The mechanical state of the operating mechanism of the high-voltage circuit breaker is a critical factor for the safe and stable operation of both the equipment and the power grid. In the context of the new power system, utilizing artificial intelligence to identify key mechanical parameters accurately is of great significance for fault diagnosis and tracing of the operating mechanism. This paper proposes a novel transfer network based on the one-dimensional convolutional neural network and an improved Transformer, which realizes mechanical parameter identification for the operating mechanism. The network consists of a common feature learning network and a specific feature learning network. The common feature learning network extracts shared knowledge between the mechanical state of the operating mechanism and composite feature factors, while the specific feature learning network learns knowledge specific to the individual feature factors. By integrating the common and the specific feature networks, the global knowledge is retained and the ability to extract local features is enhanced. The numerical experiment results show that the proposed network achieves higher accuracy compared to other methods, with the relative error of the feature factor identification being less than 4%.