Artificial Neural Network Based Maximum Power Point Tracking of a Photovoltaic System

Nonlinear behavior of photovoltaic system under changing environmental conditions, creates the requirement of designing maximum power point tracker (MPPT). MPPT is an electronic device that enables the system to draw the maximum available power, what PV panel is capable of generating under changing temperature and irradiance by using several types of algorithms and methods. Conventionally perturb and observe (P&O) method and incremental conductance method have been used for designing MPPT, as they provide a simple and cost effective way of tracking. But either they have drawback of slow operation or low tracking efficiency. However, Artificial Neural Networks (ANN) can be deployed for this purpose, as they can track the optimal operating point (MPP) quickly and accurately based on training data sets. This paper presents three ANN based MPPT methods: Direct ANN, Cascaded ANN and the proposed Hybrid method. The ANN's are trained using training data set of maximum power point voltages (V
_MPP
) from a user defined PV array over different environmental conditions (temperature and irradiance). In this research work, a user defined PV array and three MPPT models are simulated in SIMULINK.