Enhanced high-frequency stability and power quality in photovoltaic inverters via an optimized least mean square adaptive damping controller

• A hybrid control framework (GWO-LMS) is proposed, combining offline Grey Wolf optimization with real-time Least Mean Squares adaptation for intelligent damping of LCL filter resonances in PV-grid systems. • The data-driven LMS algorithm autonomously adjusts controller gains, achieving robust stability under dynamic grid conditions without requiring precise system modeling. • GWO-based offline optimization of the proportional-resonant controller and damping coefficients reduces tuning complexity by 60%, demonstrating the synergy between metaheuristics and real-time adaptive control in power electronics. The widespread adoption of grid-connected photovoltaic systems is vital for a sustainable energy future, yet the power electronic inverters that interface them with the grid are a source of detrimental high-frequency harmonics. While LCL filters are the preferred solution for attenuating these harmonics, their design introduces a problematic resonance peak that can lead to system instability and distorted output current. Conventional active damping techniques, such as capacitor current feedback with fixed controller gains, suppress this resonance but often at the cost of limited robustness and non-adaptive performance. This paper proposes a streamlined and robust control strategy that overcomes these limitations. The proposed strategy employs a two-stage process: first, an offline stage where a Grey Wolf Optimizer (GWO) concurrently derives the optimal parameters for the proportional-resonant controller and the damping coefficient; second, an online stage where a Least Mean Squares (LMS) algorithm dynamically fine-tunes the damping gains in real time based on operating conditions. The stability of the adaptive control law is rigorously guaranteed through Lyapunov stability analysis, ensuring robust performance under dynamic grid conditions. Simulation results validate the efficacy of the proposed approach, showing an extended effective damping bandwidth up to a quarter of the sampling frequency and demonstrating remarkable robustness to grid impedance variations and change in reference current.