Abstract:

For the prediction of the remaining useful life of the 5.3 A control and protection switch, it is found that with the performance degradation of the switch, the contact bounce phenomenon becomes more and more obvious, and the reaction characteristics of the switch also changes.Through the change of contact bounce,the sound signal is related to the spring stiffness and other characteristics of the switch.Through the change of the reaction force characteristics,the electrical life of the switch is related to the spring stiffness and other characteristics of the switch,so as to link the sound signal with the electric life.The remaining service life of the switch can be predicted by the change trend of the sound signal. The experimental results show that the recognition accuracy of voiceprint based on the electrical life of 0,100,120 000 and 160 000 times is 96.40%,which has guiding significance for the replacement of the switch. Moreover,due to the non-contact characteristics of voiceprint monitoring, it has reference significance for the maintenance of high voltage equipment.

Key words: switch, life prediction, voiceprint monitoring, machine learning