In order to fully exploit the useful information of winger Time-Frequency Spectrum,a fault diagnosis scheme based on winger distribution of vibration signal combining with singular value decomposition is proposed in this paper. Firstly,winger distribution is applied to analyze the original vibration signal,then,analysis to winger spectral matrix based on singular value decomposition is conducted,and characteristic sequences which reflect the mechanical fault state is achieved,finally,taking the singular value of winger spectral as eigenvector to do the fault diagnosis with Support Vector Machine（ SVM）. The experimental results show that this method can effectively extract fault features.