Implementation of Maximum Power Point Tracking (MPPT) Technique on Solar Tracking System Based on Adaptive Neuro-Fuzzy Inference System (ANFIS)

Characteristic I-V of photovoltaic is depended on solar irradiation and operating temperature. Solar irradiation particularly affects the output current where the increasing solar irradiation will tend to increase the output current. Meanwhile, the operating temperature of photovoltaic module affects the output voltage where increasing temperature will reduce the output voltage. There is a point on the I-V curve where photovoltaic modules produce maximum possible output power that is called Maximum Power Point (MPP). A technique to track MPP on the I-V curve is known as Maximum Power Point Tracking (MPPT). In this study, the MPPT has been successfully designed based on Adaptive Neuro-Fuzzy Inference System (ANFIS) and integrated with solar tracking system to improve the conversion efficiency of photovoltaic modules. The designed ANFIS MPPT system consists of current and voltage sensors, buck-boost converter, and Arduino MEGA 2560 microcontroller as a controller. Varying amounts of lamp with 12V 10W rating arranged in series is used as load. Solar tracking system that is equipped with MPPT ANFIS able to increase the output power of photovoltaic modules by 46.198% relative to the fixed system when 3 lamps is used as load.


Introduction
The main problem found in the Solar Power Generation System nowadays is low conversion efficiency of photovoltaic module. One way to improve power production from photovoltaic modules is to have it equipped with a solar tracking system. Solar tracking system will keep phovoltaic surface oriented toward sun, allow the module exposed to higher amount of solar irradiation, to produce maximum power [1,2].
Characteristics of a photovoltaic cell is expressed by current versus voltage curve (I-V curve) and power versus voltage curve (P-V) that is influenced by solar irradiation level and temperature of photovoltaic module. There is a particular point on a I-V curve where photovoltaic will produce highest possilble power output called maximum power point (MPP) [3,4]. Process of finding MPP to maximize power extraction is called maximum power point tracking (MPPT). The aim of this study is to develop a MPPT for photovoltaic system equipped with solar tracking system based on Adaptive Neuro-Fuzzy Inference System (ANFIS).

Photovoltaic
Photovoltaic is a semiconductor device that exhibit photovoltaic effect that convert sunlight energy into electrical energy. The equation that describe the single-diode model presented in Fig. 1  Characteristic curve of photovoltaic is depended on solar irradiation and operating temperature. Solar irradiation particularly affects the output current where increasing solar irradiation will tend to increase the output current. Meanwhile, operating temperature of photovoltaic module affects the output voltage where increasing temperature will reduce the output voltage [5].

Maximum Power Point Tracking (MPPT)
Maximum power point tracking is tracking method of MPP to obtain maximum possible power from photovoltaic during daylight. The goal of the MPPT is to match the equivalent resistance at the terminal of photovoltaic R eq to the optimal output resistance Ropt that is defined as [5] (3) When R eq = R opt condition met, the MPP is obtained thus maximum possible power will be produced by photovoltaic modules. Process for matching R eq toward R opt to obtain MPP is illustrated in Fig. 3. Slope of the straight line is the representation of R eq value. The intersection of straight line R eq and I-V curve is operating point of photovoltaic. MPPT will alter this operating point toward MPP. In general, MPPT is consist of a DC-DC converter, a controller, and sensors. Duty cycle of the converter is used as a control variable to change the R eq value [6]. Open-voltage method is a MPPT technique based on observation that maximum power point voltage V MPP has a fixed ratio to open-circuit voltage V OC [6][7][8][9][10] that is defined as (4) Where the constant k is found to be between 0.7 -0.8 [11]. Flowchart diagram of open-voltage method is shown on Fig. 5.

Buck-Boost Converter
Buck-boost converter is a type of DC-DC converter that outputs voltage either less or greater than the input voltage. Relationship between input voltage Vi, output voltage Vo, and duty cycle D for buck-boost converter is stated as:

Adaptive Neuro-Fuzzy Inference System (ANFIS)
ANFIS is a kind of adaptive networks that incorporate both Takagi-Sugeno Kang Fuzzy Inference System (FIS) and artificial neural network [12]. ANFIS structure is consisted of five layers represent artificial neural network architecture as illustrated in Fig. 7. The square nodes represents an adaptive parts while the circle nodes represents non-adaptive sections. Paramaters of the adaptive nodes will be changed during the training process of ANFIS [9,10].   The proposed ANFIS MPPT is depicted in Figure  9. Error value and change of error is taken as the input for ANFIS MPPT controller defined as where and are current and previous error, respectively. Maximum power point voltage V MPP is obtained using Open-voltage Method. Output of ANFIS MPPT is change of duty cycle and duty cycle D value can be written as (8) where is current duty cycle and is previous duty cycle.

Design of ANFIS
In this study, ANFIS is designed using MATLAB. The proposed ANFIS consists of five gaussian membership functions for each input as shown in Fig. 10 & 11. Moreover its output is singleton as shown in Fig.12.

Simulation of ANFIS MPPT
Simulation of ANFIS MPPT is performed in PSIM 9.0 and MATLAB/Simulink. Co-simulation is carried out by implementing the ANFIS MPPT in MATLAB/Simulink, meanwhile photovoltaic module and buck-boost converter is partially run using PSIM 9.0 as seen in Figure 13.

Realization of ANFIS MPPT
ANFIS MPPT hardware developed in this study is shown in Figure 14. In general, prototype is divided into several subsystems for the ease of realization. INA219 is used as current-voltage sensor and Arduino MEGA 2560 is functioned as controller. Specification of the buckboost converter is shown in Table 2.

ANFIS MPPT Simulation Result
ANFIS MPPT is tested and simulated with varying climatic condition. Some parameters involved are shown in Table 3. Value of V OC dan V MPP of photovoltaic module as the effect of varying climatic condition is shown in Table 4. This simulation is performed to determine the V MPP tracking performance of ANFIS MPPT. The results as seen in Fig. 15 show that ANFIS MPPT has good performance in tracking V MPP with varying climatic condition. Voltage fluctuation around V MPP is the result of ANFIS controller yield excess control signal ∆d for a small value of and .  Based on Table 5, it can be known that V MPP value predicted by open-voltage method is close to actual V MPP proven by P MPPT has a small deviation from actual P MPP . The result of the experiments show that ANFIS MPPT prototype has good performance on V MPP tracking. The prototype is able to track setpoint with execution time < 5 seconds from short-circuit condition for a given V MPP as displayed in Fig. 16.

ANFIS MPPT without Solar Tracker System
This experiment is conducted to compare power produced between ANFIS MPPT and non-MPPT system. There are three variabels measured namely Pout non-MPPT, Pin MPPT and Pout MPPT. Pout non-MPPT is power that directly delivered to the load. Pin MPPT is power obtained in the input side of buck-boost converter, moreover Pout MPPT is power obtained in the ouput side of buck-boost converter. Load used for experiment is 12V 10W incandescent lamp. The Experiments are conducted on August 3th, 2017.

Experiment I
This experiment is performed using 2 lamps as load arranged in series for each system at 11.00 -13.30. Experimental results show power delivered to the input side of buck-boost converter of ANFIS MPPT (Pin) is higher than that delivered to load in non-MPPT system. It indicated that ANFIS MPPT increases produced power from photovoltaic module. However, buck-boost converter used in ANFIS MPPT has efficiency of 70-80% thus power delivered in output side of converter is always lower than input side. According to Table 6, ANFIS MPPT system produced power gain around 34.14997 % relative to the non-MPPT system when 2 lamps were used as load.

Experiment II
This experiment is carried out using 3 lamps as load arranged in series for each system at 13.30 -16.00. When using 3 lamps as load, power produced by non-MPPT system (P non-MPPT) is almost equal to input side power of MPPT system (Pin) as shown in Figure 17. MPPT system increases produced power by 39.717277 %. It is assumed that equivalent resistance R eq of 3 lamps used is almost equal to optimal resistance R opt such that it make operating conditon of non-MPPT system is near MPP.

Experiment III
In this case, the experiment provided 4 lamps as load arranged in series for each system. It was conduted at 09.00 -11.00. Experiment result show that when 4 lamps are used as load, efficiency produced by ANFIS MPPT system is 43.21 % compared to the non-MPPT system. It can be known that operating condition of non-MPPT system has never reached MPP so that it produced less power.

ANFIS MPPT Based on Solar Tracker System
The experiment was operated using two 20 Wp PV modules as follows: Module 1 was equipped with solar tracking system and Module 2 was a fixed module. Loads used for the project consisted of 12V 10W incandescent lamps.  Result obtained from this experiment can be seen in Table 9. It shows that ANFIS MPPT can increase power produced from photovoltaic equipped with solar tracking system. Increased average power for various load are 11.75% for 2 lamps, 4,57% for 3 lamps, and 15,26% for 4 lamps.

Experiment II
In this experiment there are two systems tested as demonstrated in Figure 17. First system is module 1 equipped with solar tracking system and ANFIS MPPT. The other one is fixed system and direct-coupled with loads. Load used in this experiment is 3 lamps connected in series. The experiment conducted on August 15th, 2017 at 11.00-13.00. There are three variabels measured namely Pout tracking, Pin tracking and Pout fixed. Pout tracking is power obtained in the input side of buckboost converter and Pin tracking is power obtained in the ouput side of buck-boost converter. Both of them are measured and determined from the first system. Meanwhile Pout fixed is power that directly delivered to the load in the second system.