Abstract
Due to the dependence of Photovoltaic (PV) systems on environmental conditions, their behavior is completely non-linear. Therefore, achieving the maximum power point in the nonlinear curve of PV systems faces many difficulties. The effective way to achieve the maximum power point in PVs is to design a maximum power point tracking (MPPT) controller. Although different ideas are recommended for MPPT controllers, the use of the artificial neural network (ANN) controller is very attractive among them due to its high dynamic response and fewer oscillations. Nevertheless, the major challenge for designing the ANFIS-MPPT is obtaining precise training data. This work proposes a hybrid gravitational search and pattern search (GS-PS) algorithm trained ANN-based MPPT is implemented for different environmental conditions. Radiation and temperature are two important input variables while the optimum voltage is considered for the output in the proposed method. The optimum values are attained via the GS-PS algorithm to regulate ANN controller to maintain the tracking performance. In addition, the P&O technique is started to operate in the following cycle and initiates a precise searching procedure from that point. Less number of samples for training is needed by applying the combined ANFIS and P&O method because the P&O can cover the disadvantage of the ANFIS when it cannot detect the accurate point. The simulations show the recommended method has much enhanced presentation than the previous methods in the time response as well as high accuracy. The accuracy of the offered MPPT compared to other approaches has been proven with a 2%-8% improvement in different temperature and radiation conditions.