Neural network driven dynamic response optimization and inertia coordination control method for VSG

When there are load fluctuations or frequency changes in the power grid, virtual synchronous generators (VSG) provide virtual inertia support by adjusting their output electromagnetic power. However, traditional VSG control strategies with fixed parameters struggle to balance overshoot and adjustment speed, and may even induce frequency and power oscillations similar to those in synchronous power sources. To enhance the inertia support response of VSGs and mitigate oscillations, this study proposes a neural network driven dynamic response optimization and inertia coordination control, leveraging Radial Basis Function (RBF) neural networks. By employing a self-learning algorithm, the neural network weights are optimized to more precisely model the nonlinear relationship between inertia and angular frequency. Simulation test indicate that the proposed control method effectively reduces frequency and power overshoots, shortens the adjustment time, suppresses fluctuations during frequency tracking, and improves the dynamic response performance of VSG inertia support.