SVM Classifier Based Fault Diagnosis of the Satellite Attitude Control System

A fault detection and diagnosis scheme using SVM classifier is explored in this paper, First, NPCA method is applied to generate residual for classification. Second, one-against-one SVM classifiers and DS theory are combined to a kind of new multi-classifier, which is used for sensors or actuators faults of satellite attitude control system (ACS). The combination method we proposed not only effectively lighten the computing burden but also could keep the classification accuracy. Research result shows that this method for detection and diagnosis of the ACS faults is feasible.