Rolling bearing fault diagnosis method based on multi-feature extraction and WOA-ELM

Abstract

The invention discloses a rolling bearing fault diagnosis method based on feature fusion and WOA-ELM. The rolling bearing fault diagnosis method comprises the following steps of: step 1, taking official bearing data of the Case Western Reserve University as original vibration data; 2, performing time domain analysis, spectral analysis and wavelet packet decomposition on the original data, and extracting time domain, frequency domain and time-frequency domain features; 3, performing normalization processing on a mixed domain feature set obtained in the step 2; 4, performing dimension reduction on the normalized high-dimensional feature set by using a manifold learning LPP algorithm to obtain a low-dimensional feature sample set; and step 5, optimizing ELM network parameters by using a WOA algorithm, and inputting the low-dimensional feature data into a WOA-ELM model for fault diagnosis. According to the method, the problems of insufficient feature extraction, redundant feature information existing in a multi-feature sample, insufficient stability caused by network parameters randomly generated by an extreme learning machine and the like are effectively solved. The method has the obvious advantage of improving the fault diagnosis rate of bearings.

Description

technical field [0001] The invention relates to a bearing fault diagnosis method, in particular to a rolling bearing fault diagnosis method based on multi-feature extraction and WOA-ELM. Background technique [0002] Rotating machinery is the power to support the steady development of the national economy, and machinery failure will cause serious economic losses. Rolling bearings are important components of large-scale mechanical equipment, and are extremely important to the safe operation of mechanical equipment. Therefore, it is urgent to study effective and reliable bearing fault diagnosis methods. In view of the nonlinear and non-stationary characteristics of the early faults of rolling bearings, scholars at home and abroad have continued to conduct exploratory research on the feature extraction and diagnosis models of rolling bearings. [0003] At present, although time-domain analysis methods can effectively preserve the characteristics of the original signal, they ar...

AI Technical Summary

Problems solved by technology

[0003] At present, although time-domain analysis methods can effectively retain the characteristics of the original signal, they are not sensitive to non-stationary signals.

Fast Fourier transform is only suitable for the analysis of stationary signals, and it is difficult to reflect the overall and local characteristics of time and frequency domains at the same time

The simple wavelet packet energy cannot reflect the complexity of the energy in the vibration signal, and is not sensitive to the early failure of rolling bearings

In the case of multi-feature extraction, there is also the problem of redundant features

Images