The rolling bearing fault diagnosis based on LMD and LS-SVM

Rolling bearings vibration signal is complex and non-stationary signal. In order to diagnose the bearing failures accurately and quickly, propose an approach about rolling bearing fault diagnosis, which is based on LS-SVM and LMD. Firstly, decompose the original vibration signal by LMD (Local Mean Decomposition LMD) to get a series of PF(Production Function, PF); secondly, establish the AR model of PF components. And getting autoregressive parameters and residual variance of the AR model through the Burgrecursive algorithm, to constitute feature vector; finally, input the feature vector into the LS-SVM for determining the bearing running state. Experimental results show that: the method can diagnose the bearing failures quickly and accurately.