Abstract: The opening and closing operation process of high-voltage circuit breakers includes actions such as detent release, spring energy release, dynamic and static contact collision and other actions. Mechanical vibration signals, as the performance of energy transfer, contain rich operating status information. In this paper, based on the timing of the key action nodes of the circuit breaker, a circuit breaker operating state identification method is proposed to optimize the attractor characteristics of the vibration signal in the action timing. Firstly, the high-speed camera is used to capture the moving image of the main shaft crank arm during the operation process of the circuit breaker, and the target tracking algorithm is used to obtain the timing index of the key motion node of the moving contact. Then, the operation process is divided according to the timing index, and by reconstructing the phase space of the vibration signal at different stages, phase space attractor geometry and attribute features are extracted. Finally, the intelligent classification algorithm is used for fault identification. The experimental results verify that this method can be adopted to not only greatly save time, but also effectively improve the accuracy of classification identification. The method of this paper has broad application prospects in circuit breaker condition monitoring and live tests.