Sequence Impedance Model Identification of Grid-connected Inverter Based on RBF Neural Network under Weak Network

Abstract

The impedance model of the inverter system is one of the important tools for analyzing stability. For the grey / black box system, the impedance measurement method usually used can only obtain the impedance information under certain operating conditions. However, the change of grid impedance in practical application will lead to the change of operating conditions and frequency coupling effect of inverter system, resulting in large difference of impedance information of inverter. Therefore, this paper proposes a sequence impedance model identification method of grid-connected inverter based on RBF neural network under weak network. Firstly, the discrete impedance data under different grid impedances are obtained by impedance measurement, and then the surrogate model between grid impedance and inverter sequence impedance is established by using RBF neural network. This surrogate model can obtain the amplitude and phase information of inverter impedance at any grid impedance and frequency within the training range. The simulation results verify the effectiveness of this method. This paper provides a new method for modeling grid-connected inverter.