A Complex Multi-Working-Condition Bearing Fault Diagnosis Model Based on Sparse Representation Classification

This article proposes a new method for bearing fault diagnosis based on sparse representation classification to address the challenges of fault identification under complex working conditions with different degrees of damage. The core of this method lies in directly using the original vibration signal to construct an overcomplete dictionary without the need for signal denoising or manual feature extraction in advance, thus avoiding the information loss and subjective bias introduced by denoising and feature engineering in traditional methods. Firstly, all training samples are used as a dictionary to sequentially solve for sparse coefficients for each test sample. Secondly, the corresponding parts of each category in the sparse coefficients are filtered out. Then, the category error is calculated based on the sparse coefficients corresponding to each category. Finally, the fault classification of bearings is carried out by comparing the category errors. The experimental results show that this method can maintain high diagnostic accuracy and robustness in complex scenarios with various working conditions and damage levels, verifying its effectiveness and universality for bearing fault diagnosis.