Fault Diagnosis for a Hydraulic Servo System Using Wavelet Packet and Neural Network

Hydraulic servo system is widely used in the project, and it's crucial for the reliability of hydraulic system. Due to the nonlinear, time-varying and noise in the hydraulic system, the traditional fault diagnosis methods have been unable to meet the requirements. To decrease false alarm rate of hydraulic servo system, an adaptive fault detection method based on RBF observer and an adaptive threshold is presented. First, an RBF observer is established to obtain the residual by subtracting actual output from the estimated output by observer. Second, input command signal and output displacement signal of hydraulic servo system were decomposed into different wavelet frequency band which was input RBF neural network to get time-frequency adaptive threshold. Third, fault is detected by comparing the residual with adaptive threshold. Lastly, experimental results demonstrated that the proposed method is effective in detecting fault of hydraulic servo system.