Research The Method For Rolling Bearing Fault Diagnsis Based On Local Mean Decomposition

In modern machinery equipment, rotating machinery equipment account for a large proportion, and the rolling bearing is the key component of them. so, the research of condition monitoring and fault diagnosis for rotating machinery equipment is extremely important.Local Mean Decomposition(LMD) time-frequency analysis method as a new method for signal Decomposition, is very suitable for processing and analyzing the signal of non-stationary and non-linear properties, especially, the mechanical vibration signals of multi-component am-fm signals. This article focuses on the rotating machinery fault diagnosis method based on the LMD. Its main content is as follows:First, aiming at false components problem of LMD, studied a method based on the improvement of stopping criterion for iteration and the decomposition of residual component. Using the simulation signal and actual fault signal demonstrate the feasibility and effectiveness of this approach.Then, a rolling bearing fault diagnosis method based on Local Mean Decomposition multi-scale entropy and probabilistic neural network was studied. Machinery vibration signal is decomposed by LMD firstly in this method. Calculating the multi-scale entropy of PF components secondly. Finally the entropy as the fault feature vector input into the probability neural network to pattern recognition, realize the diagnosis of damage location and damage degree.Finally, combined the time-frequency analysis method with mathematical morphology, puts forward a new kind of fault characteristic parameter, multi-scale entropy spectrum. And combined with LMD, studied a rotating machinery fault diagnosis method based on LMD and multi-scale entropy spectrum analysis. Experimental results show that, this method can extract the fault features of rolling bearing fault vibration signal effectively, and realize the fault type diagnosis of rolling bearing accurately.