An Approach of Intelligent Compound Fault Diagnosis of Rolling Bearing based on MWT and CNN

Abstract

An approach to intelligent compound fault diagnosis of rolling bearing using multi- wavelet transform（ MWT） and convolution neural network（ CNN） was proposed. According to this approach,the vibration signals of rolling bearing are analyzed by using MWT of removed post processing,and the corresponding multi- wavelet coefficient branches are obtained. Then,all the multi- wavelet coefficient branches are used to form feature maps,and a multiple CNN classifiers is developed to identify the compound fault of rolling bearing.The tests for the proposed method are accomplished based on artificial bearing fault data sets,and the method is optimized,the vibration signals are analyzed by using MWT,the obtained multi- wavelet coefficient matrix is used to form feature map,and CNN is developed to make a comparative experimental study. The experimental results indicates that this method could effectively identify the compound fault of rolling bearing,and the improved method could effectively improve the fault recognition rate and reduce the training cost.

Keywords

• Rolling bearing
• Intelligent compound fault diagnosis
• Multi-wavelet transform（MWT）
• Convolution neural network（CNN）