A New Control Strategy for Improving the Voltage Quality of Inverter Grid Connected System

A Bayesian regularization neural network control strategy has been proposed to address the impact of changes in grid impedance on the output voltage amplitude and frequency of inverter grid connected systems, given an ideal SVPWM input voltage. The LCL type three-phase inverter grid connected system model has been built in the first step, and it has been demonstrated that a given ideal SVPWM input voltage can eliminate the influence of grid impedance changes on the common coupling point (PCC) voltage frequency of the grid connected system. Based on this, the duty cycle of the buck boost converter has been predicted through a neural network to adapt to changes in grid impedance, and direct voltage control has been achieved. Then, the performance function of the Bayesian regularization neural network has been elaborated, and the control performance of the neural network has been verified. Finally, a traditional resonant damping control system was constructed, and the performance of the proposed control strategy and traditional resonant damping control method was compared through Simulink simulation results. The experimental results show that the proposed control strategy can avoid the impact of grid impedance changes on the output voltage quality of PCC points, and expand the adaptability range of inverter grid connected systems to grid impedance changes.