Abstract: It is essential to obtain the switch sequence of vacuum on-load tap changers in order to evaluate their mechanical condition. A multi-body dynamic model of the diverter switch of a vacuum on-load tap changer is established, and the movements (e.g. operating time, duration, displacement and angle, etc.) of each contact are extracted firstly. The influence of stored energy of springs on the operation time, duration and contacts’ movement are calculated, and the criteria for the duration are obtained. A platform to acquire vibration characteristics of tap changers is established, and the contacts’ actions are extracted based on the Teager energy operator and wavelet time-frequency analysis. The relationship between vibration signals and contact movements during switching is established both from the simulation and the experiments. Switch sequences of 24 vacuum tap changers of converter transformers are detected through the vibration signals. The research results show that with a preload of 200 N on spring, the recovery force and energy storage are only 67.2% and 57.6% of those at 500 N, respectively. The switching time exceeds 210 ms, which is much longer than 105 ms under normal conditions. The peak value of the Teager energy operator of the vibration signal and the occurrence of clusters of time-frequency vibration signals are related to the contacts’ movement and can be used to extract the switching sequence. The filed test indicates that the average switching time of 24 tap changers is about 105 ms, which meets the technical specifications. The proposed method in this paper is meaningful to understand the dynamic movement of the diverter switch and improve the vibration-based mechanical condition monitoring of the on-load tap changers.