Fault Diagnosis of Power Equipment Based on Infrared Image Analysis

As the need to discover abnormal temperature rise and overheating breakdown of power equipment in substations is urgent, infrared inspection is important for its quick and accurate characteristics. To improve the accuracy of infrared inspection diagnosis, this paper proposes an operational diagnosis system based on comprehensive analysis of infrared images of power equipment. Firstly, the image is preprocessed by histogram equalization, Sobel operator, Canny operator, median filtering, etc. Then, the feature points of the infrared image are extracted by SIFT algorithm, and the extracted feature points are K-means clustered. Find the type of feature points that belong to the capacitive device after clustering, and then perform SIFT feature extraction on the processed image, and eliminate the feature points that are not electrical devices according to the found feature points, thereby completing the target —— Identification of the electrical device. Finally, the target recognition results of the 3 pretreatment methods and the original without pretreatment are compared to find the best effect, that is, the pre-processing algorithm with the highest recognition accuracy.