In view of the limit effect and prone to environmental interference of existing monitoring method in the opening and closing state monitoring of GIS disconnector

a kind of opening and closing state monitoring method of disconnector

which is based on the voiceprint characteristic analysis

is proposed in this paper. Firstly

for the angle and linear type disconnectors

the actual opening and closing as well as actual environmental noise audio are collected

the related audio is subjected to the noise addition processing so to build the openi

ng and closing dataset of the disconnector which is more closer to the actual environment. Then

the

MFCC

features as voiceprint features from the opening and closing audio of the disconnector are extracted. Finally

the neural network model is trained by the extracted voiceprint features to construct the opening and closing state monitoring algorithm of disconnector. The experimental results show that the proposed opening and closing state monitoring method of disconnector

compared to the traditional voiceprint monitoring method

can effectively improve the recognition

and can be accurately recognized under different noise conditions. The final results of system validation experiment show that the average recognition rate of opening and closing state of the two types of disconnectors

which is performed by the proposed method in this paper

can be up to 96.5%.