VMD–RP–CSRN Based Fault Diagnosis Method for Rolling Bearings

In response to the problems of low accuracy and poor noise immunity traditional fault diagnosis method for rolling bearing due complex variable operating conditions bearings large interference during signal acquisition, a model based on VMD–RP–CSRN is proposed. Firstly, initial feature extraction carried out by variational modal decomposition (VMD), which then converted into two-dimensional image with features recurrent plot (RP) coding, images are input channel split residual network (CSRN) classification. order verify proposed faults under working conditions, experiments selection parameters in CSRN were conducted dataset Jiangnan University, compared other commonly used methods. The combines VMD RP retain original maximum extent stress hidden signal. operation realizes selecting main three-channel image, makes more participate model. experimental results demonstrate that at least 1.2% better than comparison method, has immunity. addition, capability different data set sizes speed show generalization performance capability.