RBF Neural Network Based Virtual Synchronous Generator Control With Improved Frequency Stability

Abstract

The virtual synchronous generator (VSG) based on the energy storage system is proposed to compensate the loss of inertia and damping of the power grid. Due to the introduction of inertia, VSG is more prone to power oscillation. In this article, the nonlinear relationship between inertia and angular velocity is analyzed, and adaptive neural network (NN) control is first applied to VSG. Based on this concept, an adaptive control strategy is proposed in this article. First, the radial basis function NN that enjoys a simple algorithm, strong ability of learning, and fast learning rate is used to adjust virtual inertia adaptively. This strategy not only improves response but also reduces frequency overshoot in tracking the steady-state frequency. And then, based on the fixed damping ratio, the damping coefficient is tuned adaptively with the change of the inertia to further suppress power oscillation. The proposed strategy is supported by simulation results, which show that the strategy has good performance in damping of oscillation.