Rolling bearing fault diagnosis utilizing pearson’s correlation coefficient optimizing variational mode decomposition based deep learning model

Abstract

For unstable characteristics of helicopter critical system rolling bearing signal and low accuracy of fault diagnosis classification, a new method about fault diagnosis is proposed. Firstly, proposing the new method about variational mode decomposition (VMD) based on the Pearson's correlation coefficient (PCC) adaptive decomposition. The proposed method can effectively overcome the problem that traditional VMD methods requiring prior artificial setting of input parameters may lead to poor signal decomposition; Then combine the convolutional neural networks (CNN) strong feature extraction capability and support vector machine (SVM) good feature classification capability, The CNN_SVM model was constructed to fault diagnosis the rolling bearing signal decomposed by VMD. The effectiveness of this method is verified on bearing dataset from the laboratory. The results show that this paper proposed method has good diagnostic results and high recognition accuracy, and deep learning model has strong model stability and robust ability.