Abstract:The feature extraction of weak signal is of important significance for early stage fault diagnosis in spindle system.Started from the aspects of suppressing and adding noise to improve the signal to noise ratio,feature extraction methods with weak signal of spindle system using were listed based on the research results of signal processing field.These methods of signal processing are included,to improve the ratio of signal to noise to suppressing noise and using noise to strengthen weak signal.The methods of suppressing noise are Spectral Kurtosis,Empirical Mode Decomposition,Wavelet Transform,Cyclostationarity,Blind Source Separation,and Manifold Learning etc.The methods of strengthening weak signal by adding noise are Stochastic Resonance and Ensemble Empirical Mode Decomposition.References are provided for research of spindle fault diagnosis.