Grid-connected PV System with a modified-Neural Network Control

Abstract

The integration of solar energy in the field of electricity production is becoming a growing trend, especially photovoltaic grid-connected system because of the high cost of batteries. The simulation of single-phase two-stage photovoltaic grid-connected is highlighted in this study under different climatic conditions to analyse their influence on the output performances. So, for injecting the maximum amount of power in the grid, the solar photovoltaic (PV) array must extract the maximum solar energy available and operate continuously at the maximum power point (MPP). For that, an efficient maximum power point tracking algorithm (MPPT) should be integrated. The conventional techniques present a lot of drawbacks, especially in the fast variation of meteorological and solar irradiation conditions. Two efficient MPPT algorithms based on Neural Network (NN) have been performed and simulated. To evaluate the proposed MPPT algorithms and compare them with the conventional MPPT algorithms as Incremental Conductance (IncCond), Perturb and Observe (P&O) and Open Circuit Voltage (OCV) a simulation on MATLAB/SIMULINK platform has been done under several temperatures and irradiance. The study covers the stability, time response, oscillation and the overshoot. The simulation results show a very high efficiency and small response time with high accuracy for the proposed techniques.