A TFG-CNN Fault Diagnosis Method for Rolling Bearing

Abstract

It is difficult to obtain enough data to train a robust diagnosis model for different rolling bearing faults, and the existing intelligent bearing fault diagnosis algorithms have insufficient generalization ability. Therefore, a rolling bearing fault detector based on the time–frequency graph and convolution neural network (TFG-CNN) is introduced to improve the generalization performance of the fault diagnosis algorithm as much as possible under the condition of considering the diagnosis accuracy and sample size. The specific implementation method is to use Fast Fourier transform (FFT) to transform the vibration data of rolling bearing into a two-dimensional network graph, and then use CNN to classify them. Finally, the performance of the proposed method is analyzed by using the rolling bearing fault datasets of Case Western Reserve University, and analysis results show that the proposed method can simultaneously diagnose the fault location and severity of rolling bearing, and has good cross-domain diagnosis ability and anti-noise performance.