Vacuum circuit breaker is widely used in the medium voltage field for its strong arc extinguishing ability, non-pollution and compact structure. The traditional off-line vacuum testing methods does not meet the needs of intelligent development of power system. In this study, a method based on laser-induced breakdown spectroscopy (LIBS) technique combined with random forest (RF) algorithm is proposed for on-line vacuum degree monitoring on vacuum circuit breakers. In the experiment, the LIBS platform was used to collect the spectral data from the target material in the vacuum chamber under different pressure conditions ranging from 10-3Pa to 105Pa. A random forest model is built for calculating barometric pressure levels from spectral data. The spectral lines of 8 elements from target materials and environmental gas were selected as data sets and then input into RF model for training. Different data preprocessing methods (normalization, mean centering, 1-st derivative) were used to increase the computational power of the model, correlation coefficient (R2) and root mean square error (RMSE) were evaluation indexes. The effects of the number and depth of decision trees and the number of labels on the accuracy of barometric pressure estimation were investigated. The results show that the accuracy of this method is more than 99% and can reach the accuracy of traditional vacuum degree monitoring technology. This paper demonstrates that the LIBS technique combined with RF algorithms is a new method that can be used for on-line vacuum degree monitoring of vacuum circuit breakers.

vacuum breaker; vacuum degree monitoring; LIBS; random forest