Abstract: Aiming at the non-stationary characteristics of vibration signal shown by bearing of urban rail transit vehicle during speed change process, a method of order analysis based on synchrosqueezing transforms (SST) is proposed. Firstly, the bearing vibration signal in strong noise environment is processed by SST time-frequency analysis, where the instantaneous frequency in the vibration signal is estimated, and the instantaneous frequency is extracted. Then, the obtained instantaneous frequency is used to calculate the phase discrimination time scale, and the time-domain signal is resampled to obtain the quasi steady-state angular domain signal. Finally, the bearing fault diagnosis in the process of vehicle speed change is completed. The experiment results show that this method can effectively extract the rotational frequency in the bearing vibration signal, and realize the order analysis based on the instantaneous frequency estimation, which allows successful bearing fault diagnosis in the process of speed change.