TANASELAN RAMALU et l: DUAL-FUZZY MPPT IN PHOTOVOLTAIC- ANALYSIS FOR DUAL-LOAD Dul-Fuzzy MPPT in Photovoltic- Anlysis for Dul-lod Opertion with SEPIC Converter Tnseln Rmlu,* Mohd Amrn Mohd Rdzi Muhmmd Ammirrul Atiqi Mohd Zinuri 3 Noor Izzri Abd. Whb 4 Ribhn Zfir Abdul Rhmn 5,,3,4,5 Deprtment of Electricl nd Electronic Engineering Fculty of Engineering Universiti Putr Mlysi 43400 UPM Serdng, Selngor, Mlysi theseln537@yhoo.com, mrnmr@upm.edu.my, 3 mmirrultiqi@gmil.com, 4 izzri@upm.edu.my, 5 ribhn@upm.edu.my Abstrct In this pper, dul-fuzzy bsed mximum power point trcking (MPPT) is proposed for stndlone photovoltic (PV) system, working for dul-lod opertion. The proposed MPPT consists of two different fuzzy logic controllers (FLCs) to support opertion of single ended primry-inductor converter (SEPIC) in buck nd boost conditions, by depending on irrdince conditions. To relize dul-opertion of SEPIC, dul-lod pproch hs been pplied by selecting suitble lod resistnce, which could chieve mximum power point (MPP) with specific irrdince nd duty cycle. Then, reltionship between irrdince, lod resistnce nd duty cycle hs been discussed in detils. The dul-fuzzy bsed MPPT together with PV module of Kyocer KD0GH-PU connected to SEPIC ws simulted in MATLAB-Simulink, nd further the lbortory prototype with TMS30F8335 ezdsp bord ws implemented. Ech lod ws connected to direct current () supply to ensure continuity of the supply to both lods. Both simultion nd experimentl results nd comprison nlysis (with P&O) hve been presented. From the results nd nlysis, dul-fuzzy bsed MPPT with dul lods shows tht by switching from PV to, there is no chnge in output voltge, but during the switching from to PV, there re significnt chnges tht cn be noticed ccording to the irrdinces. Keywords fuzzy logic; dul lods; SEPIC dc-dc converter; MPPT; photovoltic. I. INTRODUCTION Among the long list of renewble energy, photovoltic (PV) from solr is much preferble due to esiness to implement with less mintennce. In recent dys, PV systems hve been never ending demnd due to its enormous potentil to be the nerest solution we hve right now to slowly substitute our dying fuel energy. When the PV pnel is exposed to solr irrdition, it could generte direct current electricity without ny environmentl impct nd contmintion. The only drwbcks regrding it, is the cost to mnufcture PV pnel is too high with smll rnge of efficiency which is bout 5-0% []. PV pnel hs nonliner output chrcteristics nd the min fctors ffecting PV output power re solr irrdition, temperture nd lod impednce [, 3]. When solr irrdition rises, PV current increses; however, the temperture of PV module hs more significnt effect on PV voltge opertion [4]. Due to nonliner output chrcteristic of PV pnel, n lgorithm is needed to trck the mximum power point (MPP) of the PV curve to deliver mximum power nd it is clled mximum power point trcking (MPPT). controller PV Pnel Sepic converter Duty cycle chnging controller Supply I V controller DUAL FUZZY MPPT Supply DSP Output from DSP to controllers Negtive connection Posittive connection 3 Fig. Block digrm of PV system with dul-mppt nd dul-lod opertion. DOI 0.503/IJSSST..7.4.46 46. ISSN: 473-804x online, 473-803 print
TANASELAN RAMALU et l: DUAL-FUZZY MPPT IN PHOTOVOLTAIC- ANALYSIS FOR DUAL-LOAD From Sepic converter + _ chnging controller controller controller Supply Supply Fig.b Opertion of the controllers when SEPIC is connected to lod From Sepic converter Current flowing No connection + _ chnging controller controller controller Supply Supply Fig.c Opertion of the controllers when SEPIC is connected to lod. The opertion of MPPT is bsiclly to find the operting voltge V MPP nd the operting current I MPP t which PV opertes to obtin MPP. There hve been mny MPPT methods developed nd implemented [5, 6]. Among them re perturb nd obverse (P&O), incrementl conductnce (IC), rtificil neurl network (ANN), fuzzy logic controller (FLC), constnt voltge, three point weight comprison, short current pulse, nd open circuit voltge. Specificlly for FLC, it does not require n ccurte mthemticl model nd is known to be very efficient in hndling problems tht hve non linerity vribles s in PV [6,7]. As dc-dc converters such s buck-boost nd CUK converters hve inverted polrity in their output vlue, single ended primry-inductor converter (SEPIC) is much more preferble becuse polrity of its output is not inverted [8, 9]. However, till dte, previous reported works only covered either boost or buck conditions, or both buck nd boost in their converter opertions with only single lod. Therefore, in this pper, two FLCs with ech hving different duty cycle will be used. If the duty cycle used is lower thn 0.5, SEPIC will operte in buck condition; however, when the duty cycle is higher thn 0.5, SEPIC will operte in boost condition [9- ]. To relize dulopertion of SEPIC, nother controller is built to hndle switching of the two lods depending on the PV output current nd it is clled s lod switching controller. Both resistors hve been set specificlly ccording to certin irrdince level nd duty cycle in order to chieve MPP. Therefore, through SEPIC, one of the resistors will undergo buck opertion nd nother resistor hs been set to undergo boost opertion with the help of lod switching controller nd both opertions will occur t different time. Menwhile to mke sure tht both lods hve continuous power supply, both lods hve been ttched to supply which will power the lod whenever there is no PV source connected in either lod. In this pper further, section II covers the proposed system for dul-lod opertion nd followed by dul Fuzzy logic controllers nd hrdwre implementtion in Sections III nd IV respectively which concludes the findings. II. PROPOSED PV SYSTEM The proposed stndlone PV system consists of four min components: PV pnel, SEPIC, MPPT nd controllers, s shown in Fig..The controllers consist of lod chnging controller nd intermedite controller. The following section will describe the modeling of ech component. Fig. shows the block digrm of dul-fuzzy MPPT PV system with dul-lod opertion. Menwhile, voltge nd current re mesured nd used by dul-fuzzy MPPT to produce duty cycle to operte IGBT in SEPIC [7]. A lod chnging controller hs been set up t the lod side. The function of lod chnging controller is to control the connection between SEPIC nd lods. Threshold current is set in the controller. Once the input current is more thn threshold vlue, SEPIC converter will be connected to lod by lod chnging controller. If the current is less thn threshold vlue, it remins s defult, where the lod chnging controller connects lod to SEPIC converter. The function of intermedite controllers is to control the connection of lods to either supply or SEPIC. When the irrdince is low, the current delivered from the PV pnel will be lower nd FLC, s further explined lter with higher duty cycle will be selected. With higher duty cycle, SEPIC performs boost ction nd thus higher resistor (lod ) will be connected to be ssocited with the low current supply nd to obtin high output voltge t the lod side. When the sun irrdince is higher, the input current from PV will increse. Now the trgeted lod will be lower resistor (lod ), so high current could flow through it nd of course the lod voltge will be lesser, so SEPIC will perform buck ction resulting from the lower duty cycle. The connections of resistor lods with intermedite controllers nd lod chnging controller re clerly shown in Fig. The defult settings of the controllers hve been mde bsed on irrdince 00W/m s lod chnging controller nd intermedite controller re ttched to lod, nd however, only intermedite controller is ttched to lod for connecting it to supply, s shown in Fig b. Menwhile, Fig c shows the lod chnging controller nd intermedite lod connects SEPIC to lod, while DOI 0.503/IJSSST..7.4.46 46. ISSN: 473-804x online, 473-803 print
TANASELAN RAMALU et l: DUAL-FUZZY MPPT IN PHOTOVOLTAIC- ANALYSIS FOR DUAL-LOAD intermedite lod connects lod with supply. All the connections of the controllers hve been tbulted ccordingly in tble. TABLE : OPERATION OF LOAD CHANGING AND INTERMEDIATE CONTROLLERS. III. DUAL FUZZY LOGIC CONTROLLERS (FLCS) Two FLCs re configured s MPPT in this system to perform buck nd boost opertions for dul lods. One FLC hs been set to give lower duty cycle nd nother FLC hs been set to higher duty cycle. The inputs for both FLCs re sme nd the difference is the output duty cycle. Both FLCs use two inputs; error E nd chnge of error CE t smple time k, which re defined by (5) nd (6) [, 3]. E is the chnge of output power from PV over the chnge of output voltge from PV. CE is the difference between current E from the previous E t the given smple time. E (k) = (5) CE (k) = E (k) E (k-) (6) Bsiclly, opertion of FLC cn be clssified into four bsic elements: fuzzifiction, rule bse, inference engine nd defuzzifiction [-4]. During fuzzifiction, inputs of the fuzzy, CE nd E vribles re clculted nd converted into linguistic vribles bsed on the membership function. The output D is generted by looking up in rule bse tble. The fuzzy output is converted bck to numericl vrible from linguistic vrible during defuzzifiction. According to the Fig.4, both selections of duty cycle nd switching of the lod controller re bsed on input current. When the input current is higher thn the threshold vlue tht hs been set in the lgorithm itself, FLC with lower duty cycle will be selected. At the sme time, lod switching controller will be switched to lod which hve low resistnce. The concept of designing the controller is bsed on P- V curve. In ech FLC, 7 membership functions re configured for ll inputs. All the membership functions re set s tringulr shpes with both ending sides of the universe of disclose compnied by trpezium shpe to show the continuous of the controller. Seven-term fuzzy sets, Negtive Big (NB), Negtive Medium (NM), Negtive Smll (NS), Zero (Z), Positive Smll (PS), Positive Medium (PM), Positive Big (PB), re defined to describe ech linguistic vrible. The fuzzy method is Mmdni where the mximum or minimum composition technique for the inference is used. Center-of-grvity methods re used in the defuzzifiction process. Fig. 3 shows membership ptterns for the following inputs of E nd CE nd output voltge reference. Tble shows the rules implemented for both controllers. From these seven membership functions, zero (ZE) membership is representing MPP, which divides left nd right sides ccording to the curve grdient. The positive membership functions (PB, PM nd PS) re focused t left side of MPP bsed on positive grdient with PB is the most left region. However, the negtive membership functions (NB, NM nd NS) re focused t right side of MPP bsed on negtive grdient with NB is the most right region. The res of PS nd NS re rther smller s compred to the rest of the membership functions in order to mke the controller becomes more sensitive towrds ZE which determines the duty cycle of the controller. The usge of FLC involves initil duty cycle where the initil duty cycle pplied is difference for both low irrdince nd high irrdince. Without the initil duty cycle, the result of FLC is highly unpredictble nd it could ffect the power converter by triggering high duty cycle into the converter resulting in high output voltge. The duty cycle cn be incresed nd decresed but with certin rnge. So for both buck nd boost opertion, two FLC re being implemented. Fig.3 Fuzzy Memberships of inputs nd output for first controller, nd second controller. DOI 0.503/IJSSST..7.4.46 46.3 ISSN: 473-804x online, 473-803 print
TANASELAN RAMALU et l: DUAL-FUZZY MPPT IN PHOTOVOLTAIC- ANALYSIS FOR DUAL-LOAD TABLE. RULES FOR BOTH FUZZY LOGIC CONTROLLERS VI. HARDWARE IMPLEMENTATION Lbortory prototype ws developed to investigte the performnce of the proposed dul-fuzzy bsed MPPT with dul-lod opertion. Solr simultor is used insted of solr pnel due to the experimentl needs of chnging the irrdinces frequently. The solr simultor used is Chrom 600H-600S; progrmmble power supply 600V/5A/5KW with Solr Arry Simultion. The prmeters in the solr simultor hve been set ccording to the PV module Kyocer KD0GH PU. Dul-fuzzy bsed MPPT is implemented through microcontroller TMS30F8335 ezdsp bord. The MPPT lgorithm ws designed by using MATLAB/Simulink nd by using Code Composer Studio (CCS); the control code ws generted before it ws downloded into DSP. Fig.4 Flow chrt of proposed dul-fuzzy MPPT. DOI 0.503/IJSSST..7.4.46 46.4 ISSN: 473-804x online, 473-803 print
TANASELAN RAMALU et l: DUAL-FUZZY MPPT IN PHOTOVOLTAIC- ANALYSIS FOR DUAL-LOAD Fig.5 shows the input voltge from PV. For the chnge from irrdinces 700 W/m to 00 W/m, the input voltge remin the sme s shown in Fig.5, but there is significnt voltge drop during the switching from low irrdince (00 W/m ) to high irrdince (700 W/m ). By using dul-fuzzy, voltge drop of bout 4V nd settling time of 0.3s re obtined, which re lower thn P&O MPPT with voltge drop of 5V nd higher settling time in bout 0.35s with very high ripples, s illustrted clerly in Fig.5b. Fig.6 shows the PV input powers for the chnge from 00 W/m to 700 W/m nd the chnge from 700 W/m to 00 W/m. Lrger voltge drop nd high ripples occur when P&O MPPT is used, resulting in poor qulity in input power. Due to high voltge drop during step down t time of 0s s shown in Fig. 6c, the input power of PV when using P&O is lso ffected. At period of 0s, P&O cuses the power drop nerly 0W nd then high ripples nd high oscilltions tke plce throughout the experimentl opertion. When dul-fuzzy is in use, power drop is only bout 3W with lesser ripples. Fig.7 shows the output voltges t lod nd lod. Whenever the intermedite controllers switch the lods to different supplies (PV supply nd supply), output voltges of the lods hve been observed. All the results re presented together by compring P&O nd dul-fuzzy. Fig 7 shows output voltge t lod during switched from PV supply to supply. b c Fig. 6 Input powers when irrdince chnges from )00 W/m to 700 W/m, b)700 W/m to 00 W/m, nd c)700 W/m to 00 W/m which re focused t time of 0s during the fll of irrdince. b b Fig. 5 Input voltges when irrdince chnges from )700 W/m to 00 W/m, nd b) 00 W/m to 700 W/m However, Fig 7b shows output voltge t lod during switched from supply to PV supply. Output voltge t lod which ws switched from PV supply to supply nd supply to PV supply ws presented in Fig. 7c nd Fig. 7d respectively. Tble 3 shows the comprison of dulfuzzy nd P&O in terms of percentge of ripples, overshoot, nd rising time. c DOI 0.503/IJSSST..7.4.46 46.5 ISSN: 473-804x online, 473-803 print
TANASELAN RAMALU et l: DUAL-FUZZY MPPT IN PHOTOVOLTAIC- ANALYSIS FOR DUAL-LOAD d Fig. 7 Effect of dul-fuzzy nd P&O lgorithms in output voltges during switched from ) PV supply to supply t lod, b) supply to PV supply t lod, c)pv supply to supply t lod, nd d) supply to PV supply t lod. TABLE 3. OUTPUT VOLTAGE CHARACTERISTICS OF BOTH DUAL-FLC AND P&O ALGORITHMS. Dul-fuzzy P&O Voltge drop t input voltge Power drop t input power Ripple in lod nd lod Voltge Over shoot in lod Rising time in lod (output voltge) 0.06%(.6V drop) 48%(3V drop) 0.36% 37.5% (6W power drop) (5W power drop) 0.08%(0.4V) 4.%(.5) 3%(0.5V) 0.05seconds VII. CONCLUSION 43% (7.5V) 0. seconds This pper presents dul-fuzzy bsed MPPT for SEPIC with dul lods by ensuring mximum power trnsfer could hppen either in low or high irrdinces. Higher duty cycle is more suitble to be pplied during lower irrdince nd must be connected to higher resistnce lod. However, lower duty cycle is preferred to be switched when the irrdince is high nd lower resistnce lod must be connected to it in order to obtin MPP. Dul-fuzzy bsed MPPT is needed in order to perform boost nd buck opertions seprtely in two different lods. Anlyses of the output voltges obtined t lod nd lod shows tht during the switching from supply to PV supply, if the irrdince is high, there will be high overshoot voltge due to high current of PV nd if the irrdince is low, there will be voltge drop due to low current supply by PV. [3] Krismdint, N.A.Rhim, H.W.Ping, J.Selvrj Photovoltic module modeling using simulink/mtlb Procedi Environmentl Sciences 7 ( 03 ) 537 546 [4] C.Qi, Z.Ming Photovoltic Module Simulink Model for Stndlone PV System 0 Interntionl Conference on Applied Physics nd Industril Engineering, 0 Interntionl conference on ppplied phyiscs nd Industril Engineering [5] E. Durán, M. Sidrch-de-Crdon, J. Glán, nd J.M. Andújr Comprtive Anlysis of Buck-Boost Converters used to obtin I- V Chrcteristic Curves of Photovoltic Modules Power Electronics Specilists Conference, 008. PESC 5-9 June 008. [6] S. Llouni, D. Rekiou, T. Rekiou, E. Mtgneb Fuzzy logic control of stnd-lone photovoltic system with bttery storge Journl of Power Sources 93 (009) 899 907 [7] M.H. Tghvee, M.A.M.Rdzi, S.M.Moosvin, HshimHiz.m, M.Hmiruce Mrhbn A current nd future study on non-isolted converters for photovolticpplictions Renewble nd SustinbleEnergyReviews7(03)6 7 [8] V. Sls, E. Olı s, A. Brrdo, A. L zro Review of the mximum power point trcking lgorithms for stnd-lone photovoltic systems Solr Energy Mterils & Solr Cells 90 (006) 555 578. [9] Zinl Slm, Juber Ahmed, Benny S. Merugu The ppliction of soft computing methods for MPPT of PV system: A technologicl nd sttus review Applied Energy 07 (03) 35 48. [0] A.A. Nbulsi, R.Dhoudi Efficiency Optimiztion of DSP- Bsed Stndlone PV System Using Fuzzy Logic nd Dul-MPPT Control IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 8, NO. 3, AUGUST 0 [] R.Ckmk, I.H. Alts, A.M. Shrf Modeling of FLC-Incrementl Bsed MPPT using - Boost Converter for Stndlone PV System Innovtions in Intelligent Systems nd Applictions (INISTA), 0 Interntionl Symposium -4 July 0 [] I.H. Alts, A.M. Shrfb A novel mximum power fuzzy logic controller for photovoltic solr energy systems Renewble Energy 33 (008) 388 399 [3] R. Rhmni, M.M. Seyedmhmoudin, S. Mekhilef nd R. Yusof, Implementtion of fuzzy logic mximum power point trcking controller for photovoltic system Americn Journl of Applied Sciences, 0 (3): 09-8, 03. REFERENCES [] B.Prid, S. Iniy, nd R. Goic, A review of solr photovoltic technologies Renewble nd Sustinble Energy Reviews 5 (0) 65 636. [] K. Ding, X. G. Bin, H. H. Liu, nd T. Peng, A mtlb-simulinkbsed pv module model nd its ppliction under conditions of nonuniform irrdince, IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 7, NO. 4, DECEMBER 0. DOI 0.503/IJSSST..7.4.46 46.6 ISSN: 473-804x online, 473-803 print