An Optimal Neural Network Controller for a PV Charging System using the MOGA/BPA

Abstract

This paper proposes an optimal integrated neural network controller (NNC) based on maximum power point tracking (MPPT) technique and voltage regulation (VR) for a PV charging system with lead-acid battery through the constant current and constant voltage (CC-CV) charge, denoted by NNCCC/NNC-CV. The proposed controller is optimized through the hybrid multi-objective genetic algorithm and back-propagation algorithm (hMOGA/BPA). By means of this optimization, the number of system parameters is significantly reduced while maintaining high MPPT and VR accuracy. After determining the NN parameters using the hMOGA/BP, the performances of charging control against rapidly changing ambient solar irradiance and module temperature are evaluated in terms of the tradeoff between transient response, stabilized MPPT and VR accuracy, charging time, and energy utilization and charging efficiency. As results, the proposed charge controller outperforms the non-optimal NNC and on/off controller. Furthermore, validation of charging control under weather variations (i.e. fine, rainy, and cloudy), several criteria are assessed to verify the performance of the proposed NNC.