mppt

2310 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 23, NO. 5, SEPTEMBER 2008

A Seamless Mode Transfer MaximumPower Point Tracking Controller ForThermoelectric Generator Applications

Rae-Young Kim, Student Member, IEEE, and Jih-Sheng Lai, Fellow, IEEE

Abstract—A boost-cascaded-with-buck converter-based powerconditioning system employing a seamless mode transfer max-imum power point tracking controller is proposed to maximizeenergy production of a thermoelectric generator while balancing avehicle battery, alternator output power, and vehicle load. When avehicle battery is fully charged, the proposed controller switches toa power matching mode seamlessly by a dual loop control system,which detects the input and output voltages and currents of theboost-cascaded-with-buck converter, and adjusts the commandsaccordingly. Both voltage and current loops are designed in afrequency domain using small signal models to ensure stableoperation. A mode selection and voltage and current commandsare determined by a digital signal processor-based controller. Theexperimental results with a dynamic source and load steps arepresented to show the effectiveness of the proposed approach.

Index Terms—Maximum power point tracking (MPPT) con-troller, seamless mode transfer, thermoelectric generator.

I. INTRODUCTION

A RECENTLY developed thermoelectric generator (TEG)has shown promise in vehicle applications [1], [2]. By in-

tegrating the TEG with an exhaust pipe and a heat exchangersystem, it is possible to recover the engine waste heat, and con-vert it to electric power for battery charging and subsequentlyimprove vehicle gas efficiency. In a recent study, a 10% fuelefficiency improvement was predicted by using the TEG to of-fload an alternator [2]. In order to interface between the TEGand a vehicle electrical bus, a power converter is essential. Thepower converter can operate in two different modes: 1) max-imum power point tracking (MPPT) and 2) power matching(PM). When a vehicle battery is not fully charged, the MPPTmode is activated to harness as much power as possible fromwaste heat. During this mode, if TEG output is higher than ve-hicle power demand, the TEG supplies power for both a vehicleload and a battery. If TEG output is less than vehicle power de-mand, the TEG powers the vehicle load with the assistance ofthe battery. On the other hands, when the battery is already fullycharged, the PM mode is activated to reduce power production

Manuscript received October 10, 2007; revised March 04, 2008. Current ver-sion published November 21, 2008. Recommended by Associate Editor J. Guer-rero. This work was supported in part by the Visteon Corporation and by BSST,LLC. This paper is a revision of a paper presented at the 2007 IEEE Industry Ap-plications Conference, 42nd IAS Annual Meeting, New Orleans, LA, September23–27, 2007.

The authors are with the Future Energy Electronics Center, Virginia Poly-technic Institute and State University, Blacksburg, VA 24061-0111 USA(e-mail: [email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TPEL.2008.2001904

and to avoid overcharging the battery. During the PM mode, thevehicle electrical bus is regulated at a float charging voltage bythe power converter.

For the MPPT methods, there have been significant re-searches in photovoltaic applications. Since they do notnecessarily operate in two different modes, most literaturesimply has shown the control with generic MPPT methods suchas the hill climbing method [3], the perturbation and observa-tion methods [4], [5], the incremental conductance methods[6], and the ripple correlation control methods [7], [8]. Themicroprocessor-adapted methods such as a fuzzy logic, a neuralnetwork and a state-space-based model have been proposed forperformance improvement [9]–[11].

Although most existing MPPT methods can be applied to theproposed TEG systems, it is necessary to modify the use ofthe MPPT method to operate in two different modes. In [12],a system with PV output charging to a battery was implementedwith a single-stage dc-dc converter and a two-loop control tech-nique, where a voltage loop regulates the PV voltage to tracka maximum power point, and a current loop regulates the bat-tery charging current. These loops were alternately selected ac-cording to battery charging status. In [13], a two-stage dc-dcconverter was adopted to use an electric double layer capacitorfor energy storage and to implement MPPT to regulate batterycharging voltage. There were two operating modes for the firststage converter and three operating modes for the second stageconverter. These modes were changed with the voltage statusesof the double-layer capacitor and the battery. The major problemof the control techniques found in [12] and [13] is the tran-sient over-voltage or -current associated with the dynamic modechanges. Literature [14] solved this problem by controlling solararray output power. In [15] and [16], a voltage loop control tech-nique to operate in an MPPT mode for battery charging was pro-posed. The problem with these techniques was they did not con-sider the case when the PV generated more power than needed.

In this paper, a boost-cascaded-with-buck converter is pro-posed as a power converter to operate in a wide TEG voltagerange, i.e., from 0 to 25 V. The common dc bus between twostages can adopt a double layer capacitor as the energy storageoption. The control loops allow the MPPT or the PM modeoperation to maximize the TEG output or to match the TEGoutput with battery state-of-charge and vehicle load conditions.In order to avoid transients associated with a mode transfer be-tween the MPPT and the PM modes, a double dual-loop con-troller is proposed for both boost and buck converters with a cur-rent loop as the inner loop and a voltage loop as the outer loop.

0885-8993/$25.00 © 2008 IEEE

KIM AND LAI: SEAMLESS MODE TRANSFER MAXIMUM POWER POINT TRACKING CONTROLLER 2311

Fig. 1. Thermoelectric generator power conversion system employing theseamless mode transfer MPPT controller.

The mode transfer is controlled by a voltage loop that regulatesthe battery voltage. When the battery voltage is fully charged,the power converter needs to switch to PM mode. The proposedmode transfer approach is to incorporate the voltage loop con-trol with a perturbation and observation (P&O) method. Un-like the mode transfer relying on a form of switch that requiresstop-and-go, and thus producing significant transients, the pro-posed approach allows the mode transfer seamlessly.

In designing both voltage and current loop controllers, theboost-cascaded-with-buck converter was modeled in averagemode, and the frequency domain analysis and design approachwas adopted for a proportional-integral (PI) controller to ensurecontrol loop stability. Converter stability and mode transfersmoothness were verified with a hardware prototype. The powerconverter consists of two back-to-back three-phase convertersthat achieve overall power conversion efficiency of 96% at thenominal power condition. The controller is implemented witha TMS320F2812 digital signal processor board that includesthe essential sensor conditioning and interface circuits for thepower stage.

II. THERMOELECTRIC GENERATOR MPPT CONTROLLER

A. System Description

Fig. 1 shows the proposed TEG power conversion system em-ploying the seamless mode transfer MPPT controller for vehicleapplications. The TEG produces output voltage ranging from 0to 25 V depending on temperature difference, and the powerconverter needs to regulate the output voltage between 12.3 and16.5 V to provide battery charging. The power converter workswith the alternator denoted as ALT to regulate the vehicle powerbus according to availability of the TEG output and vehicle loadconditions. When the TEG output is insufficient to supply allthe vehicle loads, the alternator needs to kick in and provideextra power to regulate the electrical bus. When the TEG outputis high enough to power all the vehicle loads, the alternatorcan be shut off. With extensive circuit topology evaluations, aboost-cascaded-with-buck converter is selected to meet the de-sired efficiency of 95% or higher. To ensure power handlingcapability and to reduce ripple and its associated losses, theproposed boost-cascaded-with-buck converter is to have threephases interleaving each other for current sharing and ripple re-duction. In addition, synchronous rectification is employed to

Fig. 2. Power-versus-current characteristics of the thermoelectric generatorunder varying TEG voltage condition.

Fig. 3. Block diagram of the seamless mode transfer MPPT controller.

reduce conduction loss. The dc-link capacitor is used as themiddle stage energy buffer, which can be a double-layer capac-itor to allow more flexibility of energy harnessing.

B. Basic Operation of the Proposed MPPT Controller

The maximum power point (MPP) of the TEG is a functionof exhaust temperature, and can be characterized with the TEGvoltage, which changes dynamically with a temperature differ-ence between its P-type and N-type materials, and source resis-tance [2], [17]. As shown in Fig. 1, the TEG electrical modelis represented by using a voltage source denoted as behind asource resistance denoted as . Given a temperature condition,the is determined, and then output power versus current plot,which is called P-I characteristics, can be obtained as shownin Fig. 2. The curve with higher or higher power representshigher temperature condition. Note that there exists only oneMPP, which is denoted as a dotted line, of the given value.

Fig. 3 shows the block diagram of the seamless mode transferMPPT controller. The boost converter controller denoted asdotted lines is responsible for the MPPT or the PM modesdepending on battery state-of-charge. The modified P&Ocompensator continuously loops back the input voltage

, the input current and the error of output voltage ,and then adjusts the input current command to capture asmuch power as possible from the TEG in the MPPT mode orto regulate the output voltage as commanded in the PM mode.The detailed explanation of the is presented in Section

2312 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 23, NO. 5, SEPTEMBER 2008

II-C. After comparing the with the , the error is fed intothe current compensator to generate the control voltage

. The is converted to the duty cycle for the boostconverter by the PWM circuit.

The buck converter controller denoted as solid lines is respon-sible for regulating the dc-link voltage . The current compen-sator regulates the output current to match the outputcurrent command , and the voltage compensator reg-ulates the according to the given voltage command . Ahigher allows more energy storage that can provide morepower to offload the alternator for better gas efficiency.

C. Algorithm of the Modified P&O Compensator

P&O methods have been widely used in photovoltaic systemsfor MPPT. The basic idea of the P&O methods is to regulate cur-rent or voltage to pursue a positive slope of P-I characteristicsbefore reaching the MPP and a negative slope of a P-I character-istics after passing the MPP. After a few iterations of the P&Oprocess, a system operating point moves toward the MPP.

For the modified P&O compensator , the basic algorithmcan be described as follows: 1) when the output voltage islower than its command , the MPPT mode is activated to cap-ture as much power as possible from the TEG, and movestoward a direction which increases TEG power; 2) when theis higher than the , the PM mode is triggered to reduce powerproduction, and the changes toward a direction which de-creases TEG power; and 3) when the equals the , thesettles down and remains unchanged. Based on the previous al-gorithm, the can be implemented as (1) in the digital control

(1)

where is the error between the and at the th sam-pling time, is a constant gain, repre-sents a sign of the instantaneous slope of the input power atthe th sampling time, and the denotes a limiting functionto set the maximum perturbation for avoiding excessive oscilla-tions. The and the are the current commandsat the th and th sampling time, respectively. The in(1) exhibits both a generic P&O method and a voltage compen-sator, and thus, the smooth mode transfer is achieved withoutany transients.

In the PM mode, there exists two possible operating pointsunder given load condition on P-I characteristics of Fig. 2, butthe operating point with smaller current is desirable for the highpower conversion efficiency. In the , the functionin (1) is forced to set a positive regardless of the calculationresult of the power slope, and thus, the PM mode operates in thedesirable point which exists in the right side of the MPP.

III. MODELING OF TEG POWER CONVERSION SYSTEM

For the design of compensators with desirable dynamics andstability, frequency domain based design techniques have beenwidely used. The proposed boost-cascaded-with-buck converterhas a common bus with a dc-link capacitor, and thus an interac-tion between two converters exists. In order to predict dynamicsand stability with the consideration of the interaction, the accu-rate model including the TEG and battery is derived.

Fig. 4. Equivalent circuit of the boost-cascaded-with-buck converter with theTEG and battery model.

Fig. 5. Subset circuits of the boost-cascaded-with buck converter: (a) boostconverter sunk with a current source and (b) buck converter sourced with a cur-rent source.

Fig. 4 shows the electrical model of the TEG system em-ploying the boost-cascaded-with-buck converter. From themodeling point of view, the complete three-phase interleavedcircuit can be represented with a single-phase one with one thirdof the original inductance. A linear electrical battery model,which consists of a voltage source and an internal resistor

, is used to represent the short-term output characteristic of abattery [18], [19] and a vehicle load is modeled as a resistor .

Considering Fig. 4 and the relationship between the inductorcurrents , and the upper switch currents , , the aver-aged voltage of the dc-link capacitor over one switching periodcan be expressed in (2)

(2)

where the math accent “ ” denotes the averaged value over oneswitching period. Equation (2) reveals that the boost-cascaded-with buck converter can be decomposed into two subset circuitsof the basic converter sunk or sourced with a current source,as shown in Fig. 5. For the each subset circuit, the convertermodeling is straightforward and easily derived by using the non-linear averaged model of the PWM switch [20]. Such converter

KIM AND LAI: SEAMLESS MODE TRANSFER MAXIMUM POWER POINT TRACKING CONTROLLER 2313

Fig. 6. Linearized small-signal circuit of the boost-cascaded with buck converter with the TEG and battery model.

modeling significantly simplifies the model effort while pro-viding a straightforward solution that approximates the conven-tional accurate modeling. Aggregating the two subset models,the nonlinear averaged model of the TEG system employing theboost-cascaded-with-buck converter is obtained as expressed in(3) and (4)

(3)

(4)

The small-signal model can be derived as follows. The statevariables , , and , input variables and , outputvariables , , and , and duty ratios , are decomposedinto the nominal values and small-signal deviations. When thedecomposed variables and the duty ratios are substituted into(3) and (4), the linearized small-signal model is derived withassumption that the nonlinear or second-order terms are ne-glected under sufficiently small deviations. Fig. 6 illustrates theresulted linearized equivalent circuit of the TEG system em-ploying the boost-cascaded-with-buck converter, where upper-case letters denote the nominal values, and the math accent “ ”denotes the small-signal deviations from the nominal values.

The major control-to-output transfer functions are obtainedfrom Fig. 6, as shown in (5)–(8), where and are dutycycle-to-boost inductor current and duty cycle-to-output voltage

transfer functions of the boost converter, and and areduty cycle-to-buck inductor current and duty cycle-to-dc linkvoltage transfer functions of the buck converter. In the equa-tions, the terms represent gain, the terms represent thezeros, and the terms represent the poles. The transfer func-tions reveal a third-order dynamics with one real pole and onepair of complex conjugate poles. Note that in (6) showsright-half-plane zero and in (8) has a negative dc gain

(5)

(6)

(7)

(8)

IV. DESIGN OF THE PROPOSED MPPT CONTROLLER

A. Buck Converter Controller

Fig. 7 represents the block diagram of the buck convertercontroller in a closed loop, where and denote feed-back current and voltage sensor of the buck converter and isPWM gain. The two-zero and three-pole PI compensator of (9)is used for the current loop compensator , and the one-zeroand two-pole PI compensator of (10) is used for the voltage loopcompensator to ensure stability under both light-load andheavy-load conditions of and

(9)

2314 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 23, NO. 5, SEPTEMBER 2008

Fig. 7. Block diagram of the buck converter controller in a closed loop.

(10)

Using the circuit parameters: 15 V, 100 m ,20 V, 14.4 V, 5 H, 3.3 H,

500 F, 10 m , 12 V, and 10 m , the cur-rent compensator parameters and voltage compensator param-eters are determined. The is designed at 4.7 kHz,

5.5 kHz, 20 kHz, 23 kHz, andas shown in (11). The is designed at 350 Hz,

2.5 kHz, and as shown in (12). Notethat the needs a negative feedback gain due to the negativegain of the

(11)

(12)

Fig. 8(a) shows gain/phase plots of the current loop for anopen and closed loop under light- and heavy-load conditions.The design result shows a phase margin higher than 100 anda gain margin higher than 20 dB for both light- and heavy-loadconditions. The crossover frequency is about 800 Hz. For thevoltage loop, the design result shows a phase margin higher than80 and a gain margin higher than 40 dB for both light- andheavy-load conditions, as shown in Fig. 8(b). The crossover fre-quency ranges from 60 to 300 Hz.

B. Boost Converter Controller

Fig. 9 shows the closed-loop controller block diagram of theboost converter controller, where and denote feed-back current and voltage sensor of the boost converter. Thetwo-zero and three-pole PI compensator of (9) is used for thecurrent loop compensator . The designed current loop com-pensator parameters under the same circuit parameters men-tioned previously with both light- and heavy-load conditions of

2.5 and 0.5 are: 2.0 kHz,2.7 kHz, 19 kHz, and

(13)

The frequency domain representation of the modified P&Ocompensator is derived from (1). With given circuit pa-rameters, the of (1) sets to a positive one

Fig. 8. Gain/phase plots of the buck converter controller for an open and closedloop: (a) the current controller loop gain and (b) the voltage controller loop gain.

Fig. 9. Block diagram of the boost converter controller in a closed loop.

because the converter operates in the PM mode. When the ef-fect of the limiting function is neglected, the -domain expres-sion of (1) is substituted by , and multiplied with

in order to consider the zero-order-hold effect[21]. The result of this calculation is given in (14), where is

KIM AND LAI: SEAMLESS MODE TRANSFER MAXIMUM POWER POINT TRACKING CONTROLLER 2315

Fig. 10. Gain/phase plot of the boost converter controller for an open andclosed loop: (a) the current controller loop gain and (b) the voltage controllerloop gain.

the sampling frequency. Note that the exhibits the inherentpole at the origin, and thus zero steady-state error is achievedwithout an additional compensator

(14)

The constant gain should be selected to provide a properphase and gain margin for stability, and its crossover frequencyshould be much less than that of the inner loop so that an inter-action between the inner loop and the output loop is avoided. Inthis paper, the is designed to 0.02 under the sampling fre-quency of 10 kHz. The switching frequency of the converteris set to 50 kHz.

Fig. 10(a) shows gain/phase plots of the current loop for anopen and closed loop under light- and heavy-load conditions.The design result shows a phase margin higher than 130 anda gain margin higher than 17 dB for both light- and heavy-load conditions. The crossover frequency ranges from 600 Hz

Fig. 11. Test setup for the thermoelectric generator power conversion system.

to 3 kHz. For the voltage loop, the design result shows a phasemargin higher than 60 and a gain margin higher than 20 dB forboth light- and heavy-load conditions, as shown in Fig. 10(b).The designed crossover frequency ranges from 70 to 300 Hz.Note that the phase of the voltage loop in Fig. 10(b) dramat-ically drops beyond the half sampling frequency , due tothe digital sampling effect.

V. EXPERIMENTAL VERIFICATION

In order to verify the effectiveness and stability of the pro-posed seamless mode transfer MPPT controller, a completeTEG prototype was built and tested. Fig. 11 shows the test setupof the power converter and associated instruments. It consistsof four major parts: a constant voltage controlled power supplywith a series connected resistor to mimic a thermoelectricgenerator, a three-phase interleaved boost-cascaded-with-buckconverter, a DSP to implement the modified P&O compensator,and a battery bank with load resistors. A precision current shuntis used to calibrate the current measurement. The three-phaseinterleaved boost-cascaded-with-buck converter consists oftwo stacked power boards, where a current compensator isimplemented by analog circuits. The power supply voltage isset to 14.9 V. For the internal resistance of the TEG, differentcombinations are obtained with a heat-suck resistor bank thatcontains six resistors. For the load resistor bank, different resis-tance values are obtained with paralleled and series connectionsof 12 resistors that are either 1 or 0.5 . A 12-V battery isconnected in parallel to the load resistor bank. The outputvoltage reference of the power converter was set at 14.4 V.

A. Mode Transfer Test

To verify the smooth and continuous mode transfer, a loadresistance is changed under a fixed power supply voltageof 14.9 V and a fixed internal resistance of 0.21 . Fig. 12(a)shows test results when the operating point is changed from theMPPT mode to the PM mode. The load is reduced by increasingthe from 0.5 to 1 . Before reduction, the output voltageis less than its command of 14.4 V, and thus the controller isoperating in the MPPT mode. The theoretical maximum power

2316 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 23, NO. 5, SEPTEMBER 2008

Fig. 12. Experimental waveforms for mode transfer test: (a) from a maximumpower point tracking mode to a power matching mode (1 s/div) and (b) from apower matching mode to a maximum power point tracking mode (1 s/div).

MPP was calculated to be 264.3 W at 35.5 A. When the loadis reduced, the reaches its command. The input currentis reduced gradually, and the operating mode is changed to thePM mode.

Fig. 12(b) indicates that the operating point is returned to theMPPT mode again when the load is increased by changing the

to 0.5 . After the is stepped to 0.5 , the dropsless than its command. The starts increasing, and the inputvoltage starts decreasing due to more voltage drop of theinternal resistor . Eventually, the reaches the previousvalue 35 A and the mode transfer is completed. Note that thedc-link voltage is always maintained at 20 V as commandeddespite a mode transfer and a converter load change.

B. MPPT Tracking to Internal Resistance Variation

To see the tracking behavior of the proposed MPPT controllerunder varying value, the steps from 0.21 to 0.26 , andvice versa. Theoretical maximum power is 264.3 W at 35.5 Ain the case of 0.21 , and 191.4 W at 25.6 A in the case of 0.26

, respectively. Small load resistance of 0.5 is used tokeep the operating point on the MPPT mode regardless of vari-ation of the . Fig. 13(a) shows experimental waveforms when

Fig. 13. Experimental waveforms when internal resistance is changed: (a) froma small resistance to a large resistance (1 s/div) and (b) from a large resistanceto a small resistance (1 s/div).

the varies from 0.21 to 0.26 . After increasing resistance,the MPPT controller decreases the input current to track theMPP smoothly. Finally, the MPPT controller reaches its theo-retical MPP.

Fig. 13(b) indicates that when the returns to 0.21 , thestarts to increase, and reaches 33 A. It is observed that the inputvoltage increases as soon as the changes due to decreasedvoltage drop. The output voltage changes depending on theinput power . The dc-link voltage is well regulated at 20 Vas commanded.

C. DC-Link Voltage Regulation

As a bonus advantage of the proposed MPPT controller, thedc-link voltage can be regulated at any given command. Itallows more flexibility of energy harnessing when the dc-linkcapacitor is replaced with an electric double-layer capacitor.Fig. 14 shows experimental waveforms when the dc-link voltagecommand steps from 20 to 30 V, and vice versa. The fol-lows the . The output voltage and the input power arenot affected by this variation. From the tested waveform of the

, it is seen that the proposed double dual-loop controllers arestable under different dc link voltages. Note that it is observedthat the input current is perturbed within 3 A peak-to-peak

KIM AND LAI: SEAMLESS MODE TRANSFER MAXIMUM POWER POINT TRACKING CONTROLLER 2317

Fig. 14. Experimental waveforms for a dc-link voltage regulation (1 s/div).

ac current of 50 Hz to track the MPP. The peak-to-peak mag-nitude was set by the limiting function of the modified P&Ocompensator.

VI. CONCLUSION

In this paper, a three-phase boost-cascaded-with-buck con-verter is adopted for a high-efficiency TEG power conditioningsystem to operate in a wide input voltage range and to maximumTEG output. A modified perturbation and observation MPPTmethod along with double dual-loop controllers is proposed toallow seamless mode transfer when the TEG output exceeds thebattery state of charge. The proposed method allows smooth andcontinuous mode transfer without a sudden transient. Duringmode transfer and load steps, the dc-link voltage is regulated ata constant value as commanded. It allows more flexibility of en-ergy harnessing when a dc-link capacitor is replaced with ultracapacitor.

The entire TEG power conversion system configuration andoperating principle of the controller were described in detail.The controller was designed by using a small-signal model ofthe power converter, and then the hardware experiments werecarried out to verify the design. From experimental results,the stability of the proposed MPPT controller was proven,and MPPT and output voltage regulation were verified under:1) load steps; 2) source impedance variation; and 3) differentdc link voltage commands.

REFERENCES

[1] J. Yang, “Potential applications of thermoelectric waste heat recoveryin the automotive industry,” in Proc. IEEE Int. Conf. Thermoelectron.,2005, pp. 19–23.

[2] J. LaGrandeur, D. Crane, S. Hung, B. Mazar, and A. Eder, “Automo-tive waste heat conversion to electric power using Skutterudite, TAGS,PbTe and BiTe,” in Proc. IEEE Int. Conf. Thermoelectron., 2006, pp.282–285.

[3] Y. Chen, Y. Liu, and F. Wu, “Multiinput converter with power factorcorrection, maximum power point tracking, and ripple-free input cur-rents,” IEEE Trans. Power Electron., vol. 19, no. 3, pp. 631–639, May2004.

[4] N. Femia, G. Petrone, G. Spagnuolo, and M. Vitelli, “Optimization ofperturb and observe maximum power point tracking method,” IEEETrans. Power Electron., vol. 20, no. 4, pp. 963–973, Jul. 2005.

[5] N. Kasa, T. Iida, and L. Chen, “Flyback inverter controlled by sensor-less current MPPT for photovoltaic power system,” IEEE Trans. Ind.Electron., vol. 52, no. 4, pp. 1145–1152, Jul. 2005.

[6] Y.-C. Kuo, T.-J. Liang, and J.-F. Chen, “Novel maximum-power-point-tracking controller for photovoltaic energy conversion system,” IEEETrans. Ind. Electron., vol. 48, no. 3, pp. 594–601, May 2001.

[7] P. Midya, P. Krein, R. Turnbull, R. Reppa, and J. Kimball, “Dynamicmaximum power point tracker for photovoltaic applications,” in Proc.IEEE Power Electron. Specialists Conf., 1996, pp. 1710–1716.

[8] Y. Lim and D. Hamill, “Simple maximum power point tracker for pho-tovoltaic arrays,” Electron. Lett., vol. 36, no. 11, pp. 997–999, 2000.

[9] M. Veerachary, T. Senjyu, and K. Uezato, “Neural-network-basedmaximum-power-point tracking of coupled-inductor interleaved-boost-converter-supplied PV system using fuzzy controller,” IEEETrans. Ind. Electron., vol. 50, no. 4, pp. 749–758, Jul. 2003.

[10] T. Kottas, Y. Boutalis, and A. Karlis, “New maximum power pointtracker for PV arrays using fuzzy controller in close cooperation withfuzzy cognitive networks,” IEEE Trans. Energy Conv., vol. 21, no. 3,pp. 793–803, 2006.

[11] E. Solodovnik, L. Shengyi, and R. Dougal, “Power controller designfor maximum power tracking in solar installations,” IEEE Trans. PowerElectron., vol. 19, no. 5, pp. 1295–1304, Jul. 2004.

[12] P. Huynh and B. Cho, “Design and analysis of microprocessor con-trolled peak power tracking system,” IEEE Trans. Aerosp. Electron.Syst., vol. 32, no. 1, pp. 182–190, Jan. 1996.

[13] Y. Nozaki, K. Akiyama, H. Kawaguchi, and K. Kurokawa, “Animproved method for controlling an EDLC-Battery hybrid stand-alonephotovoltaic power system,” in Proc. IEEE Appl. Power Electron.Conf., 2000, pp. 781–786.

[14] Y. Cho and B. Cho, “A digital controlled solar array regulator em-ploying the charge control,” in Proc. IEEE Energy Conv. Eng. Conf.,1997, pp. 2222–2227.

[15] C. Hua, J. Lin, and C. Shen, “Implementation of a DSP-controlled pho-tovoltaic system with peak power tracking,” IEEE Trans. Ind. Electron.,vol. 45, no. 1, pp. 99–107, Jan. 1998.

[16] E. Koutroulis, K. Kalaitzakis, and N. Voulgaris, “Development of a mi-crocontroller-based, photovoltaic maximum power point tracking con-trol system,” IEEE Trans. Power Electron., vol. 16, no. 1, pp. 46–54,Jan. 2001.

[17] H. Nagayoshi, “Comparison of maximum power point control methodsfor thermoelectric power generator,” in Proc. IEEE Int. Conf. Thermo-electron., 2002, pp. 450–453.

[18] Y.-H. Kim and H.-D. Ha, “Design of interface circuits with electricbattery models,” IEEE Trans. Ind. Electron., vol. 44, no. 1, pp. 81–86,Jan. 1997.

[19] C. Zhan, X. Wu, S. Kromlidis, V. Ramachandaramurthy, M. Barnes,N. Jenkins, and A. Ruddell, “Two electrical models of the lead-acidbattery used in a dynamic voltage restorer,” IEE—Proc. TransmissionDistrib., vol. 150, no. 2, pp. 175–182, 2003.

[20] V. Voperian, “Simplified analysis of PWM converters using the modelof the PWM switch, continuous conduction mode,” IEEE Trans.Aerosp. Electron. Syst., vol. 26, no. 3, pp. 490–496, May 1990.

[21] R. Ridley, “A new, continuous-time model for current-mode control,”IEEE Trans. Power Electron., vol. 6, no. 2, pp. 271–280, Mar. 1991.

Rae-Young Kim (S’06) received the B.S. and M.S.degrees in electrical engineering from the HanyangUniversity, Seoul, Korea, in 1997 and 1999, respec-tively. He is currently pursuing the Ph.D. degree inpower electronics from the Virginia Polytechnic In-stitute and State University, Blacksburg.

He is currently a Graduate Research Assistant withthe Virginia Polytechnic Institute and State Univer-sity. From 1999 to 2004, he was a Senior Researcherwith the Hyosung Heavy Industry Research and De-velopment Center, Seoul, Korea. His research inter-

ests include modeling and control of power converters, soft switching tech-niques, and power converter applications in renewable energy systems.

2318 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 23, NO. 5, SEPTEMBER 2008

Jih-Sheng (Jason) Lai (S’85–M’89–SM’93–F’07)received the M.S. and Ph.D. degrees in electricalengineering from the University of Tennessee,Knoxville, in 1985 and 1989, respectively.

From 1980 to 1983, he was the Head of the Elec-trical Engineering Department, Ming-Chi Instituteof Technology, Taipei, Taiwan, R.O.C., where heinitiated a power electronics program and receiveda grant from his college and a fellowship from theNational Science Council to study abroad. In 1986,he became a staff member with the University of

Tennessee, where he taught control systems and energy conversion courses. In1989, he joined the Electric Power Research Institute (EPRI) Power ElectronicsApplications Center (PEAC), where he managed EPRI-sponsored power elec-tronics research projects. From 1993, he worked with the Oak Ridge National

Laboratory as the Power Electronics Lead Scientist, where he initiated a highpower electronics program and developed several novel high power convertersincluding multilevel converters and soft-switching inverters. In 1996, he joinedVirginia Polytechnic Institute and State University, Blacksburg. He is currentlya Professor and the Director of the Future Energy Electronics Center. Hismain research areas include high efficiency power electronics conversionsfor high power and energy applications. He has published more than 195technical papers and 2 books and received 17 U.S. patents. He chaired the2000 IEEE Workshop on Computers in Power Electronics (COMPEL 2000),2001 IEEE/DOE Future Energy Challenge, and 2005 IEEE Applied PowerElectronics Conference and Exposition (APEC 2005).

Dr. Lai was a recipient of several distinctive awards including a Tech-nical Achievement Award in Lockheed Martin Award Night, two IEEE IASConference Paper Awards, Best Paper Awards from IECON-97, IPEC-05, andPCC-07.