Fuzzy robust model predictive fault tolerant control in wind turbine based on ST-SMO and PMIO

Abstract

Abstract Conventional wind turbine maintenance relies on post-failure diagnosis, yet sustained degraded operation remains critical. This paper proposes a fault-tolerant model predictive control (MPC) framework with switching between a super-twisting sliding mode observer (ST-SMO) and a proportional multiple integral (PMI) dual observer system. In the observation layer, the PMI observer estimates unmeasured states and various faults, while the ST-SMO specifically compensates for high-order pitch angle faults. A Takagi-Sugeno (T-S) fuzzy logic based on the $$\nu$$ -gap metric coordinates the two observers. Adaptive penalty terms in the MPC layer compensate deviations via a linear parameter-varying model. Multi-scenario case studies validate the framework under load-range transitions and simultaneous multi-fault conditions. Compared with single-observer FTC methods and conventional MPC, the proposed framework improves maximum power tracking accuracy by 18% under sensor faults, suppresses drivetrain torsional torque fluctuation by 50% under cross-load switching, and reduces tower bending moment damage equivalent load (DEL) by 19.63% under multi-fault conditions. It innovatively integrates dual observers with T-S fuzzy logic and hard-soft combined LPV switching, achieving synergistic optimization of fault tolerance, fatigue load mitigation, and active power maximization for megawatt-class wind turbines under multi-fault coupling and cross-load transitions.