Research on insulator fault inspection algorithm based on deep learning
【摘要】 针对输电线路中出现的绝缘子损坏问题,构建基于深度学习的目标控制模型,采用Faster R-CNN算法对无人机巡检中拍摄图像(简称巡检图像)进行分析和识别,以有效提高输电线路的可靠性和无人机的巡检效率。对样本集的500张图像进行识别与检测应用发现:采用无人机进行输电线路巡检并基于深度学习算法进行绝缘子故障图像识别时,缺陷绝缘子的识别准确率为92.0%,正常绝缘子的识别准确率可达97.8%。
【Abstract】 Aiming at the problem of insulator damage in transmission line,a target control model based on deep learning is constructed,and the Faster R-CNN algorithm is used to analyze and identify the images taken in UAV patrol inspection(inspection image for short),so as to effectively improve the reliability of transmission line and the efficiency of UAV patrol inspection.Through the identification and detection of 500 images of the sample set,it is found that when UAV is used for transmission line inspection and the Faster-CNN algorithm based on deep learning is used to analyze the image fault of insulator devices,the recognition accuracy of defective insulator is 92.0%,and that of normal insulator is 97.8%.
【Key words】 application of four rotor autonomous vehicle； deep learning； insulator； fault detect；
- 【文献出处】 中原工学院学报 ,Journal of Zhongyuan University of Technology , 编辑部邮箱 ,2021年05期