Artificial intelligence enhanced adaptive damping for electric vehicle-grid stability: A hybrid particle swarm optimization and affine projection mixed norm approach

The integration of electric vehicle batteries into power grids for frequency regulation presents significant control challenges for power electronics engineers and smart grid designers, particularly due to high-frequency inverter harmonics. While Inductor-Capacitor-Inductor (LCL) filters effectively suppress these harmonics without causing inrush current, their inherent resonance behavior—when coupled with parameter variations and dynamic operating conditions—may distort battery current and destabilize the system. Fixed-gain controllers in traditional damping methods based on capacitor current feedback struggle with wide-range grid impedance variations, LCL filter parameter deviations, and rapid reference current changes. To address these challenges, this study proposes an artificial intelligence-driven hybrid framework combining offline particle swarm optimization with real-time affine projection mixed norm adaptation. The proposed solution offers practical benefits for electric vehicle-grid integration, including reduced manual tuning effort, improved power quality under weak grid conditions, and compliance with grid standards, making it directly applicable to grid-connected inverter design, renewable energy systems, and vehicle-to-grid applications. Simulation and experimental validation demonstrate superior performance: a 50% expansion in the effective damping region (reaching 25% of the sampling frequency), while maintaining <5% total harmonic distortion under extreme grid impedance variations (−60% to +200%).