Fault diagnosis method for high-voltage circuit breakers based on physics-informed transfer learning

IntroductionHigh-voltage circuit breakers are core control and protection equipment in power systems, and their operational status directly affects device stability and power grid security. Improving the accuracy of their fault detection is a key demand for the operation and maintenance of power equipment.MethodsThis study proposes a fault detection method for high-voltage circuit breakers based on multi-source information and motion analysis. First, a 1-dimensional recurrent neural network (1DRNN) is used to analyze voiceprint and current signals to extract feature data related to the mechanical state of the operating mechanism. Second, a physics-informed transfer learning network model consisting of a Common Feature Learning Network (CFLN) and a Mechanical Feature Learning Network (MFLN) is constructed to explore shared features between multi-source signals and mechanical parameters and extract specific features of individual mechanical parameters in a targeted manner. Meanwhile, a multi-head attention mechanism is integrated to enhance the model’s ability to capture key features, and a physics-based loss function is designed to improve the physical consistency of the model during mechanical parameter identification.ResultsExperimental verification shows that the proposed method achieves a fault diagnosis accuracy of over 93% for high-voltage circuit breakers, and the model can still maintain high diagnostic stability and detection accuracy under noise interference conditions.DiscussionThrough the design of deep fusion of multi-source signals and embedding of physical information, this method makes up for the information defects of single-signal diagnosis, solves the problem of lack of physical consistency in data-driven models, and improves the environmental adaptability of fault diagnosis models, providing a practical technical solution for the intelligent fault diagnosis of high-voltage circuit breakers.