Data-driven active fault-tolerant control for 15MW offshore wind turbines subject to intermittent sensor faults

Sensors provide essential feedback for closed-loop control of floating offshore wind turbines, but random intermittent short-circuit faults can cause feedback information loss and degrade power regulation, structural responses, and platform stability. This study proposes a feedback reconstruction-oriented active fault-tolerant control framework for floating offshore wind turbines, where fault identification, signal reconstruction, and control compensation are coordinated according to faulty sensor information characteristics. An observer-based fault identification method distinguishes healthy and faulty sensor states. For short-circuit faults with unavailable feedback signals, a dynamic linearization model-based reconstruction method recovers fault-free feedback using historical input–output data. The reconstructed signals are incorporated into a differential-error-enhanced model-free adaptive control strategy for closed-loop regulation. The framework is validated on a 15-MW floating offshore wind turbine using OpenFAST–Simulink simulations under turbulent wind–wave loads, measured environmental excitation, intermittent faults with different severities, and measurement noise. Results demonstrate that reconstructed signals closely follow fault-free references and achieve lower reconstruction errors than Kalman-filter, moving-average, and neural-network-based methods. Compared with the baseline proportional–integral controller, the proposed framework reduces generator-power standard deviation by more than 20% under all turbulent wind–wave cases and effectively mitigates generator-speed fluctuation, tower-top fore–aft displacement, and platform pitch motion.