Bearing Fault Diagnosis Method Based on GMM and Coupled Hidden Markov Model

Abstract

Aiming at the weak characteristic of bearing fault signal, a method of coupled hidden Markov fault diagnosis based on Gaussian mixture model is proposed, which realizes the identification of common faults of bearing. Firstly, the bearing signal is collected and the running state information of the bearing under different states is obtained. Secondly, the whitening algorithm is used to whiten the signal of bearing under different states, and reduce the influence on the weak fault signal of bearing. Then the step algorithm is used to process the signal, characterizing the running state of the bearing. Finally, the corresponding Coupled Hidden Markov model is trained by using the representation information in different state, and the test data is used to verify it. The experimental results show that the Coupled Hidden Markov model based on the Gaussian mixture model can effectively identify the bearing fault states.