Basing on Vibration Signal and Improved Neural Network Research on Contact State Recognition Model of SF6 High Voltage Circuit Breaker

High voltage circuit breakers are prone to contact wear during operation, resulting in poor contact. As a result, the circuit breaker cannot conduct current normally, which adversely affects the safety and stability of power system operation. Nowadays, high-voltage circuit breaker equipment usually dectect the contact resistance before they are put into working, without a stable and accurate way to dectect breakers' contact station without damaging them. Aiming at a SF
6
high-voltage circuit breaker, this paper builds a circuit breaker vibration signal acquisition platform, completes vibration signal acquisition under different currents and different contact states, and obtains 30 vibration signals under different working conditions. The signal pretreatment method is designed, and a circuit breaker contact state recognition model based on vibration signals and improved neural network is proposed to achieve good recognition effect. It can provide reference for the contact monitoring and trouble dectection in the future.