A Fault Diagnosis Method for High-speed Train Bogie Based on Multi-source Data Fusion

The diagnostic ability of the traditional bogie fault diagnosis method with single dimension status data is limited. In order to grasp the degradation state of components accurately by using more abundant information, reduce the maintenance workload and ensure the safety of operation, this paper proposes the data-driven multidimensional data fusion fault diagnosis model. In the model, the fault of bogie can be detected and classified by feature extraction, feature selection, feature fusion, anomaly detection and fault classification. In the fault diagnosis of temperature, vibration and other states, this method can give the rate of abnormal points and anomaly score of each compartment successfully, to locate the anomalous compartment. The list of anomaly compartments given by this method has been verified by experts.