基于无人机影像的电网绝缘子自爆识别

电网因其在电能传输方面的关键性作用,在我国民生项目建设领域一直扮演着至关重要的角色。电网杆塔上的绝缘子一旦发生自爆(也称“缺陷”),绝缘子会自动剥落,输电线路就会产生安全隐患,严重时会降低输电线路的运行寿命,甚至会引发供电中断,发生大范围的停电事故,造成巨大的财产损失。目前,主流的巡检方法为人工巡检,该方法不仅耗时耗力,而且也存在一定主观出错率,已不适用于目前电路巡检的实际情况。本设计采用YOLO V5网络模型,对无人机航拍影像中绝缘子串及绝缘子自爆进行自动识别。首先通过平移、翻转、裁剪等,对航拍绝缘子影像数据集进行数据增广,并对增广后的数据集在LabelImg中进行标注,然后利用YOLO V5网络模型对绝缘子串及绝缘子自爆进行识别,最后采用PyQt5框架在PyCharm中设计了绝缘子自爆识别的系统界面,对模型进行调用,实现了绝缘子串及绝缘子自爆识别。本设计采用从网络上下载、国家电网提供、数据增广所得到的500张无人机航拍影像作为数据集,对所得数据集进行人工标注,再使用YOLO V5网络模型进行训练和测试,结果表明YOLO V5网络模型对绝缘子串具有较高的识别精度,最高识别精度为90.2%,对绝缘子自爆的最高检测精度为80.8%。这说明了YOLO V5网络模型在绝缘子串识别方面有较好的表现,但是由于训练集中绝缘子自爆的样本影像数量有限,所以该网络模型对绝缘子的自爆识别存在一定局限性,本实验能够部分代替人力实现电网绝缘子智能巡检,提高了检测效率。

Identification of Power Grid Insulator Self Explosion Based on UAV Image

Because of its key role in power transmission,power grid has always played a vital role in the field of livelihood project construction in China.Once the insulator on the power grid tower self explodes(also known as"defect"),the insulator will automatically peel off,and the transmission line will have potential safety hazards.In serious cases,it will reduce the operation life of the transmission line,even lead to power interruption,largescale power failure and huge property losses.At present,the mainstream inspection method is manual inspection.This method is not only time-consuming and labor-consuming,but also has a certain subjective error rate,which is not suitable for the actual situation of current circuit inspection.The design adopts the Yolo V5 network model to automatically identify the insulator string and insulator self explosion in the aerial image of UAV.Firstly,the aerial insulator image data set is expanded through translation,turning and cutting,and the expanded data set is marked in labelimg.Then,the insulator string and insulator self explosion are identified by Yolo V5 network model.Finally,the system interface of insulator self explosion identification is designed in pychart using pyqt5 framework to call the model,The identification of insulator string and insulator self explosion is realized.In this design,500 UAV aerial images downloaded from the network,provided by the State Grid and expanded data are used as the data set.The obtained data set is manually labeled,and then trained and tested with the Yolo V5 network model.The results show that the Yolo V5 network model has high recognition accuracy for the isolated sub string,and the maximum recognition accuracy is 90.2%,The highest detection accuracy of insulator self explosion is 80.8%.This shows that the Yolo V5 network model has a good performance in insulator string identification.However,due to the limited number of sample images of insulator self explosion in the training center,the network model has some limitations in insulator self explosion identification.This experiment can partially replace manpower to realize intelligent inspection of power grid insulators and improve the detection efficiency.