While temperature monitoring is widely used on MU train bogie bearings, it is not sufficiently effective for localized faults like spalling, especially for early fault monitoring. Since vibration information can fully reflect the working state of the bearing, a vibration method of fault monitoring was used. A dynamic model of a rolling bearing with fault was established based on common causes of faults on MU train bogie rolling bearings. The flexural stiffness of the axletree, the gap of the bearing and the nonlinear contact force between races were taken into consideration, and models of rolling bearing with faults on the outer race, inner race and rolling elements were included. The models were simulated with the Runge Kutta numerical method. The rolling bearings with different faults and faults under different degrees were tested. The results were identical, and the maximum error of both the calculation and simulation results was less than 5%, proving the correctness of the dynamic model.