Comparison of Radial Basis Function Neural Network and Back Propagation Neural Network in Controller for Photovoltaic-Array Modeling and Maximum Power Point Tracking

Photovoltaic array has highly non-linear characteristics, it is difficult to model with analytical methods and Photovoltaic array mathematical model requires detailed knowledge of physical parameters relating to the solarcell material. This information may not be readily available to the users, therefore the derived mathematical model may be inaccurate. The neural networks model does not require any mathematical equation of photovoltaic array, so it has more and more attraction for modeling photovoltaic array. Back propagation neural network has been widely used for the maximum power point tracking in photovoltaic system. However, Back propagation neural network is multiple layers network with longer training process and lower training precision. Radial basis function neural network has a simple network construction with excellent ability to approach any complex non-linear function and a fast training speed. This paper sets up the model of photovoltaic array based on the Radial basis function neural network to decrease the complexity of the modeling, in which the solar radiation and ambient temperature impacting on photovoltaic array are input variables and the output voltage and current corresponding to the maximum power output of photovoltaic array are output variables. Compared with Back propagation neural network, the radial basis function neural network is an excellent modeling method for forecasting the maximum power capacity of photovoltaic array, which is verified by the simulation results based on MATLAB SIMULINK in the paper.