Hybrid optimization of data-driven fuzzy controller for DFIG-based wind energy systems

This study proposes a data-driven fuzzy control framework for DFIG-based wind energy systems, integrating global and local optimization to enhance the performance of the rotor-side (RSC) and grid-side (GSC) converters. The methodology comprises two complementary stages: global optimization using a genetic algorithm (GA) and particle swarm optimization (PSO) for fuzzy rule learning, followed by local refinement through pattern search (PS) for membership function tuning. A 49-rule Type-1 TSK fuzzy system with triangular membership functions was adopted to model the nonlinear control behavior. The proposed approach was evaluated using multiple performance indicators, including the root mean square error (RMSE), error distribution, and convergence characteristics. Furthermore, dynamic and transient performances are thoroughly assessed in the presence of grid-side faults such as three-phase-to-ground (LLLG) faults. The results obtained at wind speeds of 10 and 15 m/s demonstrated significant improvements in transient performance, characterized by reduced overshoot, faster settling time, and smoother dynamic responses. Quantitatively, the RSC active power settling time improved from 0.11 to 0.10 s, whereas the GSC DC-link voltage overshoot decreased from 14.3% to 8.63% under real-time validation using the OPAL-RT OP5700 platform. Comparative analysis indicates that hybrid optimization strategies, particularly the combination of GA and PS, provide superior accuracy, stability, and convergence. The proposed framework maintains the interpretability of fuzzy systems while enhancing the control performance, thereby offering a robust and scalable solution for grid-connected wind energy applications.