Abstract: The circuit breaker operates by energizing the trip/close coil to drive the core, which then triggers the mechanical components to control the opening and closing actions.The existing research on the fault diagnosis of circuit breakers mainly focuses on extracting the current characteristics of the trip/close coil, but the relationship between the core movement and the coil current is not well understood. This paper uses the Lucas-Kanade optical flow method to analyze high-speed image sequences frame by frame, identifying the core movement trajectory in coordination with the coil current. The results show that the core motion precedes the current peak, and its speed, acceleration, and displacement are related to the coil current waveform and its cumulative duration. Moreover, a fault classification framework is proposed, combining the coil current and core movement features, using clustering to characterize fault severity and a random forest to assess the sensitivity of movement-current features to faults. Experiments simulating coil voltage anomalies, core jamming, and loose coil screws validate that the integrated current-core feature analysis enables more accurate fault diagnosis.