Fault Condition Detection Based on Wavelet Packet Transform and Support Vector Data Description

In order to solve the problems of correctly identifying fault condition and accurately monitoring fault development in industrial production, a new fault condition detection and identification method based on wavelet packet transform and support vector data description (SVDD) is described. For the nonlinear monitoring systems, the key to fault condition detection is main feature extracting. Wavelet packet transform, as a novel technique of signal processing, possesses excellent characteristic of time-frequency localization and is suitable for analysing the time-varying or transient signal, And support vector data description is adopted for abnormal fault detection, in which the normal and abnormal conditions can he distinguished by one-class classifier, set up only on the base of samples in normal conditions. In the experiment, According to the frequency domain feature of hoist motors of mining vibration signal, energy eigenvector of frequency domain is extracted using wavelet packet transform method, then fault condition of hoist motor is recognized using SVDD classifier. Experiment results indicate that the proposed method affords credible fault detection and identification.