Actuator Fault Diagnosis Research of Control System Based on EWT-SOM Method

As the execution terminal component of the control system, the actuator fault often leaded to a series of serious consequences, and the harsh working environment further leaded to a higher fault rate of the actuator. In this paper, a fault diagnosis method of control system actuator based on the combination of empirical wavelet transform and self-organizing map neural network was proposed. The method used the empirical wavelet transform to mine fault signal features in depth at first, and then used the self-organizing map neural network to carry on fast fault classifications. Finally, the method proposed in this paper was verified by simulation experiments on the actuator fault simulation semi-physical test platform, and we compared the proposed method with the wavelet neural network fault diagnosis method. The results showed that the proposed fault diagnosis method can efficiently complete the task of actuator fault diagnosis, and the diagnosis accuracy was better than that of wavelet neural network.