A Data-Driven Control Parameter Optimization Framework for Enhancing Frequency Stability in High-Renewable-Penetration Power Systems

As the penetration rate of renewable energy continues to rise, the equivalent inertia of power systems has significantly decreased, leading to a marked degradation in frequency stability support capabilities. Under conditions of high renewable energy penetration, the question of how to effectively enhance grid frequency support capacity has become a critical research topic in the field of power system operation and control. This paper first systematically analyzes the impact of key control parameters on the frequency dynamic response of power systems. It investigates the intrinsic relationship between these parameters and system frequency stability through both analytical frequency response modeling and time-domain simulation analysis. A frequency stability margin metric is constructed based on the grid frequency response process to quantify the system’s frequency stability performance. Building upon this foundation, an improved ResNet-based frequency stability margin prediction model is established to enable rapid estimation of the frequency stability margin. Furthermore, Bayesian optimization is introduced to optimize frequency control parameters, thereby enhancing system frequency stability. Case studies conducted on the simulation system CSEE-FS with insufficient frequency support capability demonstrate that the proposed method effectively increases the frequency stability margin and significantly improves the system’s frequency response performance.