Global Conference on ower Control and Optimization, Kuching, Malaysia, 4, December 1 COTROL AD OTIMIZATIO OF FUZZ BASED MAXIMUM OWER OIT TRACKIG I SOLAR HOTOOLTAIC SSTEM C. S. Chin,. eelakantan, H.. oong, K. T. K. Teo School of Engineering and Information Technology, University Malaysia Sabah, Kota Kinabalu, Malaysia cschin84@yahoo.com kenteo@ums.edu.my Abstract Solar photovoltaic () electrification is an important renewable energy source. The electric which is converted directly from solar irradiation via panel is not steady due to different solar intensity. To maximize the panel output power, perturb and observe (&O) maximum power point tracking (MT) has been implemented into system. Through a buckboost DCDC converter, MT is able to vary the operating voltage and search for the maximum power that the panel can produce. The implementation of fuzzy logic has been proposed in this paper. Based on the input change of power and input change of power with respect to change of voltage, fuzzy can determine the size of perturbed voltage and facilitate in maximum power tracking faster and minimize the voltage variation after the maximum power point has been identified. Simulation results show that the performance of fuzzy based MT is better than conventional &O MT. Keywords: hotovoltaic, MT, fuzzy logic, perturb & observe 1. Introduction The world today is aware of solar photovoltaic as an important renewable energy source for electricity generation especially in countries where the solar density is relatively high. Solar photovoltaic is a phenomenon where the solar irradiation is converted directly into electricity via solar cell [1] and the process does not have any materials to be consumed or emitted. Solar electrification can be applied even in rural areas where standalone system can supply adequate electricity for certain area independently without the need of having connection with utility grid. The array has a particular operating point that can supply the maximum power to the load which is generally called maximum power point (M). The maximum power point has a nonlinear locus where it varies according to the solar irradiance and the cell temperature []. To boost the efficiency of system, the M has to be tracked followed by regulating the panel to operate at M operating voltage point, thus optimizing the production of electricity. This process can draw as much power as possible that the panel can produce. There are several methods that have been widely implemented to track the M. The most widely used methods are erturb and Observe (&O), incremental conductance and threepoint weight comparison. In this paper, &O MT is investigated. &O technique apply perturbation to the buckboost DCDC controller by increasing or decreasing the pulse width modulator (WM) duty cycle, subsequently observes the effect on the output power [3]. If the power at present state is larger than previous state, the controller s duty cycle shall be increased or viceversa until the M operating voltage point is identified. roblem that arises in &O MT method is that the operating voltage in panel always fluctuating due to the needs of continuous tracking for the next perturbation cycle. In this paper, fuzzy logic is proposed to be implemented in MT. is robust and relatively simple to design since fuzzy do not require information about the exact model [4]. The power at the present state will be compared with the power at the previous state and the change of power will be one of the inputs of fuzzy. Another input is the change of power with respect to change of voltage. Based on these two inputs, fuzzy can determine the size of perturbed voltage. Therefore, fuzzy based MT can track the maximum power point faster. In addition, can minimize the voltage fluctuation after M has been recognized.. System description The fuzzy based MT solar system can be illustrated as shown in Fig. 1. The system consists of a panel, buckboost converter, fuzzy based MT control unit and a load. The power produced by panel is supplied to the load through a buckboost converter. The output voltage and current from the panel are fed to the fuzzy based MT control unit to determine the perturbed voltage reference for buckboost converter. panel I Buckboost DCDC converter based MT control unit Load WM Fig. 1: based MT solar system..1. Modeling of panel The general model of solar cell can be derived from physical characteristic called one diode model. The
Global Conference on ower Control and Optimization, Kuching, Malaysia, 4, December 1 equivalent circuit of solar cell is shown as Fig. [4][5]. The Schockley diode equation which describes the I characteristic of diode is shown in equation (1) 1 (1) where is the diode current, is the reverse bias saturation current, is the voltage across the diode, is ideality factor of the diode and is the thermal voltage. Thermal voltage however can be defined as in equation () () where is Boltzmann constant (1.38653 1 3 J/K), is temperature in degrees kelvin and is electron charge (1.617646 1 19 C). To model the I characteristic of array, equation (3) has been derived based on the equivalent circuit in Fig., 1 (3) where is the current at terminals of array, is the array current, is the array terminal voltage, is the equivalent series resistance of the array and is the equivalent parallel resistance. Unlike the electrical generators which are generally classified as either current source or voltage source, the device presents hybrid behavior. panel acts as a current source when panel operates at voltage smaller than M voltage point but it acts as voltage source when it operates at voltage larger than M voltage point [6]. The series resistance has strong influence when panel acts as voltage source whereas the parallel resistance has great influence when the panel acts as current source. is the sum of structural resistance of panel however exists due to leakage current of pn junction depending on the fabrication method of the cells. Generally, is very high and is very low. High resistance blocks the panel from shortcircuited and low resistance of allows the current flow to the load without resistance. Hence, these parameters can be neglected [6]. The characteristics of SHAR E8EEA multicrystalline silicon module with 8W have been studied. The SHAR E8EEA is modeled in MATLABSIMULIK using equation (3) with the assumption that the module has constant temperature of 5C. Since is very small and is very high, it can be assumed that I pv is equal to panel short circuit current (I sc ). The parameters obtained from SHAR E8EEA datasheet for panel modeling are shown in Table 1. Fig. 3 shows the characteristic of the panel at different solar irradiance. It can be noticed that the M operating voltage point of panel varies at different solar irradiance. As the solar irradiance increased, the M voltage point is higher. Fig. 4 shows the characteristic of panel at 6W/m solar irradiance and the corresponding I curve. The panel modeled in MATLABSIMULIK has similar characteristics that described in SHAR E8EEA datasheet... Buckboost DCDC converter Buckboost DCDC converter is an important element in system since buckboost converter is able to regulate the output voltage that may be less or greater than the input voltage. Buckboost converter allow more flexibility in modulating the energy transfer from the input sorce to the load by varying the duty cycle [7]. Fig. 5 shows the circuit diagram of buckboost DCDC converter. The operation of the buckboost converter can be Table 1: arameters of SHAR E8EEA array at 5C and 1W/m solar irradiance. arameters Symbol Typical value Open circuit voltage oc 1.3 Maximum power pm 17.1 voltage Short circuit current I sc 5.16A Maximum power current I pm 4.68A Maximum power m 8W o. of cells 36 9 8 7 6 5 4 M 6W/m owervoltage characteristics 1W/m 8W/m 3 I pv D ID Rp Rs I 1 Conn1 pv 1 5 1 15 5 Fig. 3: owervoltage characteristic of panel at different solar irradiance. Conn Fig. : Equivalent circuit for solar cell.
Global Conference on ower Control and Optimization, Kuching, Malaysia, 4, December 1 I I sc Current source oltage source opposite to the input polarity. Fig. 6 is the operation of buckboost boost converter. The relationship among the load voltage, input source voltage and duty cycle can be described as equation (4). (4) M M operating voltage Fig. 4: Currentvoltage characteristics and power at 6W/m voltage characteristics of panel. 1 WM Dm g E 1 C Conn1 IGBT s L C Load a Conn Fig. 5: Circuit diagram of buck boost converter. divided into two modes, namely on state and off state. During the on state the IGBT is turned on and the diode is reverse biased. The current from the input source flows throughh the inductor. When IGBT is turned off during off state, the energy stored in the inductor willl be transfered to the load until the next on state. By varying the duty cycle, the output voltage is changed accordingly. The duty cycle however can be delivered by MT control unit. In buckboost converter, the output polarity is 1 Conn1 oc.3. erturb and observe MT erturb and Observe (&O) MT has widely been used to track the M by continously changing the operating voltage point of solar panel. This method applies a little increase or decrease in operating voltage to the panel and compare the output power at present with previous perturbation cycle [8][9]. Fig. 7 shows the operation of &O. The operation of &O MT is started with the measurement of voltage () and current (I). Comparison has been made among two parameters (voltage and power) between actual state and previous state 1. There are total of four cases to be considered in &O MT. Fig. 8 is the powervoltage characteristic of SHAR E8EEA for the four cases under discussion. The module is operated at 6W/m solar irradiance at 5C. Case I where k > k1 and k > k1, the situation can be described as path A in Fig. 8. Therefore, a small voltage need to be added on the present voltage in order to approach M operating point. k> k1? Start Measure k, I k, k1 1, I k1 k= k I k k1= k1 I k1 k> k1? k> k1? s L C Loadd a Conn k1= k k1= k "on" state Dm L1 C1 Load1 a "off" state Fig. 6: The operation of buckboost converter. M continue track? End Fig. 7: Flowchart of &O MT. = es = o
Global Conference on ower Control and Optimization, Kuching, Malaysia, 4, December 1 Maximum power ath A ath B Maximum power operating voltage Fig. 8: rinciple for M tracking. Case II where k > k1 and k < k1 can be illustrated as path B in Fig. 8. It should have reducing of on the present voltage. Case III where k < k1 and k > k1 can be described as path B in Fig. 8, should have reduction on the present voltage. However case I where k < k1 and k < k1 can be illustrated as path A in Fig. 8, having addition on k. A common problem that arises in &O MT algorithm is the array operating voltage being perturbed every cycle [9]. In general, the tracking of M will never be ended unless the system is stopped for operation. Even if the M is reached, &O MT is still continually changing the operating voltage for module, hoping the next cycle has higher output power. The oscillation of the operating voltage has caused in the power loss in the system. Thus, the implementation of fuzzy logic is expected to reduce the oscillation of the operating voltage and hence minimize the power loss in the system. 3. logic MT logic has been introduced in M tracking in system lately [5]. logic is easy to use due to their heuristic nature associated with simplicity and effectiveness for linear and nonlinear systems. Among the advantages are fuzzy does not need accurate mathematical model; fuzzy can work with imprecise inputs; fuzzy can deal with nonlinearity; and fuzzy are more robust than conventional noncontrol can be linear controller [1]. The operation of fuzzy logic classified into four basic elements, namely fuzzification, rule base, inference engine and defuzzification. The fuzzification is the process of converting the system actual value into linguistic fuzzy sets using fuzzy membership function. The membership function is a curvature that describes each point of membership value in the input space. rule base is a collection of ifthen rules that contains all the information for the controlled parameters. It is set according to professional experience and the operation of the system control. inference engine is an operating method that formulates a logical decision based on the fuzzy rule setting and transforms the fuzzy rule base into fuzzy linguistic output. Defuzzifier is a manner to convert the linguistic fuzzy sets back into actual value. In [1], [3] and [5], the derivative / and change of / become the inputs of fuzzy controller for duty cycle tuning. In [], the author has selected the change of power and change of voltage as the inputs and a voltage reference as the output of fuzzy controller. The inputs of fuzzy in [4] are change of power and change of current whereas the output is the converter current reference. In [8], the two inputs of fuzzy controller are / and the previous duty cycle. The fuzzy decides the output duty cycle based on the fuzzy inputs. In [1], fuzzy works by tuning the duty cycle according to voltage error and change of voltage error. In this paper, change of power and change of power with respect to change of voltage / has been selected as the inputs of fuzzy controller which and / can be described as in equation (5) and equation (6) respectively. (5) (6) where is the current state and 1 is the previous state. Based on thesee two inputs, fuzzy will decide the size of perturbed voltage to &O MT for further process. The membership function of input has the range of [ 1.3] and input / has the range [ 5] whereas the range of output is set [ 1.]. In this paper, fuzzy is set to work according to the magnitude of inputs and decide the magnitude of output for &O MT. Fig. 9 shows the membership function output. It can be noticed that the membership function is set not to be averaged along the range. The output has total of six membership functions for the range [.5] whereas only three membership functions for the range [.4 1.]. The main purpose of this setting is that the fuzzy been expected to be more sensitive at range [.5]. At this range, the panel operating voltage is expected approaching the maximum power. Hence it should have minimum perturbed voltage reference to minimize the voltage fluctuation. However, the range [.4 1.] of output aims to lead &O MT track the maximum power operating voltage point faster. Fig. 9: Membership functions of fuzzy output.
Global Conference on ower Control and Optimization, Kuching, Malaysia, 4, December 1 rule base is an important element in fuzzy logic controller. rule base collects all the data which fuzzy inference engine will determine a logical conclusion based on the collected data. viewer is a tool to verify if the rules are set properly. viewer is shown in Fig. 1. Each row of plot in fuzzy viewer represents one rule. The fuzzy is set to have 19 rules and hence there are 19 rows in rule viewer. Row 15 in Fig. 1 declares that if the change of power is very low, regardless of the /, the perturbed voltage reference is set to be the lowest. The output decision of fuzzy can be checked via adjusting index line of fuzzy inputs. Fig. 1 shows the index line of input has been adjusted to.44 and the index line of input is set to.3. Through fuzzy inference engine calculation, the output perturbed voltage is.348. Subsequently, the output can be checked to validate the tuning parameter. The defuzzification method used in fuzzy based MT is centroid, which computes the centre of arc under curve. From Fig. 1, the areas of row 4, 5, 8 and 9 are accumulated and the areaa under curve is calculated as.348. 4. Simulation results and discussion The performances of &O MT and fuzzy based &O MT have been investigated and compared. Fig. 11 shows the results of maximum power operating voltage point versus time and maximum power versus time at 1W/m and 6W/m solar irradiance. From Fig. 11, it is noticed that both &O MT and fuzzy based &O MT can track the maximum power operating voltage point. However, fuzzy based &O MT can track the maximum power operating voltage faster than conventional &O MT. The M tracking of fuzzy MT is approximated 56% faster than & &O at 1W/m solar irradiance and approximated 45% faster at 6W/m solar irradiance. Fast tracking M can lead to more production of electrical energy from module. In addition, the perturbed voltage around M operating voltage of both controllers has been analyzed. When system has identified the M, both controllers are able to direct module to oscillate around M operating voltage. However, fuzzy based MT can provide less perturbed voltage compared to & &O MT for both 1W/m and 6W/m cases. Less perturbed voltage will lead to a more steady output power. Through the observation of &O MT and fuzzy based &O MT on single solar irradiation (6W/m and 1W/m ) the performance of fuzzy MT is better than conventional &O MT. Both controllers have been tested under variable solar irradiance. Fig. 1 is the results of &O MT whereas the results of fuzzy MT have been shown in Fig. 13. Initially, the solar irradiance is set to 8W/m for 18s. Both controllers are able to approach the maximum power voltage operating point and achieve the maximum power gaining. The reference of M voltage operating point and maximum power of each solar irradiation can be obtained from characteristics as in Fig. 3. However, &O MT consumes more time to track the M operating voltage point. MT can track the M voltage operating point at least 3% faster than &O. The fact that fuzzy MT can track M faster is evident at time equals to 18s and 47s of Fig. 13, where fuzzy MT reaches maximum power ahead than &O MT. When the maximum power has been successfully tracked, fuzzy based &O MT will reduce the perturbed voltage. Comparing the M operating voltage point in Fig. 1 and Fig. 13, &O MT has larger perturbed voltage. The larger perturbed voltage will lead to an unstable output power. The maximum power is still continuously being tracked in both &O MT and fuzzy &O MT cases even after the maximum power has been discovered. &O MT has larger oscillation around the M operating voltage. However in the fuzzy based MT, there are only small perturbed voltage being increased or decreased to the system. 18 16 14 1 1 8 6 M operating voltage at 1W/m M operating voltage at 8W/m Maximum power operating voltage at 1W/m solar irradiance Maximum power operating voltage at 6W/m solar irradiance 18 &O 16 14 1 1 8 6 &O 4 4 5 1 15 5 5 1 15 5 9 M at 1W/m Maximum power at 1W/m solar irradiance 5 M at 8W/m Maximum power at 6W/m solar irradiance 8 7 45 4 6 5 &O 35 3 5 &O 4 3 15 1 1 5 Fig. 1: Rule viewer for parameter verification. 5 1 15 5 Fig. 11: Comparison of performance &O MT and fuzzy based &O MT. 5 1 15 5
Global Conference on ower Control and Optimization, Kuching, Malaysia, 4, December 1 Fig. 1: &O MT maximum power operating voltage and maximum power against time. Fig. 13: MT maximum power operating voltage and maximum power against time. 5. Conclusion This paper presents the comparison of fuzzy based &O MT and conventional & &O MT. The characteristics and I characteristics of SHAR E8EEA have been modeled in MATLABSIMULIK to examine the performance of both controllers. Based on the simulation results, it can be concluded that both controllers can assist panel to deliver maximum power. However, the performance of fuzzy MT is better. MT can track M faster than conventional MT even in variable changes of solar irradiance. In addition, fuzzy MT has the capability of reducing the perturbed voltage when M has been recognized. This action directly preserves a more stable output power compared to the conventional MT where the output power fluctuates due to larger perturbed voltage around M voltage point. t t 6. References [1] oppadol Khaehintung, haophak Sirisuk, and Anatawat Kunakorn, Gridconnected photovoltaic system with maximum power point tracking using selforganizing fuzzy logic controller, IEEE ower Electronics and Drives Systems, EDS, Kuala Lumpur, 5, pp. 51751. [] S.Lalouni, D. Rekioua, T. Rekioua, and E. Matagne, logic control of standalone photovoltaic system with battery storage, Journal of ower Sources, olume 193, Issue, 5 September 9, pp. 89997. [3] M.S. Ait Cheikh, C. Larbes, G. F. Tchoketch Kebir, and A. Zerguerras, Maximum power point tracking using a fuzzy logic control scheme, Revue des energies Renouvelables, ol. 1, 7, pp. 387395. [4] Subiyanto, Azah Mohamed, M. A. Hanan, and Hamimi Fadziati Adb Wahab, hotovoltaic maximum power point tracking using fuzzy logic controller, roceeding of the Regional Engineering ostgraduate Conference, 9, pp. Elec151Elec158. [5] Wang ChangChun, Wu MingChuan, and Ou Shenguan, Analysis and research on maximum power point tracking of photovoltaic array with fuzzy logic control and three point weight comparison method, Science China ress and Springererlag Berlin Heidelberg, ol. 53, August 1, pp.183189. [6] M. G. illalva, J. R. Gazoli, and E. Ruppert F., Modeling and circuitbased simulation of photovoltaic arrays, Brazilian Journal of ower Electronics, ol. 14, 9, pp. 3541. [7] Dimosthenis eftitsis, Georgios Adamidis, anagiotis Bakas, and Anastasios Balouktsis, hotovoltaic system MTracker Investigation and implementation using DS engine and buckboost DCDC converter, IEEE ower Electronics and Motion Control Conference, 13 th EEEMC, 8, pp. 184 1846. [8] A. Chouder, F. Guijoan, and S. Silvestre, Simulation of fuzzy based tracker and performance comparison with perturb and observe method, Revue des energies Renouvelables, ol.11, 8, pp. 577586. [9] Roberto Faranda, and Sonia Leva, Energy comparison of MT techniques for systems, WSEA transactions on ower Systems, 8, pp. 446455. [1] Mummadi eerachary, Tomonobu Senjyu, and Katsumi Uezato, Feedforward maximum power point tracking of systems using fuzzy controller, IEEE transactions on Aerospace and Electronics Systems, ol. 38, Issue 3,, pp. 969981.