Measurement of Vacuum Interrupter Shell Thickness Based on Terahertz Time-Domain Spectroscopy and BP Neural Network

The vacuum interrupter serves as the core component of a vacuum circuit breaker, and the thickness of its shell is a crucial indicator for evaluating its quality. In order to achieve accurate measurement of the vacuum interrupter’s thickness, a BP neural network model is constructed. This model allows for obtaining the thickness measurement results directly through input and output relationship mapping, without the need to calculate the refractive index. This approach effectively avoids errors in refractive index caused by uneven microstructure of ceramics. Experimental results demonstrate that after BP neural network training, prediction accuracy is further improved. When compared with five other machine learning models, the BP neural network exhibits superior performance in thickness prediction, achieving an accuracy rate of 99.9975%. Therefore, this study proposes a method for measuring terahertz vacuum arc extinguishing outer shell thickness based on BP neural network technology. This method can effectively enhance measurement accuracy and precision, thereby holding significant importance for ensuring safe operation and maintenance of vacuum circuit breakers.