A Fault Diagnosis Method Based on Temperature and Vibration Characteristics for High-speed Train Axle Box Bearing

High-speed train running speed unceasing enhancement, the vehicle running status monitoring and security put forward higher request, so the temperature characteristic of axle box bearing and axle box bearing fault diagnosis method is crucial to provide certain reference and reference for the engineering practice. For the fault diagnosis of high-speed train axle box bearings, a deep learning network-based method, considering the features of both temperature and vibration, is proposed in this paper. A two-channel CNN is constructed based on 2D-CNN and 1D-CNN, in which 2D-CNN takes infrared image as input, and 1D-CNN takes vibration signal as input. Convolution and pooling are carried out respectively, and stretching is taken as feature vector. Then, splicing is carried out in the aggregation layer, and classification is carried out through the fully connected network layer. This method can realize the effective fusion of one-dimensional vibration features and two-dimensional temperature field features, and improve the classification accuracy. The performance of the proposed model is analyzed by high-speed train bearing test. The results show that in this paper, deep learning network is used for bearing fault diagnosis without artificial feature extraction, and the average accuracy of training set is 100%, and that of verification set is 98.02%; Based on infrared thermal imaging system for lubrication of high sensitivity compared with vibration system, but the infrared thermal imaging system for mechanical fault vibration system, due to the interaction characteristics of machine learning algorithms.

Keywords: Bearing fault diagnosis, Infrared image, Temperature vibration fusion, Deep learning, Temperature characteristic.

Download PDF