Fault diagnosis of electromechanical actuator based on principal component analysis and support vector machine

The communication proposes a method to classify the fault mode of EMAs. Traditional fault diagnosis just purposes on single components (BLDCM, ball screw, LVDT) signal analysis of EMA. Under actual circumstance, the signal detection for closed-loop electromechanical actuator system including controller and mechanical device cannot reflect single component fault. Data adopted by sensor is multidimensional and difficult to deal with. Principal component analysis(PCA) is capable to change the projection surface which decreases the amplitude of low correlation thereby reduce the data dimension. Support vector machine(SVM) is applied to data classification of less sample. This paper employs PCA to process the data adopted from mathematical model in Simulink, integrates the result and classifies the fault mode by SVM at the end.