An fault detection method based on compensation of SVMR

The main idea of this paper is to use the wavelet kernel SVMR compensation to reduce the impact of uncertainties on the observer output errors. The SVMR can theoretically approximate any nonlinear function. Using the wavelet kernel SVMR to approximate the uncertainties is introduced into modeling the uncertainties of satellite attitude control system. First, this paper selects SVMR identification structure; Then, we acquire modeling samples of uncertainties through observer, which is necessary to establish SVMR model; Finally, the SVMR model is added to the established observer to offset system uncertainties, which enable to detect actuator fault more efficiently. At last, the simulation results of actuator faults of satellite attitude control system are shown the effectiveness of this method.