Stabilization of Hybrid Photovoltaic Energy Storage System Based on RBF Neural Network Adaptive Control

This paper presents a new control strategy based on RBF (Radial Basis Function) neural network and traditional PI control to improve the dynamic performance of energy storage system to stabilize photovoltaic output power. This method only needs to identify the rate of change of output with input (Jacobian information), and can adaptively adjust the parameters of PI controller by combining with weight update algorithm, so that when the load and photovoltaic output power change suddenly, the energy storage battery can quickly and accurately compensate or absorb the unbalanced power and maintain the bus voltage stability. The simulation models of traditional double-closed-loop PI control and RBF neural network adaptive control are built. The experimental results show that the RBF neural network adaptive controller can stabilize the DC bus voltage at 380 V well, and has better dynamic performance and control effect than traditional double-closed-loop PI control.