Substation switching device identification method based on deep learning

Abstract

In order to realize the real-time and efficient detection of substation switching device status, and to solve some existing problems of substation inspection system, such as the limited degree of automation and the need for human intervention to solve the problem. This paper proposes a status recognition method of substation switching device based on deep learning. The method obtains image data by the optical camera of an inspection robot, and uses the deep learning technology to analyze and detect the data. In this method, firstly, the image data of substation switching device is selected and marked as the data set of model training, then the Yolov3 target detection network is used to build an automatic status recognition model of substation switch device. Experimental results show that this method can automatically identify the substation switching device status, and the recognition accuracy reached 97%.