9/8/2019 IAENG Conference Publications: Proceedings Book of The International MultiConference of Engineers and Computer Scientists 2014 Vol. I www.iaeng.org/publication/details/IMECS2014_Proc_I.html 1/1 Proceedings Book of The International MultiConference of Engineers and Computer Scientists 2014 Vol. I ISBN:978-988-19252-5-1 ISSN:2078-0958 (Print) / 2078-0966 (Online) International MultiConference of Engineers and Computer Scientists 2014 (IMECS 2014). Price: US 40 (by surface) / US 60 (by air) This volume includes the papers of the below conferences: The 2014 IAENG International Conference on Artificial Intelligence and Applications The 2014 IAENG International Conference on Bioinformatics The 2014 IAENG International Conference on Bioinformatics Special Session: The IAENG International Workshop on Biomedical Engineering The 2014 IAENG International Conference on Computer Science The 2014 IAENG International Conference on Control and Automation The 2014 IAENG International Conference on Data Mining and Applications The 2014 IAENG International Conference on Imaging Engineering The 2014 IAENG International Conference on Internet Computing and Web Services The 2014 IAENG International Conference on Scientific Computing The 2014 IAENG International Conference on Software Engineering
42
Embed
9/8/2019 IAENG Conference Publications: …p3m.ppns.ac.id/wp-content/uploads/2019/09/2014-IMECS.pdfJoin IAENG Now! IAENG Membership is free. Our societies welcome committee members
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
9/8/2019 IAENG Conference Publications: Proceedings Book of The International MultiConference of Engineers and Computer Scientists 2014 Vol. I
Proceedings Book of TheInternationalMultiConference ofEngineers and ComputerScientists 2014 Vol. IISBN:978-988-19252-5-1
ISSN:2078-0958 (Print) / 2078-0966 (Online)
International MultiConference of Engineers andComputer Scientists 2014 (IMECS 2014).
Price: US 40 (by surface) / US 60 (by air)
This volume includes the papers of the belowconferences: The 2014 IAENG International Conference on ArtificialIntelligence and Applications The 2014 IAENG International Conference onBioinformatics The 2014 IAENG International Conference onBioinformatics Special Session: The IAENGInternational Workshop on Biomedical Engineering The 2014 IAENG International Conference onComputer Science The 2014 IAENG International Conference on Controland Automation The 2014 IAENG International Conference on DataMining and Applications The 2014 IAENG International Conference on ImagingEngineering The 2014 IAENG International Conference on InternetComputing and Web Services The 2014 IAENG International Conference onScientific Computing The 2014 IAENG International Conference on SoftwareEngineering
9/8/2019 International MultiConference of Engineers and Computer Scientists IMECS 2014
www.iaeng.org/IMECS2014/committee.html 1/6
Hong Kong, 12-14 March, 2014
Conferences Publications Membership About IAENG FAQ Contact Us
The IMECS 2014 Program Committee
As the IMECS is organized by the International Association of Engineers and the Engineering Letters, some ofthe editorial board members of the Engineering Letters and the IAENG journals and the committee members ofIAENG Societies are taking part in the IMECS 2014 International Program Committee.
IMECS MultiConference Co-chairs
Prof. Craig Douglas (honorary co-chair)
School of Energy Resources Distinguished Professor,Departments of Mathematics & Director, Institute for Scientific Computation, University ofWyoming, USAVisiting Professor and Associate Director, Center forNumerical Porous Media MethodsKing Abdullah University of Science & Technology, SaudiArabia & Former Senior Research Scientist (corresponding tononteaching full professor), Computer Science Department, Yale University, USA
Prof. David Dagan Feng (honorary co-chair)Head, School of Information Technologies, University of Sydney, Australia
Prof. Alexander M. Korsunsky (honorary co-chair)Professor of Engineering Science Former Dean, Trinity CollegeDepartment of Engineering Science, University of Oxford,UK
Prof. Ping-kong Alexander Wai (honorary co-chair)Vice President and Chair Professor of OpticalCommunicationsHong Kong Polytechnic University, Hong Kong
9/8/2019 International MultiConference of Engineers and Computer Scientists IMECS 2014
www.iaeng.org/IMECS2014/committee.html 2/6
Prof. Oscar CastilloProfessor and Research Director of Computer ScienceTijuana Institute of Technology, MexicoFormer Adjunct Professor of San Diego State University,USA President of HAFSA (Hispanic American Fuzzy SystemsAssociation)Vice-Chair of the Mexican Chapter of the ComputationalIntelligence Society (IEEE)
Prof. Jeong-A LeeProfessor of Department of Computer EngineeringChosun University, South Korea
IMECS Program Committee & Conferences Co-Chairs
Yousry H. Abdelkader Alexandria University, Egypt
Daoud Ait-Kadi University Laval, Canada
Basim Al-Najjar Vaxjo University, Sweden
Ali Allahverdi Kuwait University, Kuwait
Bala P. Amavasai Sheffield Hallam University, U.K.
Carlos Henggeler Antunes University of Coimbra, Portugal
Djamel Bouchaffra Oakland University, USA
Indranil Bose The University of Hong Kong, Hong Kong
Anthony Brabazon University College Dublin, Ireland
Mietek Brdys The University of Birmingham, UK
Fidel Cacheda University of A Coruna, Spain
Alan Hoi-shou Chan City University of Hong Kong, Hong Kong
Felix T.S. Chan The Hong Kong Polytechnic University, Hong Kong
Chin-Chen Chang Feng Chia University, Taiwan
Pranay Chaudhuri University of the West Indies, West Indies
9/8/2019 Proceedings of IMECS 2014, 12-14 March, 2014, Hong Kong, IAENG Open Access Publication
www.iaeng.org/publication/IMECS2014/ 1/29
Sunday, September 8, -1781
Proceedings of the International MultiConference ofEngineers and Computer Scientists 2014
IMECS 2014, 12-14 March, 2014, Hong Kong
The International MultiConference of Engineers and Computer Scientists has been organized bythe International Association of Engineers (IAENG), a non-profit international association for theengineers and the computer scientists. The IMECS 2014 takes place in The Royal Garden Hotel,Kowloon, Hong Kong, 12-14 March, 2014.
The focus of our conference is on the frontier topics in the theoretical and applied engineering andcomputer science subjects. The IMECS conferences have been serving as good platforms for ourmembers and the entire engineering community to meet with each other and to exchange ideas.The conferences have also stroke a balance between theoretical and application development.
Our IMECS 2014 has been organized with conference committees that have been formed withover three hundred committee members who are mainly research center heads, faculty deans,department heads, professors, and research scientists from over 30 countries. The conferences aretruly international meetings with a high level of participation from many countries.
The response that we have received for the multiconference is excellent. The IMECS 2014 hasattracted more than six hundred participants from over 50 countries. All submitted papers havegone through the peer review process. The summary of submissions and accepted papers inIMECS 2014 is as followed: total number of submissions reviewed: 683; total number of acceptedpapers: 350. And the overall acceptance rate in our multiconference is 51.24%.
Editors: S. I. Ao and Oscar Castillo and Craig Douglas and David Dagan Feng and Jeong-A Lee
9/8/2019 Proceedings of IMECS 2014, 12-14 March, 2014, Hong Kong, IAENG Open Access Publication
www.iaeng.org/publication/IMECS2014/ 13/29
Recipe Recommendation Method by Considering the User's Preference and Ingredient Quantity of Target Recipe
Mayumi Ueda, Syungo Asanuma, Yusuke Miyawaki, and Shinsuke Nakajima
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp519-523 [Online Full Text]
The 2014 IAENG International Conference on Scientific Computing
Hierarchical Modelling and X-ray Analysis of Human Dentine and Enamel
Tan Sui, Michael A. Sandholzer, Nikolaos Baimpas, Alexander J.G. Lunt, Igor P. Dolbnya, Jianan Hu, Anthony D. Walmsley, Philip J. Lumley,Gabriel Landini, and Alexander M. Korsunsky
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp524-528 [Online Full Text]
Shear Properties of Graphene Containing Nitrogen Atoms and Grain Boundaries Using Molecular Dynamics Simulations
Shingo Okamoto, and Akihiko Ito
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp529-534 [Online Full Text]
Numerical Method for the Heat Equation with Dirichlet and Neumann Conditions
A. Cheniguel
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp535-539 [Online Full Text]
Mobile Application for Field Knowledge Data of Urban River Catchment Decision Support System
Fathoni Usman, Rohayu Che Omar, and Lariyah Mohd Sidek
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp540-544 [Online Full Text]
Computational Modeling of Honeycomb Structures with Shape Memory Alloys
Y. Toi, and J. He
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp545-550 [Online Full Text]
On Positive Definite Solutions of the Linear Matrix Equation X + A*XA = I
Sana'a A. Zarea, Salah M. El-Sayed, and Amal A. S. Al-Marshdy
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp551-557 [Online Full Text]
The Hybrid Extragradient Method for the Split Feasibility and Fixed Point Problems
Jitsupa Deepho, Wiyada Kumam, and Poom Kumam
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp558-563 [Online Full Text]
9/8/2019 Proceedings of IMECS 2014, 12-14 March, 2014, Hong Kong, IAENG Open Access Publication
www.iaeng.org/publication/IMECS2014/ 19/29
Yuen-Haw Chang, Chin-Ling Chen, and Tzu-Chi Lin
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp791-796 [Online Full Text]
Design of a Novel High Gain Carbon Nanotube based Operational Transconductance Amplifier
Sajad A Loan, M. Nizamuddin, Faisal Bashir, Humyra Shabir, Asim. M. Murshid, Abdul Rahman Alamoud, and Shuja A Abbasi
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp797-800 [Online Full Text]
The 2014 IAENG International Conference on Electrical Engineering Special Session: Design, Analysis andTools for Integrated Circuits and Systems
Preface of the 2014 IAENG International Conference on Electrical Engineering Special Session: Design, Analysis andTools for Integrated Circuits and Systems
Ka Lok Man, and Nan Zhang
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp801-804 [Online Full Text]
Accelerating Financial Code through Parallelisation and Source-Level Optimisation
Nan Zhang, and Ka Lok Man
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp805-806 [Online Full Text]
Parallel Computation of Value at Risk using the Delta-Gamma Monte Carlo Approach
Nan Zhang, Ka Lok Man, and Eng Gee Lim
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp807-810 [Online Full Text]
Software Risk Management Practice: Evidence From Thai Software Firms
Tharwon Arnuphaptrairong
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp811-817 [Online Full Text]
Weighted Type of Quantile Regression and its Application
Xuejun Jiang, Tian Xia, and Dejun Xie
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp818-822 [Online Full Text]
Using Coh-Metrix to Analyse Writing Skills of Students: A Case Study in a Technological Common Core CurriculumCourse
Chi-Un Lei, Ka Lok Man, and T.O. Ting
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp823-825 [Online Full Text]
NextBUS: A Bus Transportation Advisory System for Metropolis
9/8/2019 Proceedings of IMECS 2014, 12-14 March, 2014, Hong Kong, IAENG Open Access Publication
www.iaeng.org/publication/IMECS2014/ 21/29
The 2014 IAENG International Conference on Engineering Physics
A Comparative Spectroscopic Study of Graphene-coated vs Pristine Li(Mn,Ni,Co) Oxide Materials for Lithium-ion BatteryCathodes
Taehoon Kim, Bohang Song, Giannantonio Cibin, Andy Dent, Lu Li, and Alexander M. Korsunsky
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp874-880 [Online Full Text]
Relationship between Eigenvalues and Size of Time Step in Computer Simulation of Thermomechanics Phenomena
Elzbieta Gawronska, and Norbert Sczygiol
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp881-885 [Online Full Text]
Effect of Point Defects on Shear Properties of Graphene Using Molecular Dynamics Simulations
Akihiko Ito, and Shingo Okamoto
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp886-891 [Online Full Text]
The Calculation of Axisymmetric Duct Geometries for Incompressible Rotational Flow Using a Differential EquationApproach and a Boundary Integral Formula based on Green's Theorem
Vasos Pavlika
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp892-899 [Online Full Text]
Distribution of Node-to-Node Distance in a Cubic Lattice of Binding Centers
Zbigniew Domanski, and Norbert Sczygiol
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp900-903 [Online Full Text]
Characterizing the Effect of Processing Parameters on the Porosity of Laser Deposited Titanium Alloy Powder
Rasheedat M. Mahamood, Esther T. Akinlabi, Mukul Shukla, and Sisa Pityana
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp904-908 [Online Full Text]
The Electromagnetic Analysis and Design of a New Permanent Magnetic Eddy Current Damper
XIE Hu, MA Shuyuan, and LI Wenbin
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp909-913 [Online Full Text]
Radiation Effects on Heat and Mass Transfer over an Exponentially Accelerated Infinite Vertical Plate with ChemicalReaction
A. Ahmed, M. N.Sarki, and M. Ahmad
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp914-918 [Online Full Text]
9/8/2019 Proceedings of IMECS 2014, 12-14 March, 2014, Hong Kong, IAENG Open Access Publication
www.iaeng.org/publication/IMECS2014/ 22/29
Comparison of Magnetostatic Field Calculations Associated with Thick Solenoids in the Presence of Iron using a PowerSeries Approach and the Euler-Maclaurin Summation Formula
Vasos Pavlika
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp919-926 [Online Full Text]
The Electromechanical Behavior of a Micro-Beam Driven by Traveling Electrostatic Force
Yuh-Chung Hu, Wei-Hsiang Tu, Pei-Zen Chang, Chih-Kung Lee, David T.W. Lin, and Chung-Neng Huang
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp927-932 [Online Full Text]
Using CUDA Architecture for the Computer Simulation of the Casting Solidification Process
Grzegorz Michalski, and Norbert Sczygiol
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp933-937 [Online Full Text]
The 2014 IAENG International Conference on Industrial Engineering
Lot Sizing Optimisation for Stochastic Make-to-order Manufacturing
X. J. Wang, and S. H. Choi
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp938-943 [Online Full Text]
Improvement Printing Industry Plant layout for Effective Production
A. Watanapa, W. Wiyaratn, and P. Kajondecha
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp944-948 [Online Full Text]
The Capacity Planning Problem Considering the Procurement of Bottleneck Machines and Auxiliary Tools
Yin-Yann Chen, Hsiao-Yao Fan, Chiung-Wen Shih, and Po-Han Huang
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp949-952 [Online Full Text]
Study on Basket Document Factory Plant Layout for Proficient Production
Wisitsree Wiyaratn, and Anucha Watanapa
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp953-956 [Online Full Text]
A Methodology to Evaluate Fiscal Incentives for Promoting Investment of Pioneer Industries: A Case Study
Nur Atikah, Muhammad Hisjam, Wahyudi Sutopo, and Lila H. Bakhtiar
Proceedings of The International MultiConference of Engineers and Computer Scientists 2014 , pp957-962 [Online Full Text]
The power produced by a wind turbine generator (WTG)at a particular site is highly dependent on the wind regimeat that location. There is a number of ways that windspeed can and has been modeled in power system reliabilityevaluation. This method uses the ARMA model to predictwind speeds in the reliability evaluation process and isdesignated as the ARMA approach. An ARMA model with pautoregressive terms and q moving average terms is denotedas ARMA(p, q). The ARMA model created for the SwiftCurrent site in Saskatchewan, Canada based on 1996 to 2003data is shown in the following [15]:
The simulated wind speed at hour t, designated as V (t), canbe calculated as follows:
V (t) = µ(t) + σ(t)s(t). (2)
where µ(t) is the mean observed wind speed at hour andσ(t) is the standard deviation of the observed wind speed athour.
B. Dynamic Modeling of WECS
The power captured by a wind turbine is given by
Pm = 0.5ρπCp(λ, β)R2V 3 (3)
where ρ is the air density (typically 1.25 kg/m3), R is radiusof blades ( in meter), Cp(λ, β) is the wind-turbine powercoefficient, and V is the wind speed (in m/s). The coefficientCp(λ, β) depends on the pitch angle of the blades β (indegrees) and the tip-speed ratio λ, which is defined as theratio of the linear velocity of the blade tip (ωtR) to the windspeed V as follows:
λ =ωtR
V(4)
where ωt is the wind turbine shaft speed (in rad/s).The relation of Cp versus λ of a three-blade horizontal-
axis wind turbine for various blade pitch angles β is illus-trated in Fig. 1. The curves have been obtained by using thefollowing equation that is commonly used in wind turbinesimulators [5], [16]:
Cp(λ, β) = 0.5176(116
λi− 0.4β − 5)e−21/λi + 0.0068λ
(5)1
λi=
1
λ+ 0.008β− 0.035
β3 + 1. (6)
WECS can be structured into several interconnected sub-system models as shown in Fig. 2. This system consists ofwind turbine, a drive train, and a generation unit.
The objective of the proposed control is to maximize thepower that the turbine extracts. This can be achieved if Cpis maximized. To maximize Cp, λ must be kept constantat its optimum value , regardless of the wind speed. Fig. 3illustrates the steady-state power-speed characteristics (solid
0 5 10 150
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Tip−speed Ratio λ
Po
we
r C
oe
ffic
ien
t C
p
β=−0.9
β=−0.8
β=0
β=2
β=6
β=12
β=20
Fig. 1. Power coefficient versus tip-speed ratio, for various blade pitchangles β.
Fig. 2. Structural diagram of WECS systems.
curves) and the maximum power point curve (dashed curve)attained at each wind speed, for a blade pitch angle of 0o.The aerodynamic torque on the wind turbine rotor can beobtained using the following relationships:
Tm =Pm
ωt=
ρπCp(λ, β)R3V 2
2λ. (7)
0 1 2 3 4 5 60
0.5
1
1.5
V=6m/s
V=8m/s
V=9m/s
V=10m/s
V=11m/s
V=12m/s
Tur
bin
Out
put P
ower
Pm
[M
Wat
t]
Wind turbine speed [rad/s]
Fig. 3. Power-speed characteristics of wind turbine, for various wind speedsat pitch angle 0o.
The basic idea of the proposed MPPT technique is toretrieve the optimal rotor speed ωt (meaning the speedcorresponding to the maximum generable power) for anyinstantaneous value of the wind speed. In Fig. 2, the inputsignals coming from the turbine control system are thegenerator torque set point Tg,ref and the desired pitch angleβref . The measured outputs are assumed to be the turbinerotor speed ωt. The wind speed V is the disturbance signal
Proceedings of the International MultiConference of Engineers and Computer Scientists 2014 Vol I, IMECS 2014, March 12 - 14, 2014, Hong Kong
The generator torque Tg is a nonlinear function of ωg andthe control variable Tg,ref . The generator usually operatesin the linear region of its torque characteristic which can,therefore, be approximated by a linear form
Tg = Bgωg − Tg,ref . (11)
The pitch actuator is modelled as a first-order dynamicsystem with saturation in the amplitude and derivative ofthe pitch angle β as [4], [5].
β =−1
τβ +
1
τβref . (12)
It can be seen that the overall WECS model describedin (3)-(12) is nonlinear. Fig. 4 shows the block diagramof a control scheme to track the optimal rotor speed tomaximize the power that the turbine extracts. The control
Fig. 4. Block diagram of nonlinear dynamic of WECS.
system acts on the generator in order to apply the referenceelectromagnetic torque Tg,ref and on the pitch actuator inorder to control the pitch angle of the blades β. The systemparameters are given as follows [17]:Turbine and drive train parametersR=30.30m,Ks=15.66x105N/m,Bs=30.29x102N.ms/rad,Jt=83.00x104kg.m2
To control a given system, the controller design includestwo steps: the first step for identification and prediction ofWECS by quasi-ARX neural network model; and the secondstep for deriving and implementing control law. In Fig. 5, weshows the adaptive controller scheme based on quasi-ARXmodel. To regulate turbine speed at MPPT operating pointis performed by using blade pitch control, with generatortorque assumed to be constant.
Fig. 5. Block diagram of the MPPT controller of WECS.
A. System Identification
Through using Taylor expansion series [8], [10], nonlinearcontinuous function can be presented as
T areTaylor coefficients (nonlinear parameter estimation) and theinformation or input regression vector, respectively. ϕ(t) ∈Rn=nu+ny , n is the dimension of information vector, equalsto the sum of nu and ny that represent orders of time delayin input and output data. ℵ(ϕ(t)) ∈ Rn=nu+ny is a functioncalled as the core-part sub-model to parameterize the inputregression vector. e(t) and y0 are gaussian white noise addedto the system and initial condition of output, respectively.Assumption 1. The pairs of the input and output of trainingdata are bounded.Assumption 2. The input and output of nonlinear functionℵ(ϕ(t)) are bounded.
By performing Taylor expansion series, nonlinear systemis decomposed into linear correlation between the informa-tion vector and its coefficients. It is the same in form likeARX model with nonlinear coefficients. If the system islinear, then the coefficients are constant; and if the system isnonlinear, then the coefficients are not constant or nonlinear.By putting nonlinear function into its coefficients, quasi-linear ARX model is defined as follows,
The system identification are performed by quasi-ARX neu-ral network model is shown in Fig. 6. The embedded of MLPnetwork of quasi-ARX model has input dimension of ϕ(t) isequal to n, the number of hidden layer is m and the numberof output layer is n. The quasi-ARX incorporating neuralnetwork can be expressed as,
where Ω = W1,W2, B, θ,W1 ∈ Rmxn,W2 ∈ Rnxm, B ∈Rmx1 are the weights matrix the first and the second layer.θ ∈ Rnx1 is the bias vector of output nodes, and Γ isthe diagonal nonlinear operator with identical sigmoidalelements on hidden nodes.
Proceedings of the International MultiConference of Engineers and Computer Scientists 2014 Vol I, IMECS 2014, March 12 - 14, 2014, Hong Kong
Fig. 6. Quasi-ARX neural network with MLP network as embeddedsystems.
If model in (15) satisfies to mapping the input-output of thesystem, and Assumption 1. and Assumption 2. are fulfilled,then we can estimate the output of the system at time (t+d).The equation (15) is regressed at time (t + d) to calculatethe output at d step ahead prediction, described as,
T , for onlinestep ahead prediction d is equal to one.
The learning algorithm for quasi-ARX model is per-formed by the back propagation error algorithm for em-bedded MLP network and LSE algorithm for the toupdate θ. Let we introduce two sub-models zl(k) =y(t, ϕ(t))−ϕ(t)[W2(k)ΓW1(k)(ϕ(t)+B(k))]T , and zn(k) =y(t, ϕ(t)) − ϕ(t)θ(k)T , and k is the learning number. Thestep of learning algorithm of quasi-ARX neural network isdescribed by,
1) set k = 0 for initial conditions, θ(k) = 0; and smallinitial values to W1(k), W2(k), and B(k), then setk = 1, where k is the learning number.
2) calculate zl(k), then estimate θ(k) for by using a least-squares error algorithm.
3) calculate zn(k), then estimate W1(k), W2(k), andB(k). It is realized by using the well-known back-propagation (BP) algorithm.
4) use the (16) to update ℵ(k, ϕ(t))5) stop if pre-specified conditions are met and update
ℵ(ϕ(t)) by using ℵ(k, ϕ(t)), otherwise go to Step 2,and repeat the estimation of θ(k), and W1(k), W2(k),and B(k), set k = k + 1.
B. Controller Design
The quasi-ARX prediction model is improved to guaranteesystem stability expressed by
where W2ΓW1(ϕ(t) +B) is nonlinear part, θ is linear part.Obviously, through introducing the switching function χ(t),the improved quasi-ARX neural network model is differentfrom the conventional quasi-ARX model. When χ(t) = 1, it is
a nonlinear prediction model which can insure the predictionaccuracy. And when χ(t) = 0, it is a linear prediction modelwhich can insure the control stability [18].
The linear part error and nonlinear part error, respectivelyis defined as follows :
The switching criterion function are described as follows:
Ji(t) =t∑
l=k
ai(l)(∥ei(l)∥2 − 42)
2(1 + ai(l)ϕ(l − k)Tϕ(l − k)
+ ct∑
l=t−N+1
(1− ai(l)∥ei(l)∥2), i = 1, 2 (22)
χ(t) =
1, if J1(t) > J2(t)0, otherwise (23)
The value of is determined by designer where ≤ϕ(t)ℵ(ϕ(t)). The detail of switching technique and its sta-bility analysis refer to [18].
A minimum variance controller is used for WECS, defineas follows,
M(t+ 1) =
(1
2(y(t+ d)− y∗(t+ d))2 +
λ
2u(t)2
)(24)
where λ is a weight of control input, the controller can beobtained by solving,
∂M(t+ 1)
∂u= 0 (25)
In the case where a conventional neural network is used asa prediction model, a controller can not be derived directlyfrom an identified model because of the nonlinearities. How-ever, the quasi-ARX neural network model is linear in theinput variable u(t). Therefore, a controller is derived fromthe proposed model [8], [18]:
u(t) =b1(t)
b21(t) + λ((b1(t)− b(q−1, ϕ(t))q)u(t− 1)
+ y∗(t+ 1)− a(q−1, ϕ(t))y(t)) (26)
IV. SIMULATION AND RESULTS
To further demonstrate the effectiveness of the proposedMPPT control strategy, the control action is to arrange bladepitch ratio β to track angular velocities of turbine operating inMPPT point. The pitch angle command signal is determinedby the wind speed and pitch angle. Wind speed is generatedby ARMA model with the mean observed wind speed ofµ(t) = 12 m/s and the standard deviation of the observedwind speed of σ(t) = 1.5. The results of simulation in detailare shown in Fig. 7 - Fig. 13. In order to obtain maximumoutput power from a wind turbine generator system, it isnecessary to drive the wind turbine at an optimal rotor speedfor a particular wind speed.
The kernel of MIMO multi layer parceptron neural net-works has one hidden layer, nu=3, ny=4, and m=nu+ny=7.The parameter of switching criterion c=1.2 and N=3. Fig. 7and Fig. 8 illustrate the WECS response in the MPPToperating point. Before t = 0s,V = 12.48m/s, MPPTpower tracking 1.45MW , β = 0deg, angular velocity
Proceedings of the International MultiConference of Engineers and Computer Scientists 2014 Vol I, IMECS 2014, March 12 - 14, 2014, Hong Kong
Fig. 10. Trajectory of ωt of minimum variance controller with switchingbased quasi-ARX model.
0 5 10 15 20 25 30−1
−0.5
0
0.5
1
1.5
Ro
tor
spe
ed
tra
ckin
g e
rro
r (r
ad
/s)
Time (sec)
Fig. 11. Tracking error of turbine angular velocity.
0 5 10 15 20 25 30−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Sw
itch
ing
Time (sec)
Fig. 12. Switching sequance.
0 5 10 15 20 25 300.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
RMS
Time (sec)
Fig. 13. RMS error versus time.
ωt = 4.12rad/s. When the wind speed change to decreaseor increase the MPPT power also should be change in orderto keep maximum operating point of WECS with arrangeturbine rotor speed ωt by controlling blade pitch ratio β.Fig. 9 and Fig. 10 shows the control signal and wind turbinerotor speed tracking. The dot dash line denotes the output ofsystem using proposed method and solid line denotes rotorspeed reference ωt in MPPT operating point, respectively.
The tracking error of turbine rotor speed is shown inFig. 11. Switching function between nonlinear and linear partto keep system stability and control accuracy is shown inFig. 12. The performance of the proposed controller is alsomeasured by the rooted mean squared (RMS) error indexversus time shown in Fig. 13 defined as,
RMS =
√∑Nt=1(y
∗(t)− y(t))2
t(27)
where y∗(t) is the reference signal and y(t) is the controlledsystem output.
V. CONCLUSION
In this paper, quasi-ARX neural network model is used toidentification and prediction nonlinear system. The controllerdesign is derived from the proposed model with switchingfunction to keep system stability. Switching law a made bylogical signal 0 for linear part and 1 for nonlinear part,as we know quasi-ARX neural network model is dividedinto two part; nonlinear and linear. The quasi-ARX modelalso has good properties, it is used to modeling a systeminto linear correlation between regression vector and itscoefficients, so it is easy to derive the controller law by usinglocal linear properties in nonlinear system such as minimumvariance controller. By using minimum variance controllerwith switching law, the proposed model successfully is usedto track maximum power point tracking (MPPT) of WECS.
Proceedings of the International MultiConference of Engineers and Computer Scientists 2014 Vol I, IMECS 2014, March 12 - 14, 2014, Hong Kong
This research has been supported by Indonesian Gov-ernment Scholarship with Directorate General of HigherEducation, Ministry of National Education, (Beasiswa LuarNegeri DIKTI Kementrian Pendidikan dan Kebudayaan Re-publik Indonesia) and Shipbuilding Institute of PolytechnicSurabaya (Politeknik Perkapalan Negeri Surabaya).
REFERENCES
[1] S. Rahman and A. de Castro, “Environmental impacts of electricitygeneration: A global perspective,” IEEE Trans. Energy Convers., vol.10(2), pp. 307–314, 1995.
[2] B. Bose, “Global warming: Energy, environmental pollution and theimpact of power electronics,” IEEE Ind. Electron. Mag., vol. 4(1), pp.6–17, 2010.
[3] S. Sarkar and V. Ajjarapu, “MW resource assessment model for ahybrid energy conversion system with wind and solar resources,” IEEETrans. on Sustainable Energy, vol. 2(4), pp. 383–391, 2011.
[4] E. Muhando, T. Senjyu, A. Yona, H. Kinjo, and T. Funabashi,“Disturbance rejection by dual pitch control and self-tuning regulatorfor wind turbine generator parametric uncertainty compensation,” IETControl Theory Appl., vol. 1, pp. 1431–1440, 2007.
[5] M. Soliman, O. Malik, and D. Westwick, “Multiple model multiple-input multiple-output predictive control for variable speed variablepitch wind energy conversion systems,” IET Renew. Power Gener.,vol. 5(2), pp. 124–136, 2011.
[6] K. S. Narendra and K. Parthasarathy, “Identification and control ofdynamical systems using neural networks,” IEEE Trans. on NeuralNetworks, vol. 1(1), pp. 4–27, 1990.
[7] F. Chen and H. Khalil, “Adaptive control of a class of nonlinear dis-crete time systems using neural networks,” IEEE Trans. on AutomaticControl, vol. 40(5), pp. 791–801, 1995.
[8] J. Hu, K. Kumamaru, and K. Hirasawa, “A Quasi-ARMAX approachto modelling of non-linear systems,” Int. J. Control, vol. 74(18), pp.1754–1766, 2001.
[9] J. Hu, X. Lu, and K. Hirasawa, “Training quasi-ARX neural networkmodel by homotopy approach,” in Proc. SICE Annual Conference inSapporo, Hokkaido Institute of Tecnology, Japan, 2004, pp. 367–372.
[10] J. Hu and K. Hirasawa, “A method for applying multilayer perceptronsto control of nonlinear systems,” in Proc. 9th International Conferenceon Neural Informassion Processing (Singapure), 2002.
[11] M. A. Jami’in, I. Sutrisno, and J. Hu, “Lyapunov learning algorithmfor quasi-ARX neural network to identification of nonlinear dynamicalsystem,” in Proc. IEEE International Conference on Systems, Man, andCybernetics (Seoul), 2012, pp. 3141–3146.
[12] M. A. Jami‘in, I. Sutrisno, and J. Hu, “Deep searching for parameterestimation of the linear time invariant (LTI) system by using quasi-ARX neural network,” in Proc. IEEE International Joint Conferenceon Neural Network (Dallas), 2013.
[13] Y. Cheng, L. Wang, and J. Hu, “Quasi-ARX wavelet network forSVR based nonlinear system identification,” Nonlinear Theory andits Applications (NOLTA), IEICE, vol. 2(2), pp. 165–179, 2011.
[14] A. Mesemanolis, C. Mademlis, and I. Kioskeridis, “High-efficiencycontrol for a wind energy conversion system with induction generator,”IEEE Trans. on Energy Conv., vol. 27(4), pp. 958–967, 2012.
[15] R. Billinton, R. Karki, YiGao, D. Huang, P. Hu, and W. Wangdee, “Ad-equacy assessment considerations in wind integrated power systems,”IEEE Trans. Power Syst., vol. 27(4), pp. 2297–2305, 2012.
[16] M. Pucci and M. Cirrincione, “Neural mppt control of generators withinduction machines without speed sensors,” IEEE Trans. Ind. Electron.,vol. 58(1), pp. 37–47, 2011.
[17] E. Kamal, A. Aitouche, R. Ghorbani, and M. Bayart, “Robust fuzzyfault-tolerant control of wind energy conversion systems subject tosensor faults,” IEEE Trans. on Sust. Energy, vol. 3(2), pp. 231–241,2012.
[18] L. Wang, Y. Cheng, and J. Hu, “A quasi-ARX neural network withswitching mechanism to adaptive control of nonlinear systems,” SICEJournal of Control, Measurement, and System Integration, vol. 3(4),pp. 246–252, 2010.
Proceedings of the International MultiConference of Engineers and Computer Scientists 2014 Vol I, IMECS 2014, March 12 - 14, 2014, Hong Kong