Fault Diagnosis Based on IGA-SVMR for Satellite Attitude Control System

Support Vector Machine Regression is a nonlinear modeling method with a simple structure and shows excellent performance compared with other nonlinear-linear regression methods. Unfortunately, most users always select the SVMR parameters by rule of thumb, so they frequently fail to get the optimal model. This paper propose to use the immune genetic algorithm to adjust the SVMR parameters and use the RMSE of the cross validation as the fitness of IGA. At last, this method was applied to modeling satellite attitude control system to detect the faults of the system. Simulation shows the high fitting precision to the system models which insures the correctness of the fault detection.