Rolling Bearing Fault Diagnosis Based on Convolutional Neural Network and Multi-sensor Information Fusion

Abstract

Condition monitoring and fault diagnosis of bearings are of great research significance for the safety and reliability of the rotating mechanical system. However, the mechanical failure caused by rolling bearings is difficult to be accurately identified by traditional methods based on physical mechanism and signal analysis which are also time-consuming. Therefore, a bearing fault diagnosis model based on multi-sensor information fusion and one-dimensional convolution neural network (1D-CNN) was proposed. The bearing fault vibration feature from an aeroengine was extracted and analyzed by using the 1D-CNN. The waveform signals collected by different sensors were input and the final classification results through convolution and pool operation were output, which abandons the traditional tedious steps based on signal analysis fault diagnosis. The experimental results showed that the accuracy of the model can reach 100% by using four accelerometers. Compared with support vector machine (SVM) and feedforward neural network (FNN), the accuracy of the method is improved by 36.92 % and 18.9% respectively, which provides a feasible method for aeroengine bearing fault diagnosis.