Neural Network Assisted Variable-Step-Size P&O for Fast Maximum Power Point Tracking

Abstract

This work proposes an ultra-fast Maximum Power Point Tracking (MPPT) algorithm for Photovoltaic (PV) system. The objective is to combine the Variable Step Size Perturb and Observe (VSS P&O) algorithm and the Neural Network (NN) algorithm to rapidly track the Maximum Power Point (MPP) of a PV. The role of the NN is to propose a new starting point for the P&O algorithm on every sudden climatic variation. This will reduce the searching time required by the P&O to reach the MPP. The proposed method is verified using MATLAB-Simulink simulations. Moreover, an experimental validation is carried out using a boost-converter in conjunction with a Microcontroller based system. The performance of the proposed method is compared with the conventional P&O and the VSS P&O on MATLAB-Simulink, and then with the experimental test. The results show that the proposed method tracks faster the MPP by 3 to 7 times compared to the two other methods.