A maximum power point tracking control for wind energy conversion systems using regularized data-enabled predictive control

Abstract

This paper applies Artificial Intelligence (AI) in the sense of data-driven learning and optimization to wind energy control. Specifically, the implemented AI method uses Data-Enabled Predictive Control (DeePC) combined with quadratic regularization for maximum power point tracking (MPPT) of a permanent magnet synchronous generator (PMSG)-based wind energy conversion system (WECS). The contribution is the use of regularized DeePC to construct a predictive controller directly from measured input/output data without explicit model identification, while improving robustness to noise and nonlinearity. The method is applied to MPPT via rotor-speed tracking and direct-axis current regulation under wind variations, disturbances, and parameter uncertainty. The quadratic regularization penalizes initial-condition mismatch and limits the trajectory parameter, mitigating prediction errors and yielding smoother control actions. The method is evaluated in a simulation with a linear quadratic regulator (LQR) and sliding mode control (SMC). DeePC achieves a settling time of 0.03 s (s) with minimal overshoot under step changes, compared with approximately 0.2 s for LQR and 0.5 s for SMC. In addition, DeePC reduces rotor-speed root mean square error (RMSE) to 0.15/0.23 radians per second (rad/s) (nominal/distorted parameters) under a realistic wind profile, compared to 0.42/0.53 rad/s for LQR and 0.76/0.81 rad/s for SMC. These results indicate that regularized DeePC is an effective data-driven alternative to model-based MPPT control within the validated operating regimes