A wavelet packet analysis based method for fault diagnosis of dynamic circuits

A novel method based on the wavelet packet analysis is proposed for the fault diagnosis of dynamic circuit. In the proposed method, a fixed unstable AC signal is used as the stimulant of a dynamic circuit, and the responses are captured and disassembled by the wavelet packet analysis. Many faults in dynamic circuit are sensitive to frequencies, which are the chief difficulties of traditional DC based diagnosis method. By the stimulation of the unstable AC signal, the responses of dynamic circuit should be different from normal and fault conditions, and all of dynamic fault in the CUT can be sensitized. In the further step, the fault characters are obtained by the wavelet packet analysis of the output signals. Compared to the DC based fault diagnosis method and traditional AC fault dictionary, the superiority of the presented strategy is exemplified in the success fault diagnosis. The test results demonstrate that the applying of energy criterion after wavelet packet analysis is powerful and reliable method.