputs. The output
control is a smooth control function despite a wide range of input variations.
• It can be easily modified to improve system performance by generating appropriate governing rules.
• Any sensor data that provides some indication of a system's actions and reactions is
sufficient. This allows the sensors to be inexpensive and imprecise thus keeping the
overall system cost and complexity low.
• Its rule-based operation enables any reasonable number of inputs to be processed and
numerous outputs generated. The control system can be broken into smaller units that
Page 51
51
use several smaller fuzzy logic controllers distributed on the system, each with more
limited responsibilities.
DESIGN OF FUZZY LOGIC CONTROLLER PARAMETERS
CONTROLLER STRUCTURE
The basic scheme of a fuzzy logic based maximum power point tracker is shown in Figure-.
The dc-dc converter is represented by a “black box” from which only the terminals
corresponding to input voltage Vm, input current Im from the PV module, and the controlled
switch S are extracted. As observed, only two state variables are sensed; the input voltage and
input current. The two values are used to calculate the input power. From these
measurements, the fuzzy logic controller provides a signal proportional to the converter duty
cycle which is then applied to the converter through a pulse width modulator. The modulator
uses the value of D to perform Pulse Width Modulation (PWM), which generates the control
signals for the converter switch. The fuzzy logic controller scheme is a closed loop system.
Figure 24-Fuzzy control scheme for a maximum power point tracker
A functional block diagram representation of the fuzzy controller is shown in Figure-. The
inputs to the controller are the scaled change of power, kp∆∆∆∆Pk and the previous change of
duty, ∆∆∆∆Dk-1, where kp is a power scale factor, and k is the sampling instant. The output of the
fuzzy controller is the duty cycle Dk at the k -th sampling instant, and is defined as
Page 52
52
Dk = Dk-1 + kd ∆∆∆∆Dk (23)
Where ∆∆∆∆Dk is the inferred change of duty cycle by the fuzzy controller at the k-th sampling
instant and kd is a duty-cycle scale factor. The block containing the term 1 Z − in Figure-
indicates a unit time delay.
Figure 25-Functional block of the fuzzy controller
MEMBERSHIP FUNCTIONS
Fuzzy sets for each input and output variable are defined as shown in Figure-. Five fuzzy
subsets Negative Big (NB), Negative Small (NS), Zero (ZE), Positive Small (PS), and
Positive Big (PB) are chosen for the input variable ∆∆∆∆Pk, eleven subsets are used for the input
and output variable ∆∆∆∆Dk-1. The subsets are NB, Negative Medium (NM), Negative Medium
Medium (NMM), NS, Negative Small Small (NSS), ZE, Positive Small Small (PSS), PS,
Positive Medium Medium (PMM), Positive Medium (PM), and PB. Eleven fuzzy subsets
were chosen for ∆∆∆∆Dk-1 in order to smooth the control action. As shown in Figure-, triangular
and trapezoidal shapes have been adopted for the membership functions; the value of each
input and output variable is normalized. The same membership function is used for the output
value ∆∆∆∆Dk and the input value ∆∆∆∆Dk-1.
Page 53
53
Figure 26-Membership functions for (a) change in power (b) change in duty cycle
The membership functions for ∆∆∆∆Pk and ∆∆∆∆Dk-1 are made denser at the centre in order to
provide more sensitivity as the variation of power approaches zero. The duty cycle is
internally limited to a maximum value of 90% to prevent operation at low efficiencies. It is
also limited to a minimum value of 10% to ensure that the converter switching process does
not stop as operation at D = 0 will indicate a false maximum power point.
SCALING FACTORS
For simplicity, the universe of discourse for each fuzzy variable was normalized to be in the
range [-1 1]; this procedure involves scale mapping for the input and output data. The choice
of the scale factors kp and kd greatly affects the bandwidth and the overall performance of the
controller. The factor kp determines the sensitivity of the controller to changes in power, and
kd determines the sensitivity to change in duty cycle. Suitable values of kp and kd were
chosen based on simulation results.
DERIVATION OF CONTROL RULES
Fuzzy control rules are obtained from the analysis of the system behaviour. The different
operating conditions are considered in order to improve tracking performance in terms of
dynamic response and robustness. The algorithm can be explained as follows: the tracking
process is started with an initial duty cycle, D0. The converter input current Im, and voltage
Vm, are then measured and used to compute the module power Pk. Then, the duty cycle is
increased by the controller based on the initial changes in power and duty cycle. At stage two,
m I and m V are measured and used to compute Pk+1. After gathering the past and present
information of the module power, the controller makes a decision on whether to increase or
decrease the duty cycle. This tracking process repeats itself continuously until the peak power
point is reached. The control rules are divided into four categories.
Page 54
54
Category I
These rules are used to guide the controller under constant operating conditions i.e. when
there are no variations in solar radiation, temperature, or load. Operation is based on the
meta-rule:
“IF the last change in the duty cycle has caused the power to increase, keep moving the duty
cycle in the same direction; otherwise if it has caused power to decrease, move it in the
opposite direction.”
The duty cycle is changed in adaptive steps which depend on the change in power to ensure
the maximum power point is approached at a fast speed and to prevent oscillations around it.
There are forty rules in this category and they are shown in Table 5-1 with a degree of
support of 1. These rules are given the strongest degree of support since the task they
represent describes the normal system operation. The rules are read as,
• IF ∆∆∆∆Pk is NB AND ∆∆∆∆Dk- 1 is NB THEN ∆∆∆∆Dk is PM
• IF ∆∆∆∆Pk is NB AND ∆∆∆∆Dk- 1 is NM THEN ∆∆∆∆Dk is PMM
• IF ∆∆∆∆Pk is NB AND ∆∆∆∆Dk- 1 is NS THEN ∆∆∆∆Dk k is PSS
• …………………………………………………………..
• IF ∆∆∆∆Pk is PB AND ∆∆∆∆Dk- 1 is PB THEN ∆∆∆∆Dk is PM.
Table 5-1 Fuzzy controller rule base
Page 55
55
Category II
The rules in this category guide the controller when there are sudden changes in solar
radiation, temperature, or load leading to an overall shift in the optimum point. There are four
rules in this collection and are found in the vertical column of Table 5-1 corresponding to
∆∆∆∆Dk-1 = 0, and a degree of support of 0.5. This condition is rare and the rules are used to
return the system to normal operation where category I rules are activated to search for the
new optimum operating point.
Category III
The rules in this group are used to ensure that the maximum power transfer search only stops
when the true maxima has been reached. Several false maxima are introduced due to the
quantization effect shown in Figure 27. Since the input signals are digitized, the continuous
curve is broken into a series of plateaus (points with constant power). It is observed from
Figure 27 that the steeper the curve, the shorter the plateau. Since the optimum point satisfies
the condition δP / δD = 0, the controller might recognize any large plateau as a maximum
power point and stop. There are ten rules in this category and are found in the horizontal
column of Table 5-1 corresponding to ∆∆∆∆Pk = 0. The rules are given a degree of support of 0.5
since the condition is rare.
Category IV
There is only one rule in this category. The rule is activated when the system reaches the
optimum point and it is used to stabilize operation at the maximum power point. It is given a
weight of 0.25 as shown in Table 5-1. The rule is read as,
• IF ∆∆∆∆Pk is E AND ∆∆∆∆Dk-1 is ZE THEN ∆∆∆∆Dk is ZE
Page 56
56
Figure 27-Quantization effect during maximum power search
DECISION MAKING
From the membership functions in Figure 26, it is observed that every input and output
belongs to at most two fuzzy sets. A maximum of four rules are therefore activated at any
sampling instant during which the input signals ∆∆∆∆Pk and ∆∆∆∆Dk-1 are processed. For instance, let
∆∆∆∆Pk = 0.2 and ∆∆∆∆Dk-1= -0.35. Change in power belongs ∆ ∆ ∆ ∆Pk to the fuzzy set PS and PB with a
degree of membership µPS(∆∆∆∆Pk) = 0.33, and µPB(∆∆∆∆Pk) = 0.71. Change in duty cycle ∆∆∆∆Dk-1
belongs to the fuzzy set NMM and NM with µNMM(∆∆∆∆Dk-1) = 0.5, and µNM(∆∆∆∆Dk-1) = 0.25. The
degree of membership for the other membership functions is zero. Therefore, the following
four rules are activated,
• Rule 1: IF ∆∆∆∆Pk is PS AND ∆∆∆∆Dk-1 is NM THEN ∆∆∆∆Dk is NMM.
• Rule 2: IF ∆∆∆∆Pk is PS AND ∆∆∆∆Dk-1 is NMM THEN ∆∆∆∆Dk is NS
• Rule 3: IF ∆∆∆∆Pk is PB AND ∆∆∆∆Dk-1 is NM THEN ∆∆∆∆Dk is NMM
• Rule 4: IF ∆∆∆∆Pk is PB AND ∆∆∆∆Dk-1 is NMM THEN ∆∆∆∆Dk is NS
Page 57
57
The inference result of each rule consists of two parts, the degree of fulfilment, DOFi of the
individual rule, and the weighting factor wi, according to the rule. The degree of fulfilment
(DOF) is obtained by means of Mamdani’s min-fuzzy implication and wi is retrieved from the
control rule table. The degree of fulfilment of each rule using Mamdani’s min fuzzy
implication is given by,
DOFi = min { µ∆∆∆∆P (∆∆∆∆Pk), µ∆∆∆∆D k-1(∆∆∆∆Dk-1) } (24)
And the output of the each rule is given by,
zi = (DOFi) wi (25)
Where, zi denotes the fuzzy representation of change in duty cycle inferred from the i-th rule.
Since the inferred output is fuzzy, the defuzzification operation is performed to obtain a crisp
output.
The fuzzy inference system with the Mamdani’s min fuzzy implication method for inputs ∆∆∆∆Pk
= 0.2 and ∆∆∆∆Dk-1= -0.35, is illustrated in Figure 4-6. The degree of fulfilment of Rule 1 is given
by:
DOF1= µPS (∆Pk) ∧ µNM (∆Dk-1) = 0.33 ∧ 0.25 (26)
Where Ù = minimum (AND) operator. The rule output z1, is given by the truncated
membership function NMM. Similarly, degrees of fulfilment for Rules 2, 3 and 4 are
evaluated using Equation (27) to give:
DOF1 = 0.33 (27)
DOF2 = 0.25 (28)
DOF3 = 0.5 (29)
Page 58
58
The corresponding outputs for Rules 2, 3 and 4 are the truncated membership functions NS’,
NMM’ and NS’ respectively, as shown in Figure 28. The total fuzzy output is the union of all
the component membership functions and is given by,
µPS ( Ζ ) = µNMM’ ( Ζ ) ∨∨∨∨ µNS’ ( Ζ ) ∨∨∨∨ µNMM’ ( Ζ ) ∨∨∨∨ µNS’ ( Ζ ) (30)
Where Ú = maximum (OR) operator. The fuzzy output ( ) out µ Z is shown in Figure 28.
DEFUZZIFICATION
The output of an inference process is a fuzzy set specifying a distribution space of fuzzy
control actions defined over an output universe of discourse. Defuzzification is the conversion
of this fuzzy output to crisp output suitable for a control action. A defuzzification strategy is
aimed at producing a non-fuzzy control action that best represents the possibility distribution
of an inferred fuzzy control action. Unfortunately, there is no systematic procedure for
choosing a defuzzification strategy. This process involves the operation:
Z0 = defuzzifier (Z) (31)
The term Z0 in Equation (31) is the non-fuzzy control output and defuzzifier is the
defuzzification operator. There are various defuzzification methods which include, centre of
area (COA), bisector, mean of maxima, sum of maxima, etc. The COA method commonly
known as centre of gravity is used in this work. The method is computationally intensive but
more accurate than other methods. In this method, the crisp output Z0of the Z variable is taken
to be the geometric centre of the output fuzzy value µout ( Z ), where µout ( Z ) is formed by
taking the union of all the contributions of rules whose DOF > 0 as shown in Figure 4-6. The
general expression for the COA method in the case of a discretized universe of discourse is
given by,
(32)
Page 59
59
Figure 28-Fuzzy inference and defuzzification using Mamdani method
A crisp value for the change in duty cycle is calculated by applying Equation (32) to the
output fuzzy value µout ( Z ) in Figure 28.
Page 60
60
SIMULATION OF TYPE-1 FUZZY CONTROL FOR MPPT
Using matlab program and feeding the data manually in the program we got the simulation
result for the duty cycle signal.
Figure 29-Fuzzy Membership function for Power
Figure 30-Fuzzy Membership function of output
Page 61
61
Figure 31- Final output of Duty Cycle Signal
Page 62
62
CHAPTER 6: TYPE-2 FUZZY LOGIC AND DESIGN OF
CONTROLLER FOR MPPT ALGORITHM
Type-2 fuzzy sets were introduced by Zadeh in 1975 as an extension of type-1 fuzzy sets.
Mendel and Karnik have developed the theory of type-2 fuzzy sets further. The theoretical
background of interval type-2 fuzzy system and its design principles are described. T2FLSs
appear to be a more promising method than their type-1 counterparts for handling
uncertainties such as noisy data and changing environments. The effects of the measurement
noise in type-1 and type-2 FLCs (T2FLCs) and identifiers are simulated to perform a
comparative analysis. It is concluded that the use of T2FLCs in real world applications which
exhibit measurement noise and modeling uncertainties can be a better option than type-1
FLCs (T1FLCs).
When a system has large amount of uncertainties, T1FLSs may not be able to achieve the
desired level of performance with a reasonable complexity of the structure. In such cases, the
use of T2FLSs is suggested as the preferable approach in the literature in many areas, such as
forecasting of time-series, controlling of mobile robots, and the truck backing-up control
problem are discussed. It is shown that when the parameters are tuned properly, T2FLSs can
result in a better ability to predict as compared to T1FLSs. T2FLS is applied to real time
mobile robots for indoor and outdoor environments. The real time implementation studies
show that a traditional T1FLC cannot handle the uncertainties in the system effectively and a
T2FLC using type-2 fuzzy sets results in a better performance. Moreover, with the latter
approach the number of rules to be determined may be reduced (it should be noted that this
may not mean a corresponding decrease in the parameters to be updated).
Type-2 FLSs use type-2 fuzzy sets which are described by membership functions which
themselves are fuzzy. This allows Type-2 FLSs to model and handle the uncertainty of
measurement and any rule uncertainty. Examples are the variability of expert opinion on a
fuzzy set, and their self-referencing variability over time; opinions do change. Noise of the
system and errors of measurement also have an effect.
According to a nonlinear current-voltage characteristic Photovoltaic for exploitation of this
optimization we need to track maximum output instantly. The aim of this paper is to search
maximum power point (MPP) Based on the Type2 Fuzzy logic controller (T2FLC) which is a
novel method in maximum power point tracking (MPPT). Solar cells’ MPP varies with solar
insolation and ambient temperature. With the improved efficiencies of power electronics
Page 63
63
converters, it is now possible to operate photovoltaic (PV) power systems at its MPP in order
to improve the overall system efficiency. The results of simulation show that the T2FLC
significantly improves the robustness of controller’s PV during the tracking phase as
compared to a conventional type1 fuzzy logic control (T1FLC) in present of noise in
photovoltaic power systems. This noise maybe insert to system when voltage and current are
measured. We show that the results obtained with T2FLC are better than the ones obtained
with the T1FLC method.
TYPE-2 FUZZY SETS
A type-2 fuzzy set, Ã may be represented as,
à = {((x, u), µÃ (x, u))| ∀x ∈ X ∀u ∈ Jx ⊆ [0, 1]} (33)
Where, µÃ (x, u) is the type-2 fuzzy membership function in which 0 ≤ µÃ (x, u) ≤ 1. à can
also be defined as
(34)
Where, ∫ ∫ denotes union over all admissible x and u.
Jx is called primary membership of x. There is a secondary membership value corresponding
to each primary membership value that defines the possibility for primary memberships.
Whereas the secondary membership functions can take values in the interval of [0, 1] in
generalized T2FLSs, they are uniform functions that only take on values of 1 in interval
T2FLSs. Since the general T2FLSs are computationally very demanding, the use of interval
T2FLSs is more commonly seen in the literature, due to the fact that the computations are
more manageable.
If the circumstances are so fuzzy, the places of the membership functions may not be
determined precisely. In such cases, the membership grade cannot be determined as a crisp
number in [0, 1], then the use of type-2 fuzzy sets might be a preferable option.
In Figure 29, the membership function does not have a single value for a specific value of x.
The values that the vertical line intersects the membership functions do not need all is
weighted same. Moreover, an amplitude distribution can be assigned to all of those points.
Hence, a three-dimensional membership function-a type-2 membership function- that
Page 64
64
characterizes a type-2 fuzzy set is created if the amplitude distribution operation is done for
all x ∈ X.
The footprint of uncertainty (FOU), the union of all primary memberships, is said to be the
bounded region that represents the uncertainty in the primary memberships of a type-2 fuzzy
set (Figure 29). An upper membership function and a lower membership function are two
type-1 membership functions that are the bounds for the FOU of a type-2 fuzzy set.
Figure 32-A Gaussian type-2 fuzzy membership function (FOU)
INTERVAL TYPE-2 FUZZY SETS
When all µÃ (x, u) are equal to 1, then à is an interval T2FLS. The special case of Equation
(35) might be defined for the interval T2FLSs:
(35)
As general T2FLS has huge computational burden. So, interval T2FLSs is commonly used in
literature. Both the general and interval type-2 fuzzy membership functions are three-
dimensional. As can be seen from Figure 33, the only difference between them is that the
secondary membership value of the interval type-2 membership function is always equal to 1.
Page 65
65
TYPE-2 FUZZY LOGIC SYSTEM BLOCK DIAGRAM
The type-2 FLSs are shown in Figure 30. As can see from Figure 30, an additional block
(type reduction) is needed in type-2 FLS design. Although the structure in Figure 30 brings
some advantages when dealing with uncertainties, it also increases the computational burden.
The followings are the basic blocks of a T2FLS:
Fuzzifier: The fuzzifier maps crisp inputs into type-2 fuzzy sets which activates the inference
engine.
Rule base: The rules in T2FLS remains the same as in T1FLS, but antecedents and
consequents are represented by interval type-2 fuzzy sets.
Inference: Inference block assigns fuzzy inputs to fuzzy outputs using the rules in the rule
base and the operators such as union and intersection. In type-2 fuzzy sets, join (⊔) and meet
operators (⊓), which are new concepts in fuzzy logic theory, are used instead of union and
intersection operators. These two new operators are used in secondary membership functions.
Type-reduction: The type-2 fuzzy outputs of the inference engine are transformed into type-
1 fuzzy sets that are called the type-reduced sets. There are two common methods for the
type-reduction operation in the interval T2FLSs: One is the Karnik- Mendel iteration
algorithm, and the other is Wu-Mendel uncertainty bounds method. These two methods are
based on the calculation of the centroid.
Defuzzification: The outputs of the type reduction block are given to defuzzification block.
The type-reduced sets are determined by their left end point and right end point, the
defuzzified value is calculated the average of these points.
Page 66
66
Figure 33-T2FLS block diagram
TYPE2 FUZZY LOGIC CONTROLLER
A T2FLC composed of four basic elements: the type1 fuzzyfier, the fuzzy rule-base, the
inference engine, and the type1 defuzzifier. The fuzzy rule-base is a collection of rules, which
are combined in the inference engine, to produce a fuzzy output. Table1 shows these rule
bases with Lower Standard Deviation (LSD) and Upper Standard Deviation (USD) which are
one of important characteristics of T2FLC, which these shows in Fig. 4. The type1 fuzzifier
maps the crisp input into type1 fuzzy sets, which are subsequently used as inputs to the
inference engine, whereas the type1 defuzzifier maps the type1 fuzzy sets produced by the
inference engine into crisp numbers. A T1FLCs are unable to handle rule uncertainties
directly, because they use type1 fuzzy sets that are certain. On the other hand, T2FLCs are
very useful in circumstances where it is difficult to determine an exact, and measurement
uncertainties. It is known that type 2 fuzzy set let us to model and to minimize the effects of
uncertainties in rule base FLC. Unfortunately, type2 fuzzy sets are more difficult to use and
understand than type1 fuzzy sets; hence, their use is not widespread yet. A T2FLC Ã is
characterized by the membership function:
à = {((x, u), µÃ (x, u))| ∀x ∈ X ∀u ∈ Jx ⊆ [0, 1]} (36)
(37)
Page 67
67
Figure 34-Structure of a type2 FLC
Hence, a type2 membership grade can be any subset in [0, 1], the primary membership, and
corresponding to each primary membership, there is a secondary membership (which can also
be in [0, 1]) that defines the possibilities for the primary membership. This uncertainty is
represented by a region called footprint of uncertainty (FOU) which can be described in terms
of an upper membership function (UMF) and a lower membership function (LMF) as shown
in Fig. 32.
Figure 35-Membership function of input
Page 68
68
Figure 36-Membership function of output
FUZZIFICATION AND THE RULES
The fuzzifier maps a crisp point x=(x1,..........,xp)T
∈ X1 × X2 ×........ × XP ≡ X into a type2
fuzzy set ÃX in X. As a type2 singleton fuzzifier in a singleton fuzzification with the input
fuzzy set having only a single point on nonzero membership is used, we have:
(38) The structure of rules in a T1FLC and a T2FLC is the same. However, in the latter the
antecedents and the consequents will be represented by T2FLCs. Suppose a T2FLC has p
inputs x1, x2, x3,...........,xp an output y, and a multiple input single output (MISO). It is
assumed that there are M rules and the lth rule in the type2 FLC can be written as follows:
(39)
INFERENCE
In T2FLC, the inference engine combines rules and gives a mapping from input T2FLCs to
output T2FLCs. It is necessary to compute the join Џ (unions), the meet Π (intersections), and
the extended sup-star composition (sup-star compositions) of type2 relations.
(40)
Page 69
69
Rl Rule is described by the membership function µRl (x, y) =4 µR
l(x1,........, xp, y) where:
(41)
In general, the p-dimensional input to Rl is given by the type2 fuzzy set Ãx whose membership function is:
(42) Where, Xt= (i=1; . . . ; p) are the labels of the fuzzy sets describing the inputs. Each rule Rl
determines a type2 fuzzy set Bl = Ax ◦Rl such that:
(43) This equation is the input / output relationship between the T2FSLC that activates one rule in
the inference engine and the T2FLC at the output of that engine as described in Fig. 3. In the
FLC we used interval T2FLCs and meet under product t-norm, so the result of the input and
antecedent operations, which are contained in the firing Set = ∏P i=1 µFi(xi’), is an interval
T1FLC set :
(44)
Where
(45)
Note * is the product operation.
TYPE-REDUCER AND DEFUZZIFICATION
The type-reducer generates a T1FLC output which is then converted into a crisp output
through the defuzzifier. This T1FLC is also an interval set. For the case of our FLC we used
center of sets (cos) type reduction, Ycos which is given by:
Page 70
70
Ycos = [yl, yr] =
(46)
This interval set is determined by its two end points, yl and yr, which correspond to the
centroid of the type2 interval consequent set Gi expressed as:
(47)
Before the computation of Ycos (X), we must evaluate (47), and its two end points, yl and yr.
Let the values of fi and yi that are associated with yl are denoted by fil and yi
l, respectively, and
the values of fi and yi that are associated with yr are denoted by frl and yr
l, respectively, then
from (46), we have:
(48)
(49)
From the type-reducer we obtain an interval set Ycos (X), to defuzzify it we use the average of
yl and yr, so the defuzzified output of an interval singleton T2FLS is:
(50)
Page 71
71
SIMULATION OF TYPE-2 MPPT FUZZY CONTROLLER
Using matlab program and feeding the data manually in the program we got the simulation
result for the duty cycle signal.
Figure 37- Fuzzy Membership function of Power
Figure 38-Fuzzy Membership function of Voltage
Page 72
72
Figure 39-Fuzzy Membership function of Duty cycle
Figure 40-Final output of Duty Cycle Signal in Type-2 Fuzzy Controller
Page 73
73
CHAPTER7: CONCLUSION AND FUTURE WORK
CONCLUSION
The aim of this thesis was to develop a method to optimize the energy extraction in a
photovoltaic power system. The concept of PV module maximum power point tracking has
been presented and various methods of addressing existing challenges are explored. A fuzzy
logic based algorithm for tracking the maximum power is proposed in this work. In order to
formulate and implement the algorithm, a system model is needed. The various components
and subsystems are analyzed, modeled, validated, and combined together to produce a
complete maximum power point tracker model. A hardware implementation was then carried
out to determine the performance of the algorithm in a practical setup.
Analysis of different dc-dc converter topologies showed that the buck-boost topology is the
most suitable for a maximum power tracker. The PV module and the buck-boost converter
were modeled and validated in Simulink while the fuzzy logic algorithm was formulated
using the Fuzzy Logic Toolbox in Matlab. The complete Maximum Power Point Tracker
model was formed by combining the PV module and the converter model with the fuzzy logic
controller. The MPPT model was used to tune the fuzzy logic controller rules and
membership functions. The PV module model was found to be sufficiently accurate and can
model any solar panel using information supplied in manufacturer data sheets. Simulation
results show that the proposed fuzzy logic algorithm has an average efficiency of 98% under
rapidly varying conditions and in the presence of measurement noise. The results show that
compared to other MPPT techniques, it provides improved performance in terms of
scillations about the maximum power point, speed and sensitivity to parameter variation. This
is possible since fuzzy logic controller rules can be assigned separately for the various regions
of operation resulting in effective small-signal and large-signal operation.
A hardware design and implementation of the MPPT was then carried out in order to test the
performance of the proposed algorithm. The design was broken down into several smaller
components, which are described in turn. For each subcircuit, a thorough description of the
relevant design issues and decisions was provided. Experimental results show that the
proposed algorithm is able to transfer peak power from a PV module to the load. The results
indicate that a significant amount of additional energy can be extracted from a photovoltaic
module by using a fuzzy logic based maximum power point tracker. This results in improved
Page 74
74
efficiency for the operation of a photovoltaic power system since batteries can be sufficiently
charged and used during periods of low solar radiation. The improved efficiency is expected
to lead to significant cost savings in the long run.
FUTURE WORK
The overall control goal in a photovoltaic power system is not to deliver the maximum
amount of possible power to the load or batteries, but only as much as is needed at any given
time; this can be referred to as power matching. Usable power that is not extracted from a
particular solar panel (because it is not operating at its current maximum power point) is
dissipated as heat through its surface. In a practical pplication like a water pump in a remote
village, the MPPT circuit must include a way to determine how much power is needed at any
given time and perform maximum power tracking only to a level where power requirements
are met. The MPPT should also include a second voltage regulation stage that maintains a
steady output regardless of variations in the load demand. These improvements will greatly
boost the functionality of the system in a practical setup.
As part of future work, ways of implementing a fuzzy logic algorithm in a dedicated single-
chip microcontroller needs to be addressed. The control algorithm is fairly complex and
modifications need to be made in order to meet the limited memory space and speed of
microcontrollers. The microcontroller must also incorporate timers, PWM input and output,
A/D and D/A interfaces, and interrupts for timing control and communications. The overall
circuit also needs to be modified to include ways of supplying power to the control circuit
using batteries charged by the MPPT. The redesigned circuit should be implemented in a
printed circuit board.
On the whole, it is concluded that the overall objective of formulating and implementing a
fuzzy logic based maximum power point tracker for a photovoltaic power system has been
met. Although there is a large amount of work that can and should still be done, the work in
this thesis has created a solid foundation to allow that work to continue.
Page 75
75
REFERENCE
1. Ocran, T.A. et al, Artificial Neural Network Maximum Power Point Tracker for Solar
Electric Vehicle, Tsinghua Science & Technology, Vol. 10, No. 2, pp. 204-208, 2005.
2. Hua C., Lin J., A modified tracking algorithm for maximum power tracking of solar
array, Energy Conversion and Management 45 (2004) 911-925, Elsevier Ltd. 2003.
3. Salas V., et al, New algorithm using only one variable measurement applied to a
maximum power point tracker, Solar Energy Materials &Solar Cells 87 (2005) 675-
684, Elsevier B.V. 2004.
4. C. Hua and C. Shen, Comparative study of peak power tracking techniques for solar
storage systems, in Proc. IEEE Appl. Power Electron. Conf. and Expo., Feb. 1998,
vol. 2, pp. 676–683.
5. Koutroulis E., Kalaitzakis K., Voulgaris N.C., Development of a Microcontroller
Based Photovoltaic Maximum Power Point Tracking Control System, IEEE
Transactions on Power Electronics, Vol. 16, No. 1, 2001.
6. J. H. R. Enslin, D. B. Snyman, Simplified feed-forward control of the maximum power
point tracker for photovoltaic applications, Proc. Int.Conf. IEEE Power Electron.
Motion Control, 1992, vol. 1, pp. 548–553.
7. M. Bodur and M. Ermis, Maximum power point tracking for low power photovoltaic
solar panels, in Proc. IEEE Electro Tech. Conf., 1992, vol. 2, pp. 758–761.
8. C. R. Sullivan and M. J. Powers, A high-efficiency maximum power point trackers for
photovoltaic array in a solar-powered race vehicle, in Proc.IEEE PESC, 1993, pp.
574–580.
Page 76
76
9. M. Veerachary, T. Senjyu, and K. Uezato, Feed-forwardmaximum powerpoint
tracking of PV systems using fuzzy controller, IEEE Trans. Aerosp. Electron. Syst.,
vol. 38, no. 3, pp. 969–981, Jul. 2002.
10. Ocran, T.A. et al, Artificial Neural Network Maximum Power Point Tracker for Solar
Electric Vehicle, Tsinghua Science & Technology, Vol. 10, No. 2, pp. 204-208, 2005.
11. Bidyadhar Subudhi, and Raseswari Pradhan, A Comparative Study on Maximum
Power Point Tracking Techniques for Photovoltaic Power Systems, IEEE
TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 4, NO. 1, JANUARY 2013
12. Qilian Liang and Jerry M. Mendel, Interval Type-2 Fuzzy Logic System: Theory and
Design, IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 8, NO. 5,
OCTOBER 2000
13. Chian-Song Chiu, T-S Fuzzy Maximum Power Point Tracking Control of Solar
Power Generation Systems, IEEE TRANSACTIONS ON ENERGY CONVERSION,
VOL. 25, NO. 4, DECEMBER 2010
14. Pongsakor Takun, Somyot Kaitwanidvilai and Chaiyan Jettanasen, Maximum Power
Point Tracking using Fuzzy Logic Control for Photovoltaic Systems, Proceeding of the
International MultiConference of Engineers and Computer Scientists 2011 Vol II,
IMECS 2011, March 16-18, 2011, Hong Kong
Page 77
77
APPENDICES
A-1 MATLAB CODE FOR TYPE-1 FUZZY MPPT SYSTEMS
clear all;
clc;
seed(1,:)=[-1.41872966171615,-1.50937336396472,1.35940535370735,0.552050153997157,-3.56300846530129,-3.92921289153443,3.59350530002252,2.91377613583114,-4.06830134654488,-9.38315840253697,5.33163240608823,2.55427454937803];
x=-5:0.01:5;
m11=seed(1,1);
nmf_1=1.*(x>=-5 & x<=-2)+((m11-x)/(m11+2)).*(x>-2 & x<=m11)+0.*(x>m11);
m12=seed(1,2);
m13=seed(1,3);
mmf_1=0.*(x>=-5 & x<m12)+((m12-x)/m12).*(x>=m12 & x<=0)+((m13-x)/m13).*(x>0 & x<=m13)+0.*(x>m13);
m14=seed(1,4);
pmf_1=0.*(x>=-5 & x<=m14)+((m14-x)/(m14-2)).*(x>m14 & x<=2)+1.*(x>2);
plot(x,nmf_1,x,mmf_1,x,pmf_1);
m21=seed(1,5);
nmf_2=1.*(x>=-5 & x<=-4)+((m21-x)/(m21+4)).*(x>-4 & x<=m21)+0.*(x>m21);
m22=seed(1,6);
m23=seed(1,7);
mmf_2=0.*(x>=-5 & x<m22)+((m22-x)/m22).*(x>=m22 & x<=0)+((m23-x)/m23).*(x>0 & x<=m23)+0.*(x>m23);
m24=seed(1,8);
pmf_2=0.*(x>=-5 & x<=m24)+((m24-x)/(m24-4)).*(x>m24 & x<=4)+1.*(x>4);
figure,plot(x,nmf_2,x,mmf_2,x,pmf_2);
x1=-10:0.01:10;
Page 78
78
m31=seed(1,9);
nmf_3=((m31-x1)/(m31+10)).*(x1>=-10 & x1<=m31)+0.*(x1>m31);
m32=seed(1,10);
m33=seed(1,11);
mmf_3=0.*(x1>=-10 & x1<m32)+((m32-x1)/m32).*(x1>=m32 & x1<=0)+((m33-x1)/m33).*(x1>0 & x1<=m33)+0.*(x1>m33);
m34=seed(1,12);
pmf_3=0.*(x1>=-10 & x1<=m34)+((m34-x1)/(m34-10)).*(x1>m34 & x1<=10);
figure,plot(x1,nmf_3,x1,mmf_3,x1,pmf_3);
N=[1.38482138927092,0.331695099753793,-0.240571581094563,2.20154050244822,-0.985997992943328,0.797006278255187,2.54603755463328,-0.520374576999944,0.798959298794457,0.402293283259269,-0.438415331019449,-0.00447452677584946,0.777744982882119,2.93138207608262,1.11594667083867,-0.307285420662612,-2.48401344836948,0.313548106522630,0.328551210929899,-0.996415148210786,1.27067528727686,0.492079604011546,0.738863773478882,-0.612072816570621,-0.860921659303211,-0.588452929194514,-1.39813810236020,-0.657840044896512,0.915040345406705,-2.83931333524345,-2.23078107238103,0.0334746917300934,0.413548212695413,-1.19450604408623,0.675340681022256,1.58501469306781,0.860781214720267,-1.96819725194340,-0.919725180931747,-1.07488393366207,-0.426464469193661,-0.845097270811198,-1.05953001399227,2.40443850189278,-0.419183396909719,-0.191893646091304,0.360508723751472,-2.18073828376819,-1.43723087054719,0.591524735185831,0.442373364363211,-1.46163551257501,1.41711950219533,-0.0840194870036627,0.178377958396273,0.137579969481291,-0.692374305594608,-2.30544630152871,-0.164764290116440,-1.97018180362483,-0.207303081972442,0.261033736690033,0.249219702822809,-1.74521294286126,0.461522032143667,-1.52388653756162,0.316836855719714,0.759385473279033,2.01227378752132,-2.58180950742198,1.04397922994940,-0.204141190654463,-0.0948098401765951,-0.708024898446485,0.294783332926228,-0.325581450216494,-0.778977027986079,-0.678064431097863,0.775853002347159,0.618921803752649,1.38559834740879,0.456920211723040,-0.183669740772424,-1.43312498597510,0.396289483615839,-0.385967739623539,0.278871258925728,1.10103051869061,-0.620458179923327,-0.607129928473020,-0.326305208067380,0.267642873780794,2.56435187859992,-1.09890948557486,1.68686800418637,-0.271213340121641,1.13038441475108,0.0751663592513872,-0.672532835522489,-
Page 79
79
0.326313069554429,1.20715091109962,1.59478856853022,1.20985451794234,-0.709733517885276,-1.62253316166248,0.504662521251894,0.664687758803999,-3.16311397468132,0.0305049438818347,2.63815385047041,-0.237227564710709,-0.810989836660408,-1.26870294985077,-2.27986325737011,-0.616168452531363,0.379476424687043,-1.23866577748961,0.319095006076758,0.991993482703499,-0.746915320515773,0.619647796416560,0.711903023399526,-0.0538889834616139,2.46302275482213,-1.60160766888651,0.346287672168541,-1.16670174165339,-0.908644284160093,0.555153221342472,1.37587269088625,0.214147065107326,1.64270854545363,-1.17096593249108,0.425715806624958,-0.668355859092558,-0.953455700447712,-3.31293774445661,0.957012593902646,2.36029663518581,0.629010196960602,1.73268806468481,0.0742519574849388,-1.54900649455698,-1.09837682340403,-1.01049651017435,1.38780936079022,-0.604739218714607,0.115119018704842,-1.67833464939734,-0.230465085880884,-2.04684504233435,-0.884298841604003,0.956144095483507,2.15232623483605,0.611134208728999,0.122415697272733,1.35115572576624,0.916522195581153,-3.08294560475902,-2.25544875236212,-1.04913259864545,-1.09980791326250,-2.01457100885465,-0.853870757405701,-1.17228176747221,1.38509687958932,-0.767913447832517,-1.60640927798435,-0.524048117771788,-0.862322890137972,0.330040567392754,-0.291171495747007,-1.87015498579144,0.730583582771480,-0.458187629546408,-0.755392975482703,-1.07234780120883,0.999445753113964,-0.137134537773299,-1.95487580833494,2.73286818905448,-0.430169976298140,0.341241619992428,0.446700891720196,-1.64481849451659,1.53732736004414,0.316138108925763,0.180849151231216,1.70035504065231,0.200288920770145,-2.16822816746923,-0.0202180624284108,1.46915766449766,1.52661429361812,0.333207355174898,-2.13195674555119,1.38033890677280,-0.0722427447820588,0.241578668250488,1.61302629291383,0.507040819164125,-4.68598632713944,-2.09451999908524,-1.04240098247441,-1.62564163023764,0.193199883541485,0.191305877638837,-0.872577808736155,1.47225988375540,-0.992137065435284,-0.435075815807036,-0.840179042254998,-1.12858042648021,0.0687588334093583,-1.75513853204869,-0.393026027672157,1.04531647679078,1.02448855076122,-2.44996562819912,0.512029121886063,-0.576567692922137,0.567191547821422,0.216684387745333,-0.397921416741962,-0.0167642343395225,0.765303717680543,0.914010981885747,-0.943802840946499,-1.67103885400230,-0.527555626199160,1.28410055389022,0.431575334880573,0.496015609411085,-0.557116242203661,-0.963908547825090,2.61482895891171,-0.132394911920200,-0.0418560718158022,-0.344563907114072,0.0877658996804111,-0.402238830184576,-1.31459134111788,0.620578141078599,1.81258652177899,0.232736050551499,-0.936828193582948,0.0710388883600386,1.28448596066744,2.24484441868597,-
Page 80
80
1.76909706539437,0.143530009393017,-0.545472219995132,0.685612789422788,0.212653221227279,-0.741039005791648,-2.27511514718797,-1.52855613527878,-0.763412876820536,-0.344570600821842,1.48017292741012,-0.748521835886504,0.468362944669008,1.14036432368427,0.291646396072781,0.106754322079998,-0.768633679355067,1.05915862292141,2.05030769543055,0.598666879262569,0.386885662668656,0.296313001682898,1.30177733034342,-0.416575211880557,-0.644812776504438,0.669576876548693,-0.385837054714682,1.12273950769089,-1.84290166633205,1.66979591137321,-0.919794957209801,0.300315920938156,-0.589025122794665,0.898332180141968,-1.78452205387631,-1.34364320022215,2.05504428231793,-1.22794200952554,-0.879185134153416,0.0784525833047537,-1.00550998703650,0.184692752840272,1.48040320478173,0.0864680037024968,2.17313990076000,-1.82511473183257,-0.358344231145863,-1.10035312328395,0.463914261232699,-1.58834651728004,-2.21519823700076,-0.495428858200163,-2.28499688781302,-1.13064733767543,-0.689012644380276,-2.10289906069769,0.709932431204590,0.384743166507179,-0.789533445080592,-1.75285118036256,-0.346576661696050,-0.211310789339244,-2.73557842264944,1.64594116804254,0.548299403368277,-2.20758347930695,2.49025451547024,-0.193303674254264,-1.85488901966005,-1.16582117315815,0.330728774934612,-1.81259403855072,-0.596157138402061,-0.127788291412250,-0.535004691627498,0.256401389690463,-0.896371835959952,-0.315381261983667,0.240111530182619,-2.79492386598159,-2.41487958140448,-1.34337210024082,0.100530022799672,0.255303364958176,-0.920427480834600,1.65225589947731,1.42636788706646,-0.222963825631180,4.18660155874998,-0.459898019136465,1.23801784891321,-0.0793394582787517,-2.17145983927161,2.36657126583045,-1.33893204834467,0.0928219928919509,2.41816270189124,0.809150578909097,0.467057143698821,-0.183072725021343,0.308690901185450,2.37958051500274,-0.580314306138870,-0.278901121813475,-0.573538914351919,-0.518739490720163,-0.217141466533065,0.779116367510426,0.111049139965645,-0.957689937883396,2.07969626585448,-1.09472340320257,-0.415995060539202,0.100836660136994,-2.11777085160188,-1.27035859514546,-1.62862378691149,-3.35415115093708,2.68528791471479,-0.919460493323576,1.05757222390175,0.0216460661706082,-1.25154600732988,-0.0508477475736162,0.509543225470446,-1.43871703485559,1.50832413484387,1.50673068429782,-0.122589722632457,0.773417560248220,-1.53412783839406,0.719946448584196,-0.161532745972635,2.02500412701583,-1.20488369102054,2.53234006762520,0.0252853577291452,0.560861055158647,1.82484927445748,1.38853221483562,-1.10264690252973,-1.15221817338068,-1.15169268362569,1.35303045937092,-0.170833770910056,-0.559857820407075,0.103882776922012,1.30395483367127,-
Page 81
81
0.200030250311312,0.389363198677451,-2.91492212093074,-0.840438292722718,-1.34275121118885,-0.804649766806675,-0.959260445230660,-0.531399691881970,-0.618545443970905,0.828052290457057,1.80792141618184,-1.93249207041947,2.20670052420605,1.61526532213254,0.117888821361295,1.98922656795367,0.308164390102684,-1.28442413489953,1.41986729480852,2.81051157936995,-0.974719340939937,-1.26076002675356,1.43306796735528,-0.0301532636828884,-1.71696313429771,1.55642986030246,-0.193969365712743,-1.00110577936122,-0.732156504386608,-0.626521851553325,1.18015193829293,-1.45497459431202,0.590717841827051,1.26447899039292,-1.08247772869104,0.420967691671818,-2.08159440885561,0.877318207824867,-1.41277796500988,1.01515124514750,0.545826078532877,-0.517921878218660,-1.86304365511266,1.67636849360081,0.192241494966741,2.11159460969132,0.0107828131199507,-2.23885434740051,2.12766776895631,-0.366836682891734,1.16375091168049,-0.909246647359768,0.770254099390070,1.05118819463518,1.27539984142651,0.628265367935514,-0.0513779150639639,0.181886703562448,0.380629795439442,-0.929668244170861,0.692941793154290,-1.34009389481400,-2.39872855038055,0.209693824619781,0.248929633576670,0.277890539078031,2.23187821257961,1.55210707722581,2.34637631838160,0.277405016835427,1.14390426417808,-0.744239213853409,-0.290786408320864,-2.20031142724091,0.0879135419050389,1.16699980796735,-1.33216471111545,-0.280091458826250,-0.911564512656003,-1.01772574216173,1.13604608478516,0.0846842332222333,0.931092727553606,0.499510520214393,1.17809172900377,-2.84165143411804,-1.28677646228305,-0.216390149826244,0.437995425035824,1.44579088690221,-0.805238556680096,-0.761583686530278,-0.358988895162106,0.875230733052720,0.133993444040174,1.41717441814692,-0.971804688293527,0.517586590556182,0.487441365598103,-2.06968315539206,0.373131409047017,2.19014455836489,-1.50792224073924];
N=.001*N;
y1(1)=-1.5;
y2(1)=-3;
t=1:500;
n=length(t);
u(1)=0;
for k=2:n
p1=y1(k-1);
Page 82
82
p2=y2(k-1);
if(p1>2)
p1=2;
end
if(p1<-2)
p1=-2;
end
if(p2>4)
p2=4;
end
if(p2<-4)
p2=-4;
end
n12=nmf_1(uint16(100*p1+501));
z12=mmf_1(uint16(100*p1+501));
p12=pmf_1(uint16(100*p1+501));
n22=nmf_2(uint16(100*p2+501));
z22=mmf_2(uint16(100*p2+501));
p22=pmf_2(uint16(100*p2+501));
p=max(max(min(p12,p22),min(p12,z22)),min(z12,p22));
z=max(max(min(n12,p22),min(z12,z22)),min(p12,n22));
n=max(max(min(n12,z22),min(z12,n22)),min(n12,n22));
mf=max(min(n,nmf_3),max((min(p,pmf_3)),(min(z,mmf_3))));
Page 83
83
num=0.0;
den=0.0;
for k1=1:2001
num=num+x1(k1)*mf(k1);
den=den+mf(k1);
end
u(k)=num/den;
y1(k)=y1(k-1)+0.05*y2(k-1)+N(k);
y2(k)=0.05*(y1(k-1)+N(k))+y2(k-1)-u(k);
end
sum=0;
for k=1:500
sum=sum+(y1(k)*y1(k));
end
RMS=sqrt(sum);
subplot(311),plot(t,y1)
title('Signal response')
subplot(312),plot(t,y2)
title('Error-dot response')
subplot(313),plot(t,u),xlabel('Time(s)')
title('Control signal')
Page 84
84
A-2 MATLAB CODE FOR TYPE-2 FUZZY MPPT SYSTEMS
clear all;
clc;
seed(1,:)=[-1.44486191714372,-1.33296143708818,-0.0637895310191314,0,0,0.0338396663311227,0.264521471975207,0.318430228036970,-4,-3.51852513113980,-1.21733109857411,-0.655547464226103,1.54298297536309,1.73450196171244,2.06437212068236,2.38667974339272,-10,-9.29849944705924,-8.46461202642989,-7.51611612278146,0,1.66721382815816,3.92216833878827,4.06015421266160];
x=-5:0.01:5;
m11=seed(1,4);
numf_1=1.*(x>=-5 & x<=-2)+((m11-x)/(m11+2)).*(x>-2 & x<=m11)+0.*(x>m11);
m12=seed(1,1);
m13=seed(1,8);
mumf_1=0.*(x>=-5 & x<m12)+((m12-x)/m12).*(x>=m12 & x<=0)+((m13-x)/m13).*(x>0 & x<=m13)+0.*(x>m13);
m14=seed(1,5);
pumf_1=0.*(x>=-5 & x<=m14)+((m14-x)/(m14-2)).*(x>m14 & x<=2)+1.*(x>2);
m21=seed(1,12);
numf_2=1.*(x>=-5 & x<=-4)+((m21-x)/(m21+4)).*(x>-4 & x<=m21)+0.*(x>m21);
m22=seed(1,9);
m23=seed(1,16);
mumf_2=0.*(x>=-5 & x<m22)+((m22-x)/m22).*(x>=m22 & x<=0)+((m23-x)/m23).*(x>0 & x<=m23)+0.*(x>m23);
m24=seed(1,13);
pumf_2=0.*(x>=-5 & x<=m24)+((m24-x)/(m24-4)).*(x>m24 & x<=4)+1.*(x>4);
Page 85
85
x1=-10:0.01:10;
m31=seed(1,20);
numf_3=((m31-x1)/(m31+10)).*(x1>=-10 & x1<=m31)+0.*(x1>m31);
m32=seed(1,17);
m33=seed(1,24);
mumf_3=0.*(x1>=-10 & x1<m32)+((m32-x1)/m32).*(x1>=m32 & x1<=0)+((m33-x1)/m33).*(x1>0 & x1<=m33)+0.*(x1>m33);
m34=seed(1,21);
pumf_3=0.*(x1>=-10 & x1<=m34)+((m34-x1)/(m34-10)).*(x1>m34 & x1<=10);
m11=seed(1,3);
nlmf_1=1.*(x>=-5 & x<=-2)+((m11-x)/(m11+2)).*(x>-2 & x<=m11)+0.*(x>m11);
m12=seed(1,2);
m13=seed(1,7);
mlmf_1=0.*(x>=-5 & x<m12)+((m12-x)/m12).*(x>=m12 & x<=0)+((m13-x)/m13).*(x>0 & x<=m13)+0.*(x>m13);
m14=seed(1,6);
plmf_1=0.*(x>=-5 & x<=m14)+((m14-x)/(m14-2)).*(x>m14 & x<=2)+1.*(x>2);
plot(x,numf_1,x,mumf_1,x,pumf_1,x,nlmf_1,x,mlmf_1,x,plmf_1);
m21=seed(1,11);
nlmf_2=1.*(x>=-5 & x<=-4)+((m21-x)/(m21+4)).*(x>-4 & x<=m21)+0.*(x>m21);
m22=seed(1,10);
m23=seed(1,15);
mlmf_2=0.*(x>=-5 & x<m22)+((m22-x)/m22).*(x>=m22 & x<=0)+((m23-x)/m23).*(x>0 & x<=m23)+0.*(x>m23);
m24=seed(1,14);
plmf_2=0.*(x>=-5 & x<=m24)+((m24-x)/(m24-4)).*(x>m24 & x<=4)+1.*(x>4);
Page 86
86
figure,plot(x,numf_2,x,mumf_2,x,pumf_2,x,nlmf_2,x,mlmf_2,x,plmf_2);
m31=seed(1,19);
nlmf_3=((m31-x1)/(m31+10)).*(x1>=-10 & x1<=m31)+0.*(x1>m31);
m32=seed(1,18);
m33=seed(1,23);
mlmf_3=0.*(x1>=-10 & x1<m32)+((m32-x1)/m32).*(x1>=m32 & x1<=0)+((m33-x1)/m33).*(x1>0 & x1<=m33)+0.*(x1>m33);
m34=seed(1,22);
plmf_3=0.*(x1>=-10 & x1<=m34)+((m34-x1)/(m34-10)).*(x1>m34 & x1<=10);
figure,plot(x1,numf_3,x1,mumf_3,x1,pumf_3,x1,nlmf_3,x1,mlmf_3,x1,plmf_3);
N=[0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.3,0,0.5,0.1,0,0,0,0,0.1,-0.4,0,-0.2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0];
y1(1)=-1.5;
y2(1)=-3;
t=1:length(N);
n=length(t);
u(1)=0;
for k=2:n
p1=y1(k-1);
p2=y2(k-1);
if(p1>2)
p1=2;
end
if(p1<-2)
Page 87
87
p1=-2;
end
if(p2>4)
p2=4;
end
if(p2<-4)
p2=-4;
end
n11=numf_1(uint16(100*p1+501));
z11=mumf_1(uint16(100*p1+501));
p11=pumf_1(uint16(100*p1+501));
n12=numf_2(uint16(100*p2+501));
z12=mumf_2(uint16(100*p2+501));
p12=pumf_2(uint16(100*p2+501));
n21=nlmf_1(uint16(100*p1+501));
z21=mlmf_1(uint16(100*p1+501));
p21=plmf_1(uint16(100*p1+501));
n22=nlmf_2(uint16(100*p2+501));
z22=mlmf_2(uint16(100*p2+501));
p22=plmf_2(uint16(100*p2+501));
p_1=max(max(min(p12,p11),min(p11,z12)),min(z11,p12));
z_1=max(max(min(n11,p12),min(z11,z12)),min(p11,n12));
n_1=max(max(min(n11,z12),min(z11,n12)),min(n11,n12));
p_2=max(max(min(p22,p21),min(p21,z22)),min(z21,p22));
Page 88
88
z_2=max(max(min(n21,p22),min(z21,z22)),min(p21,n22));
n_2=max(max(min(n21,z22),min(z21,n22)),min(n21,n22));
umf=max(min(n_1,numf_3),max((min(p_1,pumf_3)),(min(z_1,mumf_3))));
lmf=max(min(n_2,nlmf_3),max((min(p_2,plmf_3)),(min(z_2,mlmf_3))));
%figure,plot(x1,umf,x1,lmf)
upperx=0;
upper=0;
for m=1:2001
upperx=upperx+umf(m)*x1(m);
upper=upper+umf(m);
end
cl=(upperx)/(upper);
cr=cl;
a=uint16(100*cl+1001);
lowleft=0;
lowleftx=0;
for m=1:a-2
lowleftx=lowleftx+lmf(m)*x1(m);
lowleft=lowleft+lmf(m);
end
lowright=0;
lowrightx=0;
for m=a+2:2001
Page 89
89
lowrightx=lowrightx+lmf(m)*x1(m);
lowright=lowright+lmf(m);
end
b=uint16(100*cl+1001);
while(b>1)
upperx=0;
upper=0;
lowerx=0;
lower=0;
for m=1:b
upperx=upperx+umf(m)*x1(m);
upper=upper+umf(m);
end
for m=b+1:a+1
lowerx=lowerx+lmf(m)*x1(m);
lower=lower+lmf(m);
end
c=(upperx+lowerx+lowrightx)/(upper+lower+lowright);
if (c<cl)
cl=c;
b=uint16(100*cl+1001);
else
break;
end
end
Page 90
90
b=uint16(100*cr+1001);
while(b<2001)
upperx=0;
upper=0;
lowerx=0;
lower=0;
for m=b:2001
upperx=upperx+umf(m)*x1(m);
upper=upper+umf(m);
end
for m=a-1:b-1
lowerx=lowerx+lmf(m)*x1(m);
lower=lower+lmf(m);
end
c=(upperx+lowerx+lowleftx)/(upper+lower+lowleft);
if (c>cr)
cr=c;
b=uint16(100*cr+1001);
else
break;
end
end
cl1(k)=cl;
cr1(k)=cr;
u(k)=(cl+cr)/2;
Page 91
91
y1(k)=y1(k-1)+0.08*y2(k-1)+N(k);
y2(k)=0.08*(y1(k-1)+N(k))+y2(k-1)-u(k);
end
sum=0;
for k=1:length(N)
sum=sum+(y1(k)*y1(k));
end
RMS=sqrt(sum);
figure,
subplot(311),plot(t,y1)
title('Signal response')
subplot(312),plot(t,y2)
title('Error-dot response')
subplot(313),plot(t,u),xlabel('Time(s)')
title('Control signal')