Rolling Bearing Fault Diagnosis Method Based on Bayesian Residual Transformation-Singular Value Decomposition-Gaussian Mixture Hidden Markov Model Framework

Abstract

The invention belongs to the field of fault diagnosis of mechanical equipment. The invention discloses a fault diagnosis method for frame rolling bearings based on Bayesian residual transformation, singular value decomposition, and Gaussian mixture hidden Markov model. The acceleration sensor on the bearing cover under test amplifies the signal and transmits it to the multi-channel data acquisition analyzer, and the analyzer sends the collected signal to the PC; secondly, the collected signal is processed as follows: the first step, through Bayeux The Sri Lankan residual transform decomposes the signal, and denoises and reconstructs the residual signal. The reconstructed signal contains relatively clear signal features; the second step uses singular value decomposition to extract the singular value vector of the reconstructed signal obtained in the previous step, and improves The stability of fault characteristics; the third step is to classify bearing faults according to the hidden Markov model of Gaussian mixture. The invention has the advantage of accurately diagnosing rolling bearing faults.

Description

technical field [0001] The invention belongs to the technical field of mechanical equipment fault diagnosis, in particular to a rolling bearing fault diagnosis method based on Bayesian residual transformation-singular value decomposition-Gaussian mixed hidden Markov model framework. Background technique [0002] Bearings are widely used in modern mechanical equipment and are an indispensable part of rotating machinery. Its main function is to support the mechanical rotating body, reduce the friction coefficient during its movement, and ensure its rotation accuracy. Due to harsh working conditions such as high speed and heavy load, the life of the bearing becomes shorter, which causes various mechanical failures. These failures can lead to damage to machines, ranging from financial loss to catastrophic accidents. [0003] As a new discipline, mechanical equipment fault diagnosis technology was first developed in the United States in the 1960s. my country's fault diagnosis ...

AI Technical Summary

Problems solved by technology

However, these diagnostic methods have inevitable defects.

For example, the transform scale adopted by the wavelet transform method is small, when there is strong energy interference in the low frequency band, the effective value of the feature quantity will decrease, resulting in unsatisfactory diagnosis results

Empirical mode decomposition is difficult to avoid the endpoint effect, that is, the upper and lower envelopes diverge at both ends of the data sequence, and this divergence will gradually inward as the operation proceeds, thus affecting the entire data sequence

[0006] Wigner-Ville distribution has good time-frequency aggregation, but for multi-component signals, according to the convolution theorem, there will be cross terms, resulting in "false signals", and defects in the diagnosis process

Images