Abstract:In order to solve the problems of high cost and poor stability of isolating switch state detection methods based on various sensors in substations in the past, this article explores two types of isolating switch state detection algorithms based on image recognition. Compared to traditional isolating switch state recognition algorithms based on image similarity, deep learning based object detection algorithms have higher accuracy in identifying isolating switch states and can effectively detect isolating switch states in substations. After annotating 483 images containing various isolating switch states, this article used Yolov5"s pre trained model for training. The trained model was tested on a test set containing 80 different isolating switch states, and the results showed a comprehensive accuracy rate of 89.31% and a comprehensive recall rate of 98.32%. The deep learning based isolating switch recognition algorithm proposed in this article can effectively identify the status of substation isolating switches, with high recognition accuracy and simple deployment, playing an important role in ensuring the safe and stable operation of substations.