Abstract
Rapid fluctuations in solar irradiance pose substantial challenges to the transient stability, power quality, and operational efficiency of grid-tied photovoltaic (PV) inverter systems. Conventional linear controllers and standard finite-control-set model predictive control (FCS-MPC) algorithms often suffer from degraded dynamic tracking, elevated current total harmonic distortion (THD), and parameter mismatch vulnerability during severe atmospheric transients. To address these limitations, this paper proposes an adaptive model predictive control (A-MPC) strategy designed for three-phase grid-connected PV inverters operating under dynamic weather conditions. The proposed framework integrates an online recursive least-squares parameter estimator that continuously updates the internal model of the grid interface filter and equivalent grid impedance. Concurrently, an adaptive cost function dynamically adjusts its weighting factors according to the instantaneous operational regime, mitigating DC-link voltage fluctuations and minimizing current distortion. Experimental validation on a 15-kW hardware-in-the-loop (HIL) test platform demonstrates that the proposed A-MPC scheme achieves a current THD of 1.82% under steady-state conditions and suppresses DC-link voltage overshoot by 64% during steep irradiance transitions (from 1000 W/m² to 200 W/m² within 10 ms). Furthermore, the scheme preserves maximum power point tracking efficiency above 99.5% while ensuring robust low-voltage ride-through compliance, proving its efficacy for modern distributed renewable generation infrastructures.
