Abstract: The operating mechanism of power circuit breakers is prone to jamming faults during the opening and closing process due to mechanical friction, which affects the safe operation of the power system. Therefore, a fault diagnosis method for power circuit breaker operating mechanism jamming based on multi-scale convolutional neural network is proposed. Real time monitoring of the vibration signal of the circuit breaker operating mechanism using piezoelectric acceleration sensors, and introducing wavelet thresholding to denoise the signal, resulting in a pure signal. Using the wavelet packet transform method to perform frequency band decomposition on the denoised vibration signal, obtain the energy distribution of the signal components, and construct the modulation signal using the Hilbert transform method and mother wavelet function. The signal is reconstructed and decomposed to obtain the variational mode components of the vibration signal. Constructing a fault diagnosis model using multi-scale convolutional neural networks, determining the type of circuit breaker jamming fault through probability distribution calculation, and completing fault diagnosis. The experimental results show that the AUC value of the jam fault detection result of this method is 1.00, indicating high diagnostic accuracy.