Research on Fault Early Warning Method of Control Torque Gyroscope Based on Support Vector Data Description

The control torque gyroscope(CMG) is a high-speed rotating component in the attitude control system of spacecraft. Long-term high-speed operation makes it prone to malfunction If early warning is given before obvious faults occur, it is of great significance for improving the service life of spacecraft. This paper proposes a CMG fault early warning model based on Support Vector Data Description(SVDD). Firstly, introduces the Control Moment Gyroscope(CMG) and its common fault early warning methods, and points out that the traditional fault early warning methods rely too much on expert experience and prior knowledge, which leads to the SVDD method. Then the mathematical principle of the SVDD algorithm and the steps of the SVDD algorithm for fault warning are introduced. The evaluation index of fault diagnosis model is briefly mentioned. Finally, the fault warning ability of SVDD method is tested by using the fault data of a CMG. By comparing with the fault warning results of KPCA-SPE algorithm, it is shown that SVDD algorithm is superior to KPCA-SPE algorithm in terms of alarm accuracy and false alarm number, indicating that it has good fault warning ability.