For the problem of abnormal opening and closing states of high-voltage isolation switches in substations due to factors such as mechanical wear and electrical quantity changes，an atom search optimization（ASO）algorithm is proposed to identify the abnormal opening and closing status of high-voltage disconnectors in substations.Infrared and visible light cameras are used to capture the status images of high-voltage isolation switches，and the mapping relationship between image feature points is established through bilateral filtering and image registration.The joint weighted average method is used to achieve decision level fusion of images.The optimal segmentation threshold is determined by combining the gradient size and attribute vector of the centroid pixel neighborhood points of the image，and the high-voltage isolation switch feature area is extracted accordingly.Support vector machine algorithm is adopted to construct an abnormal state recognition model，and ASO algorithm is introduced to obtain model parameters，and to optimize model recognition performance，and identify the opening and closing abnormal states of the isolation switch by inputting the pixel values of the isolation switch feature area.Experiment results show that under the application of the studied method，the false positive rate of the obtained recognition results is less than 2%，and the recognition accuracy is relatively high.

substation  /  high voltage isolation switch  /  abnormal state of splitting and merging  /  image decision level fusion  /  atom search optimization algorithm  /  support vector machine algorithm