Adaptive control optimization in multi energy synergy scenarios robustness verification of fuzzy RBF neural network in microgrid frequency regulation

Grid-connected renewable energy systems with high penetration rates exacerbate frequency fluctuations in microgrids, making traditional control methods ill-suited for random source-load variations. This paper proposes an adaptive fuzzy RBF neural network control architecture: dynamically allocating multi-energy priorities through a fuzzy inference layer, enhancing robustness via adaptive scaling of the RBF kernel function and training with Levy noise, and utilizing NSGA-II optimization to balance frequency regulation accuracy with equipment losses. OPAL-RT hardware-in-the-loop experiments validate that under ±50% disturbances, the maximum frequency deviation drops to 0.29 Hz (a 40.8% reduction compared to adaptive RBF control), and stable control is maintained even with 20% communication packet loss. This architecture provides a robust dynamic frequency regulation solution for high-renewable-penetration microgrids.