Application of improved radial basis function neural network method in global MPPT for PV array

Abstract

To solve the problem that the structure and parameters of neural network are hard to be tuned, a modified radial basis function neural network (RBFNN) method based on improved particle swarm optimization algorithm (IMPSO-RBFNN, for short) is proposed. In the proposed method, the IMPSO algorithm is utilized to optimize RBFNN, and the nearest neighbor cluster algorithm (NNCA) is introduced into RBFNN. Finally, the experimental results show that the proposed method is effective for MPPT under partially shaded (PS) conditions and has a more stable performance in searching precision when compared to the other methods.