Optimized data-driven type-2 fuzzy logic control for DFIG-based wind turbines: Metaheuristic rule learning and real-time implementation

This study introduces a data-driven two-stage optimization method for a type-2 fuzzy logic controller to improve the performance of doubly-fed induction generator wind farms under varying wind and fault conditions. The developed methodology uses genetic algorithms (GA) and particle swarm optimization (PSO) for systematic fuzzy rule learning, followed by the optimization of membership function parameters. In the two-stage approach, GA and PSO are first applied for fuzzy rule learning at fixed membership function parameters, which decreases the computational complexity and accelerates the convergence. In the second stage, both the membership function parameters and rule weights are optimized using pattern search methods, resulting in improved tracking performance and control precision. The applicability of the developed method was demonstrated for different wind speeds and three-phase-to-ground fault conditions. Real-time experiments on the OP5700 simulation model-in-the-loop were performed, and the results showed that the developed method successfully bridged theory-based simulations and practical applications. A comparison of the results of this study with those obtained using conventional proportional–integral (PI) controllers showed that the developed method significantly improved the dynamic response, steady-state error, and fault ride-through capability. The developed data-driven method also has lower root mean square errors (RMSE) and faster convergence rates than the traditional method. Therefore, this study provides an applicable and scalable solution for operationally stable and efficient modern wind farms, provides valuable insights for designing intelligent controllers, and serves as a reference for renewable energy systems.