Fault Diagnosis Method of Rolling Bearing Based on Enhanced Mel's Linear Frequency Cepstral Coefficient

Abstract

The invention provides a fault diagnosis method for a rolling bearing based on an enhanced Mel linear frequency cepstral coefficient, belonging to the technical field of rolling bearing fault diagnosis. The method includes: acquiring a time domain vibration signal of the rolling bearing; using a preset Mel linear frequency filter The feature extraction is performed on the time-domain vibration signal to obtain a two-dimensional feature map of the enhanced Mel linear frequency cepstral coefficients; the final diagnosis result is obtained according to the two-dimensional feature map of the enhanced Mel linear frequency cepstral coefficients and a preset Transformer model; the present invention Accurate and rapid diagnosis of rolling bearing faults can be achieved under the condition of data distribution offset.

Description

technical field [0001] The invention relates to the technical field of fault diagnosis of rolling bearings, in particular to a fault diagnosis method of rolling bearings based on enhanced Mel linear frequency cepstral coefficients. Background technique [0002] The statements in this section merely provide background related to the present disclosure and do not necessarily constitute prior art. [0003] With the vigorous development of modern industry, more and more machinery and equipment are deployed to industrial sites to perform production tasks, and rotating machinery is a typical representative. However, due to the long-term high-load operation and the harsh working environment, the rotating machinery often breaks down and stops production. The components that fail most frequently in rotating machinery are the bearings. Therefore, it is of great practical significance to realize timely diagnosis of bearing faults so that maintenance can be carried out in advance to a...

AI Technical Summary

Problems solved by technology

However, CNN has a limited receptive field during feature extraction. When it wants to extract features in a large range of data, it needs to continuously stack convolutional layers, which makes the model parameters large and difficult to train.

In addition, the current fault diagnosis method based on deep learning often has poor diagnostic effect when applied in a new environment, because the distribution of the data in the new environment is not exactly the same as that of the training set, and this deviation of the distribution makes the parameters learned by the model no longer Perfectly applicable; in addition, the weak fault information contained in the actual collected bearing vibration signals is often covered by noise, and it is difficult to extract obvious fault features with existing signal processing methods

Images