A Fault Diagnosis Method of Rolling Bearing Based on Convolutional Neural Network

Rolling bearing is a vital part of mechanical components, and its working environment is complex, which is prone to failure. Once a fault occurs, the rotating machinery will be shut down or even scrapped, which directly affects the economic benefits and personal safety. Therefore, this paper presents a rolling bearing fault diagnosis method based on convolution neural network (CNN). Through a series of comparative experiments, the best parameters of the model, such as learning rate, BatchSize, Dropout layer position and Dropout rate, are determined. In this article, after Fourier transform of primitive vibration signal of the bearing, the frequency domain signal is transformed into a two-dimensional condition as input of CNN to diagnose and identify inner circle fault, outer circle fault, cage fault and mixed fault. Finally, the experimental data set of rolling bearings published by Xi 'an Jiaotong University is used, the diagnostic accuracy rate is as high as 99.89%, which shows that the model can effectively diagnose rolling bearing faults. In addition, the proposed method is compared with Gates recurrent neural (GRU) and long-short term memory neural networks (LSTM), and the results show the advantages of the proposed method.