基于人工智能技术的输电线路红外图像持续管过热检测

Transmission Line Infrared Image Continuous Tube Overheating Detection Based on Artificial Intelligence Technology

本文指出随着现代化社会的发展,人工智能技术不断应用于电网设备的运行维护当中。以往在输电线路红外图像中发现持续管过热的主要方法是人工诊断,效率低,主观性强,效果达不到预期。本文采用人工智能图像分析技术对红外图像中输电线路持续管过热故障进行自动检测。首先检测输电线路红外图像中的颜色条,自动建立颜色与温度间的映射关系;然后采用Yolov5算法自动检测图像中的持续管,再根据前述温度映射算法得到持续管每个像素的温度,当发现温度异常时,自动报警;最后在包含了100个点的数据集上对算法进行了测试。实验表明该方法能较为准确地检测图像中持续管的温度异常情况并主动报警,具有较强的实用价值。

This paper points out that with the development of modern society,artificial intelligence technology is continuously applied to the operation and maintenance of power grid equipment.In the past,in the infrared image of transmission lines,it was found that the main manual diagnosis experience of continuous tube overheating was low efficiency,subjective and the effect was not up to expectations.In this paper,artificial intelligence image analysis technology is used to automatically detect the overheating fault of continuous tube of transmission line in infrared image.Firstly,It detects the color bar in the infrared image of the transmission line,automatically establishes the mapping relationship between color and temperature;secondly,it uses the Yolov5 algorithm to automatically detect the continuous tube in the image,and then obtains the temperature of each pixel of the continuous tube according to the aforementioned temperature mapping algorithm,and automatically alarms when the temperature is found to be abnormal;finally,the algorithm is tested on a dataset containing 100 points.Experiments show that the proposed method can accurately detect the temperature abnormality of the continuous tube in the image and actively alarm,which has strong practical value.