Abstract:In order to accurately get fault bearing of the machine tool running state, a new method of Daul-Tree Complex Wavelet TransformLocal Mean Decomposition (DT-LMD) was proposed based on integration of DT-CWT and LMD, applying for bearing fault vibration signal extraction. Firstly, DT-CWT was used to reduce the signal noise and refactoring. Secondly, using the LMD to decompose, it was used to decompose the actual vibration signal of bearing of machine tool. The features energy of the machine tool bearing was extracted, and normalized the eigenvalue. Thirdly, the energy value of each component is gotten, finally the type of bearing fault is judged.