Research On Early Anomaly Detection Method Of Control Moment Gyro Based On Data Drive

Control moment gyro plays an important role in the near orbit of large spacecraft.As the attitude control component of large spacecraft,the life of control moment gyro seriously affects the life of the spacecraft as a whole.Therefore,the early abnormal detection of control moment gyro plays a vital role in the maintenance of spacecraft system.In this paper,the basic working principle of control moment gyro(CMG)is studied,and the possible fault location is understood.Data preprocessing operations such as removing outliers,filling in missing data and resampling at equal intervals are carried out on the test data.Sampling characteristics and time domain signals of the test data are analyzed.In this paper,the principal component analysis(PCA)model is built to reduce the dimension of the test data,and the abnormal detection of the control moment gyro test data is carried out by calculating the statistics.The support vector description(SVDD)method is used to map the test data to the high-dimensional space,and the hypersphere in the space is constructed to get the final decision result.The synthetic false alarm rate,false alarm rate and early warning time are used to compare and analyze the two methods.In this paper,the early healt h state of the control moment gyro was divided by constructing the convolutional neural network(CNN),and the model was constructed and parameters were optimized through convolution,pooling,and full connection operations,so as to obtain the final judgment results.The experimental results were verified by using the visual analysis method.