Rolling bearing fault diagnosis based on CEEMD energy entropy and extreme learning machines

DOI:	10.11809/bqzbgcxb2023.08.040
Keywords:	rolling bearing; fault diagnosis; complementary ensemble empirical mode decomposition; energy entropy; extreme learning machine
Abstract:	Considering the nonlinear and non stationary characteristics of vibration signals of rolling bearings and the fact that it is difficult to obtain a large number of typical fault samples, this paper proposes a rolling bearing fault diagnosis method based on complementary ensemble empirical mode decomposition (CEEMD) and extreme learning machines (ELM). Firstly, the non stationary original acceleration vibration signals of rolling bearings are decomposed by CEEMD method, and several stationary intrinsic mode function (IMF) components are obtained. Then, the energy entropy of each IMF component after CEEMD under different states is calculated, the energy features are extracted from the IMF components containing the main fault information, and the T Stochastic Neighbor Embedding (Tsne) cluster visualization analysis is carried out on theses features. Later, the reflection of the features on the rolling bearing state is observed, and these features are used as input to establish an extreme learning machine to judge the working state and fault type of the rolling bearing. Finally, the artificially damaged bearing data published by Case Western Reserve University in the United States are used. The simulation experiment shows that the accuracy of different running states of the bearings can reach 95%, which is significantly higher than that of the multi classification support vector machine, K nearest neighbor and other methods. It also shows that the method is effective in dealing with rolling bearing fault diagnosis.
Issue:	Vol. 44 No. 8 (2023)
Published:	2023-08-28
PDF HTML