Enabling Safety Guarantees of Urban Rail Transit: A New Digital Twin Framework for Data-Model Driven Track Condition Diagnosis

Urban rail transit has become a significant part of transportation. With the growth of its service period, various defects have begun to appear in the track, which threatens the safety of the vehicle operation. Some research has been implemented to detect the track condition based on visual instruments, which is cost-consuming. Thus, a method based on digital twin is proposed to diagnose the track condition. At first, in the virtual environment, a vehicle-track coupling dynamics model considering axle-box bearings is constructed as the digital twin model of the vehicle-track system existing in the physical environment. Then, the models of rail corrugation and rail welds are coupled into the track random irregularity for simulating the axle-box vibration characteristics in the case of track defects. The axle-box accelerations simulated under different track conditions are considered as templates after de-noising and enveloping. These templates are transferred to the physical environment from the virtual environment later. Finally, when a vehicle passes through a section of track, the track condition is diagnosed through comparing the processed axle-box signals collected from the vehicle and templates. The experimental results present that the proposed method can accurately diagnose different track conditions.