The Fault Diagnosis Model for Variable Speed Rolling Bearings Based on GCRA-FMD, COT, and Deep Convolutional Neural Networks

The high-precision fault diagnosis of rolling bearings under variable speed conditions faces significant challenges. Therefore, it is necessary to effectively extract the features of variable speed fault signals to achieve accurate classification. To address this, this paper combines the Greater Cane Rat Algorithm (GCRA), Feature Mode Decomposition (FMD), Computation Order Tracking (COT), and Deep Convolutional Neural Networks (DCNNs) are combined to propose a fault diagnosis model. First, the important parameters of FMD are selected using GCRA to enhance the ability to reconstruct signals and eliminate noise interference. Second, the COT method is employed to extract the fault features of variable speed rolling bearings. Finally, the extracted features are input into the DCNNs model for classification. The experimental data were provided by the University of Ottawa, and the results of the analysis show that this method achieves excellent classification accuracy.