Isolated singularity points recognition of hydroelectric generators fault signals based on CWT

In order to extract fault feature information from low frequency vibrating transient signal of the main shaft of hydro-generator, an elective method is put forward. With bi-orthogonal spline wavelet, the method is designed to locate the isolated singularity points and estimate their singularity degree by applying WTMM (wavelet transformation maximal module) and MRA (multi-resolving analysis) of continuous wavelet transform (CWT) and least square algorithm and taking account of the length, intensity and Lipschitz exponent of WTMM lines. Result shows that the method has perfect effect.