Infrared Image Recognition Technology Based on Visual Processing and Deep Learning

Infrared image recognition in substation is always a difficult problem. In order to solve the problem of recognition of knife gate, insulator and other components in infrared image, the target location technology of deep learning is proposed to realize the detection and recognition of typical components in infrared image, and the multi-target detection algorithm YOLO is selected to locate and identify the defects. Firstly, the image preprocessing technology is used to process the collected image, so as to filter the interference of background and other factors on the equipment identification. Then, the infrared image is detected by the YOLO target detection model based on multi feature fusion, so as to locate the position of inspection equipment in the infrared image. Then, the type of equipment is identified by the trained equipment classification model. Finally, the algorithm is tested with a large number of pictures in the substation scene.

I. INTRODUCTION

In recent years, the research on the detection and fault diagnosis of electrical equipment [1] using infrared thermal imaging technology is booming at home and abroad. This technology is also widely used in the existing substation inspection robot system. Infrared thermal imaging technology is a non-contact, passive measurement technology. It can detect and diagnose a large number of internal and external defects of power equipment, especially many faults that are not easy or cannot be detected by conventional test methods. However, there are still some deficiencies in the current fault diagnosis of electrical equipment in substations with infrared thermal image sensors, which are mainly shown as follows: (1)

Infrared thermal imaging technology reflects the information of temperature field, which will cause serious deformation of object shape. This technology can not locate and recognize the infrared image of substation equipment effectively, which leads to the inaccuracy of mapping the corresponding relationship between temperature field information and electrical equipment in the image;

(2)

In order to identify the substation equipment by infrared image intelligent recognition technology, we have accumulated a large number of infrared image data resources, but the traditional machine learning algorithm can not effectively use these data resources, and can not effectively diagnose the fault of substation equipment;

(3)

Through infrared image analysis, the temperature information of an area is often obtained. The equipment location affects the accuracy of data judgment. Therefore, an accurate method of target location and recognition of electrical equipment is needed to reduce the interference of background information.