Research On Intelligent Recognition System Of Air Switch And Connector In Substation

In order to adapt to the development of economic and power grid integration mode,this paper studies the intelligent identification system of air switches and connectors in substation.Air switches and connectors status verification is one of the monthly tasks of substation operation and maintenance personnel.Through this work,air switches and connectors status errors can be found,and the operation risk of power grid can be reduced.In this paper,the traditional manual work mode is replaced by intelligent,electronic and paperless recognition system.It can effectively improve work efficiency,reduce labor time,and lay a good foundation for the network operation and maintenance of power equipment.It is of great significance.The intelligent identification system of air switches and connectors in substation studied in this paper mainly includes three modules: air switches and connectors positioning module,air switches and connectors status recognition module based on convolution neural network,and air switches and connectors name character recognition module.Firstly,a unified image localization method is proposed for air switches and connectors.The image localization function is realized by edge and template matching.Secondly,in order to identify the splitting and closing state of the air switches or the throwing and withdrawing state of the connectors,a convolutional neural network is designed,which can actively learn the features.Convolutional neural network consists of input layer,multiple convolution and pooling layers,full connection layer and output layer.Thirdly,Baidu cloud character recognition method is applied to identify the names of air switches and connectors.Finally,the operation results of these three modules are summarized and integrated.It realizes the function of text output from input image to air switches and connectors status and corresponding name.The experiment proves that the recognition effect of this system is excellent.The application effect of air switch and connector positioning module is good.The positioning accuracy of air switch is 0.97087.And the positioning accuracy of the connector is 0.99471.The air switches and connectors state recognition module based on convolution neural network has excellent application effect.The accuracy of air switches status recognition is as high as 0.995.The accuracy of state recognition of the connectors is as high as 0.99734.Theapplication effect of air switches and connectors name character recognition module is excellent.The accuracy of air switches name recognition is 0.87118.And the accuracy of text recognition is 0.95058.When the location of a single air switch or connector device is accurate,the state recognition is accurate and the accuracy of character recognition exceeds 80%,the device is defined as the overall recognition accuracy device.According to statistics,the overall recognition accuracy of a single air switch is 0.83738,and that of a single connecting piece is 0.90212.This system can accurately realize the identification function of air switch and connector,and the effect is excellent.