Data-driven fault-tolerant controller design for hypersonic vehicles with sensor fault

A data-driven fault diagnosis and fault-tolerant control method are proposed for the problem of sensor faults in the attitude system during hypersonic vehicle flight. Firstly, a sliding window-based fault detection method is proposed by analyzing the residuals of the real-time data. Then, based on the support vector machine (SVM), a fault identification model is designed to classify the faults using the time domain feature parameters in the fault process. The fault dynamic response process is analyzed and the fault estimation method is proposed. Finally, the controller is designed based on the improved full-form dynamic linearization model-free sliding mode control (FFDL-MFSMC) method. And the effectiveness of the proposed fault tolerance strategy is illustrated by simulation.