Simulation and Experimental validation of Distributed MPPT algorithms for partially shaded Photovoltaic systems

Transcription

1 M. Muthuramalingam 1, P.S.Manoharan P.T.R College of Engineering and Technology, Madurai, India (1), Thiagarajar College of Engineering, Madurai, India () Simulation and Exerimental validation of Distributed MPPT algorithms for artially shaded Photovoltaic systems Abstract. This aer analyzes the erformance of two different maximum ower oint tracking (MPPT) algorithms for hotovoltaic (PV) system: artificial neural network () and adative neuro fuzzy inference () as used by an interleaved soft switching boost converter (ISSBC) system with different conditions, such as artially shaded, condition, changing solar insolation and PV cell temerature. However, under artially shaded conditions, when the PV module characteristics get more comlex with multile eaks of outut ower. Both algorithms are methodically investigated by means of Matlab simulation and hardware exerimental validation, comare in terms of arameters tracking seed, ower extraction, and harmonic analysis. In this toology, each cascaded H-bridge inverter (CHBMLI) unit is connected to an individual PV module through an interleaved soft switching boost converter (ISSBC). The simulation and hardware results show that algorithm is outerforming than the algorithm. Streszczenie. Streszczenie. W artykule analizowane są dwa algorytmy śledzenia maksymalnej mocy (MPPT) stosowane w systemach fotowoltaicznych: jeden () wykorzystuje sieci neuronowe a drugi wykorzystuje. Układ ozwla na załączanie rzekształtnika w zależności od warunków, n. zacienienia czy temeratury. Symulacja i ekserymentalna weryfikacja algorytmów śledzenia mocy w systemach fotowoltaicznych. Keywords: Photovoltaic (PV) system, Maximum ower oint tracking (MPPT), Microcontroller, Interleaved soft switching boost inverter (ISSBC), Cascade H-bridge inverter (CHBMLI)..Słowa kluczowe: systemy fotowoltaiczne, śledzenie mocy MPPT,, scieci neuronowe. doi:1.1915/e Introduction Photovoltaic energy has increased interest in electrical ower alications, since it is considered as a basically limitless and generally on hand energy resource In general, there is a unique oint on the P-V or I-V curve, called the maximum ower oint (MPP), at which the entire PV system oerates with maximum efficiency and roduces its maximum outut ower. The erformance of a hotovoltaic module is highly affected by the artial shaded condition [1]. Many MPPT techniques have been reorted in the literature such as article swarm otimization [] erturb and observation [3], incremental conductance [4], fuzzy logic based controller,[5] genetic algorithm, and artificial neural net work etc [6]-[8]. This aer resents a unique combination of an interleaved soft switched boost converter (ISSBC) run by a set of two hotovoltaic anels (PV) with a distributed MPPT, suitable to guarantee MPPT even under artial shadowed conditions, managed by an adative neuro fuzzy inference () unit. The ISSBC is followed by a, single hase cascaded H bridge five-level inverter (CHBMLI) driven by the individual DC oututs of the ISSBC, with selective harmonic elimination scheme to eliminate tyically the high order harmonics[9]. A comarison of maximum ower oint tracking (MPPT) algorithms for hotovoltaic (PV) system under artial shadow conditions is carried out [1]. The exerimental system is constituted by three main elements are shown in Fig.1: PV anel, dc-dc interleaved soft switching boost converter, and cascade multileveled H-bridge inverter. Fig.1. General Diagram of load connected hotovoltaic system Table 1. PV module arameters Parameter Maximum Power (Pmax) 15W Voltage at Pmax (Vm) 34.5V Current at Pmax (Im) 4.35A Oen-circuit voltage (Voc) 43.5V Short-circuit current (Isc) 4.75A. PV array modeling and simulation A 15 W rated PV anel consisting of 7 multicrystalline silicon solar cells in series arallel-connected combination used for the alication. In this model, a PV cell is reresenting by a current basis in arallel with a diode and a series resistance. The basic current equation is given in Eq. (1). qv (1) I Iv - I ex - 1 akt where I PV = current generated by the incident light, T = Temerature of the PN junction, a = Diode ideality constant I, = leakage current of the diode, q = electron charge C, k = Boltzmann constant ( J/K. To develo embedded simulink model based on current equation and manufacturer s data sheet arameter as shown in Table 1 of BP SX 15S PV module. 3. MPPT control algorithms The MPPT algorithm is using for extracting the maximum ower from the PV module and convey to the load. The ISSB DC-DC converter serves transferring maximum ower from the solar PV module to the load by changing the duty cycle. Energy conversion efficiency of MPPT algorithm calculated as Eq. () () MPPT t P t P vmax t vmppt where P v-mppt reresents the outut ower of PV with MPPT, P v-max is the outut ower true maximum ower oint. dt t dt 9 PRZEGLĄD ELEKTROTECHNICZNY, ISSN 33-97, R. 9 NR 8/14

2 3.1 MPPT algorithm Artificial neural networks () are electronic models based on the neural structure of the brain. This function ermits s to be used in the design of adative and intelligent systems since they are able to solve roblems from revious examles. models involve the creation of massively aralleled networks comosed mostly of nonlinear elements known as neurons. The is used to estimate the otimal duty cycle in real time, which corresonds to maximum ower at any given inut voltage and current. The MATLAB / SIMULINK based neural network fitting tool is used to select data, create and train a network and evaluate its erformance using mean square error and regression analysis. The is using to estimate the otimal voltage (V ref ) in real time, which corresonds to the maximum ower at any given inut insulation (G) and Temerature (T). The oeration of this technique is exlained in the block diagram in Fig.. The develoed configuration is a multilayer-ercetron structure including an inut layer, a hidden layer and outut layer, whose function is activated sigmoid. 3. MPPT algorithm The system is used to formulate the neural network architecture in the inference engine of a fuzzy controller. The functional block diagram and structure of is shown in Fig.3. The structure comrises of three distinct layers namely inut layer, hidden layer and outut layer. The inut signals error (e) and change in error signal (ce) whose membershi functions are selected as Gaussian membershi function; the outut is modulation index (d). Fig. 3 (a) Adative neuro fuzzy control system Fig. 4. Training error and Surface view Fig.. MPPT control system In this work, database engaged consists of 15 atterns of PV insulation, temerature and reference voltage variables, which have divided into two subdatabases, 7%, of the samles are used to train the, and the rest 3% are second-hand to test and validate the network. The erformance is measured by calculating the mean- square error as shown in Eq. (3). 1 ( i) ( i) (3) e y v i1 Where, = number of training data entries; y = outut vector; v = desired outut For a set of inut insolation and temerature, a welltrained would give as an outut the reference voltage that is very close to the desired value, giving an error nearly zero. Each neuron a j comutes a weighted sum of its n inuts V k, k = 1,... n, and generates an outut as shown in Eq. (4). n (4) a j tan sig wkvk bias k 1 The tan sigmoid function of the resultant weighted sum that usually has a bias associated with it that can be considered as an additional inut gives the outut. In (3), w k reresents the synase weight associated with each one of the n inuts. These training sets chosen to cover all the tyical inut sace in order to get good erformance where temerature ranges from 15 C to 55 C and solar irradiation ranged from 5 W/m to 15 W/m. The inut membershi functions are maed to the outut membershi function by 49 rules through grid artitioning method using the FIS generator in Matlab Simulink. The 5 data sets to train are obtaining from worksace from the revious MPPT algorithm, such as insolation, temerature and reference voltage. The learning data trained through back roagation technique for 5 eochs for minimum error tolerance. The network training is erforming reeatedly until the erformance indexes E Vref -V v reduce below a secified value ideally to zero. In other words when E leads to - V, then the trained connecting weights ref V v are adjusted in such a way that the estimated array voltage is identically equal to the MPP voltage. The trained surface rule hase view shown in Fig.4, the trained data set exorts the simulation and observes the erformance different artial shading condition. 4. Soft switching boost inverter The interleaved boost inverter consists of two singlehase boost converters that are connected in arallel and converters oerating 18 degrees out of hase with 3 khz switching frequency. It is ointed out that in interleaved converter mode 6 khz effect is achieved by hase shifting of the two 3 khz switching signals. The inut-inductorrile-current cancellation occurs at 5% duty cycle. Hence, the design value of the duty ratio is to.5 in this system. Therefore, the interleaved converter have lower switching losses, therefore the outut voltage of the solar cell can be boosted with high efficiency [11]. 5. Single hase CHBML inverter In this aer, selective harmonic elimination ulse with modulation technique is imlemented to generate the switching duty cycle for CHBMLI [1], [13]. Equation (5) shows the contents of the outut voltage at infinite frequencies, the module voltage V v1 -V v are associated to their resective switching angle α 1 - α. (5) 4 V. V cos( n. ab n 1,5,7, n V cos( n. v1 1 v ) PRZEGLĄD ELEKTROTECHNICZNY, ISSN 33-97, R. 9 NR 8/14 93

3 These trigonometric transcendental equations can be solved by GA fitting tool. To find the switching angle (offline) for a set of redetermined modulation indices to get the required fundamental outut voltage in a five level cascaded multi level inverter. The switching angles (α 1, α ) lie in between zero and π/. The collected the set of data trained in Simulink tool and exorted to the system. The is train, to outut the set of angles for each inut voltage situation. 6. Simulation Results The simulink software validates the erformance of the MPPT techniques under different oerating conditions. The PV module arameters are obtained from the 15- Watts multicrystalline PV Module technical data sheet. Such arameters are considered in the Standard Test two techniques are able to extract the MPP. Higher ower extracted from algorithm comared to also gives a fast steady state resonse with less oscillation. Current(A) G1=1,G=1(W/m) G1=1,G=8(W/m) 1 G1=1,G=6(W/m) G1=1,G=4(W/m) Voltage Fig. 7. I-V Curve at 5 C Table. Dynamic resonse of shaded irradiation attern Irradiation Irradiation Cell G1 (W/m) G(W/m) Temerature (from t=s to (from t=s to T( C) t=s) t=4s) PD PD Fig.5. Simulation block diagram of the system Condition (STC): 1W/m and cell temerature of 5 C. The simulation block diagram is shown in Fig.5. First, the characteristics of the PV module are validated and connected with converter system then the erformance of the MPPT techniques under various conditions is evaluated. The simulation validation of PV module and converter results of the I-V and P-V characteristics of PV module as a function of irradiation and temerature shown in Fig It can be observed quite similar to the PV module as er data sheet. In order to achieve the maximum ower oint of PV modules, and MPPT controller has been develoed using Matlab Simulink model. The simulation result is resented for the following configurations. Power(W) G1=1,G=1(W/m) Gi=1,G=8(W/m) G1=1,G=6(W/m) G1=1,G=4(W/m) Voltage Fig. 6. P-V Curve at 5 C 6.1. Dynamic Changing the Solar Radiation Simulations are carried out converter alone with two techniques under dynamically changing solar irradiations at temerature of 5 C.The simulation attern and corresonding result arameter detailed in Table and Table 3 resectively. Fig. 8 shows simulation outut voltage, ower and efficiency of sudden changes in solar irradiation from 5 to 1 W/m of PV module 1 (PD1) and PV module (PD) at 4 W/m. In this analysis, the Fig. 8. Dynamic resonse change in irradiation 6. Effect of Partially Shaded Solar Irradiation In order to verify the erformances of the and algorithm, the ISSBC-CHBMLI configuration and connected to an RL load (R=1 ohm and L=mH). The ISSB Converter is controlled by MPPT algorithms and the CHBML inverter controlled by SHE PWM technique (detailed exlained in chater 5). The simulation carried out insolation configuration of artial shaded condition is PV1 equal to 8 W/m constant (non-shaded), the PV initially 5% artificially closed condition (shaded) after 3s interval of time fully oened. The simulated detailed outut result ISSBC and CHBMLI tabulated in Table.4. The steed outut voltage and corresonding harmonic sectrum of 7.5 khz both non-shaded and shaded condition of and algorithm shown in Fig.9 (a-b) to Fig.1 (a-b). Observe the simulation result more DC ower extraction, higher AC outut voltage with less harmonics in controlled algorithm comare to algorithm both nonshaded and artially shaded condition. 94 PRZEGLĄD ELEKTROTECHNICZNY, ISSN 33-97, R. 9 NR 8/14

4 Table 3. Dynamic resonse of simulation Simulation time configuration From t =s to t=s From t =s to t = 4s MPPT Resonse Efficiency Resonse Vdc Pdc MI Vdc Pdc MI Time (ms) (%) Time(ms) Efficiency (%) PD PD PD PD Table 4. Simulation result of artially shaded and non-shaded connected with inverter Insolation ISSBC CHI (G1/G) MPPT W/m PV1 PV Steed THD at T= Voltage Vrms (%) 5 C 8/ / / / Exerimental validation The exerimental set u of the roosed system is shown as in Fig.13. The both ISSB converter and CHBML inverter, consists of IRF84 MOSFETs four for converter and eight for inverter through roer otical isolation using MCTE oto couler ICs, rogram is then downloaded, into the memory of ARM rocessor LPC 148, using the Phillis flash tool, 16F877A microcontroller, serial ort MA3, digital oscilloscoe, rheostat is used as DC resistive load for dynamic resonse and RL load (R=1 ohm and L=mH) AC load for artially shaded exerimental validation. Fig.11.Simulation results for MPPT under unbalanced condition (a) outut voltage (b) voltage harmonic sectrum Fig.1.Simulation results for MPPT under unbalanced condition (a) outut voltage (b) voltage harmonic sectrum Table 5. Non-shaded and shaded attern for exerimental condition Insulation Insulation Cell G1 (W/m) G (W/m) Temerature ( C) D D Fig.9.Simulation results for MPPT under balanced condition (a) outut voltage (b) voltage harmonic sectrum Table 6. Comarative analysis of exerimental condition Converter Inverter (DSO outut) MPPT PV1 PV Total (W) Efficiency (%) Steed voltage Vrms D D D D Fig.1 Simulation results for MPPT under balanced condition (a) outut voltage (b) voltage harmonic sectrum The validation artially shaded exerimental condition both and algorithm. The exerimental attern as shown in Table 5, which insolation attern D1 both PV PRZEGLĄD ELEKTROTECHNICZNY, ISSN 33-97, R. 9 NR 8/14 95

5 module G1 = 84W/m ( non-shaded), attern D shaded test condition G = 84 W/m changed after 3s from fully illuminated to 45%-5% suddenly closed condition. During exerimental, the PV module in artially shaded resonse detailed result as shown in the Table 6. Obtained exerimental measurements take through DSO (MAKE UNI-T) arameters of the non-shaded and shaded voltage and corresonding harmonic sectrums shown in Fig.14. Fig.13. Exerimental arrangements Hence, in both non-shaded and artially shaded oeration modes, algorithm imroves the voltage quality, ower extraction, harmonics elimination as comared to the algorithm. Fig.14.Exerimental result for (a) shaded outut voltage and harmonic sectrum (b) Non-shaded voltage harmonic sectrum (c) shaded out voltage and harmonic sectrum (d) non-shaded out voltage and harmonic sectrum 8. Conclusion This aer analyzes the erformance of and MPPT algorithms by stand-alone PV system. The configuration for the roosed system is designed and simulated using MATLAB/Simulink and imlemented in 16F877A microcontroller latform. The roosed system shows a good dynamic erformance, algorithm to track the MPP of the PV units even under the raid change in irradiation cell temerature and artial shaded condition. can rovide the overall efficiency higher than algorithms. The CHBMLI integrate with SHE modulation technique imroved outut voltage quality and reduction in THD ercentage even in artially shaded of PV modules with the based MPPT algorithm. REFERENCES [1] Xianwen Gao., Shaowu Li., Rongfen Gong., 13, Maximum ower oint tracking control strategies with variable weather arameters for hotovoltaic generation systems, Solar Energy, 93 (5), [] Kashif Ishaque., Zainal Salam., 13, A Deterministic article swarm otimization maximum ower oint tracker for hotovoltaic system under artial shading condition IEEE Transactions on Industrial Electronics, 6 (8), [3] Patel.H V., Agarwal.V., 8, MATLAB based modeling to study the effects of artial shading on PV array characteristics, IEEE Transaction on. Energy Conversion, 3, 3 [4] Hohm.D.P., Ro.M.E.,, Comarative study of maximum ower oint tracking algorithms using an exerimental, rogrammable, maximum ower oint tracking test bed, Proc. Photovoltaic Secialist Conference, [5] Esram.T., Chaman.P.L., 7, Comarison of Photovoltaic array maximum ower oint tracking techniques, IEEE Trans. Energy Conversion, (), [6] Azadeh Safari., Saad Michele., 11, Simulation and hardware imlementation of incremental conductance MPPT with direct control method using cuk inverter, IEEE Transactions on Industrial Electronics, 58 (4), [7] Mellit.A, Kalogeria SA., 11 -based modeling for hotovoltaic ower suly system. Renewable Energy 36 (1) 5 58 [8] Ravi.A., Manoharan.P.S., Vijay Anand.J., 11, Modeling and simulation of three hase multilevel inverter for grid connected hotovoltaic systems, Solar Energy, 85(11), [9] Faete Filho., Leon.M., Tolbert., Yue Cao., 11, Real-time selective harmonic minimization for multilevel inverters connected to solar anels using artificial neural network angle generation, IEEE Transactions on Industry Alications, 47 (5), [1] Yi-Hua Liu., Chun-Liang Liu., Jia-Wei Huang., Jing-Hsiau Chen., 13, Neural-network-based maximum ower oint tracking methods for hotovoltaic systems oerating under fast changing environments, Solar Energy, 89 (5), [11] Whei-Min Lin., Chih-Ming Hong., Chiung Hsing Chen., 11, Neural network based MPPT control of a stand-alone hybrid ower generation system, IEEE Transactions on Power Electronics, 6 (1), [1] Doo-Yong Jung., Young-Hyok Ji., Sang-Hoon Park., Yong Chae Jung., Chung-Yuen Won., 11, Interleaved Soft- Switching boost inverter for hotovoltaic ower-generation system, IEEE Transactions on Power Electronics, 6 (4), [13] Beser.E., Arifoglu.B., Camur.S., Beser.E.K., 1,A gridconnected hotovoltaic ower conversion system with singlehase multilevel inverter, Solar Energy, M.MUTHURAMALINGAM is the Assistant rofessor at P.T.R. College of Engineering and Technology, Madurai. His research interests include renewable energy resources. muthuramjaya@yahoo.in Dr.P.S.MANOHARAN is the Associate rofessor at Thiagarajar College of Engineering, Madurai. He has ublished more than 6 aers in international journals and conferences. His research interests include Renewable energy resources, ower system management, and Evolutionary algorithms. smeee@tce.edu 96 PRZEGLĄD ELEKTROTECHNICZNY, ISSN 33-97, R. 9 NR 8/14