Bearing Fault Diagnosis Based on Wide Deep Convolutional Neural Network and Long Short Term Memory

Abstract

Mechanical fault can cause economic loss of different degrees, even casualties. Timely fault diagnosis is an essential condition for ensuring safe production in modern industry. With the growth of intelligent manufacturing, more and more attention is paid to fault diagnosis methods that are based on deep learning. However, the diagnostic accuracy of existing diagnostic methods has still to be improved. Therefore, a fault diagnosis method called WDCNN-LSTM is proposed by combining Wide First-layer Deep Convolutional Neural Network with Long and Short Term Memory. Feature information is extracted adaptively from one-dimensional original vibration signals by Convolutional Neural Network. The extracted features are further extracted by Long and Short Term Memory, so that the fault feature information can be fully obtained. Experiments are performed on CWRU datasets to verify our proposed method. By analyzing the experimental results, we find that the average accuracy of the proposed WDCNN-LSTM model is 99.65%.