Frequency coordinated control and parameter optimization for photovoltaic–energy storage systems based on a GA-BP hybrid algorithm

Introduction Frequency oscillations induced by stochastic disturbances pose significant challenges to grid-connected photovoltaic (PV) systems. This study proposes an adaptive optimization strategy for photovoltaic-energy storage systems (PV-ESS) based on a GA-BP neural network to address this issue. Methods First, the working principles and characteristics of virtual synchronous generator (VSG) technology are elaborated. Second, the power control point positioning under deloading operation of PV systems and the virtual inertia control of energy storage systems are analyzed. Subsequently, a GA-BP neural network is introduced and applied to the adaptive parameter design of the PV-ESS system, enabling real-time dynamic adjustment of the moment of inertia J , damping coefficient D , and virtual inertia coefficient K , thereby enhancing the dynamic response performance of active power. Results The experimental results demonstrate that under active power command mutation scenarios, compared with fixed-parameter control strategies, the proposed strategy reduces the frequency nadir deviation by 14.81%, overshoot by 62.5%, and steady-state recovery time by 44.44%. Discussion The adaptive parameter adjustment mechanism effectively mitigates frequency oscillations, offering a robust solution for grid stability in PV scenarios.