Fault Diagnosis of Shaft Misalignment and Crack in Rotor System Based on MI-CNN

Abstract

Focusing on the difficulty in distinguishing the shaft misalignment and crack in a rotor system, a fault diagnosis method based on multi-input convolutional neural network (MI-CNN) is proposed in this paper. The time-domain vibration signals are directly taken as the input of a one-dimensional convolutional neural network, which are collected on the test bench in the conditions of health, shaft misalignment, crack and misalignment-crack coupling of the rotor system. Kernels of different sizes are adopted to extract the signal features of diverse dimensions at different input ends to fully use the information of the raw vibration signals, and then the extracted features from each input end are fused adaptively. Finally, the classification of shaft misalignment and crack of the rotor system is completed by softmax function. The results show that the intelligent diagnosis of shaft misalignment and crack in the rotor system can be realized effectively by the proposed method, and eventually the recognition rate reaches 99.42%, which has better accuracy and stability compared with other intelligent algorithms. The study achievements can provide a basis for intelligent fault diagnosis of rotating machinery.