A Data-Driven Fault Diagnosis for Fast Steering Mirrors Based on Gap Metric

This paper addresses the fault diagnosis issue of fast steering mirrors (FSM). Based on a data-driven gap metric (DDGM) calculation method, a monitoring process is designed that directly utilizes both offline and online data to compute the gap metric between the nominal system and the observed system for fault detection (FD). Additionally, an iterative implementation method based on data-driven stable image representation (SIR) is proposed, which facilitates the calculation of the gap metric and enables FD with reduced computational complexity. Moreover, due to factors such as the actual operating environment of FSM, actuator and sensor fault frequently occur. Therefore, a method for fault classification and gradation (FCG) is developed. Finally, the accuracy of the proposed methods is validated through experiments.