On the basis of the theory of local mean decomposition（LMD）

the influence of random noises and local strong disturbances embedded in signal on LMD results is analyzed.A novel integrated morphology-singular value decomposition filter-LMD of bearing fault diagnosis method is presented.In the method

the mathematical morphology transform combining with singular value decomposition is taken as the pre-filter process unit in order to reduce the unnecessary decomposition levels and boundary accumulated errors caused by local stronger disturbances.The process of this method is described in detail using a simulated signal analysis example.Moreover

the method is applied in the signal processing of the actual bearing fault data and the results verify its practicality and validity.