Adaptive Takagi–Sugeno–Kang Fuzzy inference with reinforcement learning to withstand extreme low voltages of DFIG-WT

Doubly fed induction generator (DFIG) wind turbines (WT) are prone to low voltages due to high wind penetration inducing generator oscillations. Adaptive Takagi–Sugeno–Kang-fuzzy (TSKF) conditional inference (CI) to withstand severe low voltage for a DFIG-WT is presented in this article. Detailed dynamic model of the DFIG-WT is derived to investigate its ability to ride through different levels of low voltages. TSKF infers the phase angle and modulation indices of the static compensator (STATCOM) connected to DFIG. Double Q -learning is designed to adapt TSKF-CI rule-base weights due to DFIG transients. The proposed method is selected to overcome the limitations of fixed-parameter controllers and conventional fuzzy systems, by combining the interpretability and nonlinear mapping capability of TSKF inference with the adaptive learning and reduced overestimation bias of Double Q-learning under highly dynamic and uncertain operating conditions. Numerical investigation showed the developed scheme effectively eliminates electromechanical transients in an event of symmetrical three-phase to ground fault while rapidly restoring stability. Double Q -learning accurately maps between the inference-scheme output and system error intensity. The proposed scheme effectively withstand severe low-voltages across different levels of electromechanical transients relative to conventional optimized controls such as PID and MPC, respectively. Practically, the proposed scheme offers a viable solution for improving grid compliance and operational reliability of wind–integrated power systems without significant hardware modification. Socially, enhanced stability of renewable energy systems facilitates higher wind energy penetration, supporting sustainable energy transitions and reduction of carbon emissions.