基于深度学习的高压输电线路防振锤检测  

Damper Detection of High Voltage Transmission Line Based on Deep Learning

作　　者：贾雁飞 陈广大 杨淼 邢砾云 赵立权 李帅洋 JIA Yanfei;CHEN Guangda;YANG Miao;XING Liyun;ZHAO Liquan;LI Shuaiyang(Department of Electrical and Information Engineering,Beihua University,Jilin Jilin 132013,China;School of Electrical Engineering,Northeast Electric Power University,Jilin Jilin 132012,China)

机构地区：[1]北华大学电气与信息工程学院,吉林吉林132013 [2]东北电力大学电气工程学院,吉林吉林132012

出　　处：《机床与液压》2022年第13期21-25,共5页Machine Tool & Hydraulics

基　　金：国家自然科学基金青年科学基金项目(61901007);吉林省科技发展计划项目(YDZJ202101ZYTS172);吉林省教育厅“十三五”科学技术项目(JJKH20210042KJ,JJKH20220054KJ);北华大学青年科技创新团队(202016003)。

摘　　要：为识别距离较近的防振锤,提出基于改进YOLOv4的防振锤自动检测方法。YOLOv4方法采用具有固定阈值的非极大值抑制方法选取检测框,较低的阈值会导致丢失高度重叠的目标,而较高的阈值则会导致更多的误检。为此,提出动态非极大值抑制方法,并将其应用于YOLOv4目标检测。该方法根据目标周围检测框的统计特性确定出动态阈值,提高边界框选择的准确性,降低高度重叠防振锤检测中的错检和漏检概率。为进一步提高防振锤检测精度,采用分段线性函数作为激活函数,克服YOLOv4算法中Leaky ReLU函数对负值处理不理想且函数曲线不平滑的问题,增强了模型的非线性表达能力。结果表明:基于改进YOLOv4的防振锤目标检测方法能够很好地检测出重叠的防振锤,且检测精度更高。In order to identify the closer damper,a damper automatic detection method based on improved YOLOv4 was proposed.In YOLOv4 method,non maximum suppression method with fixed threshold was used to select the detection frame,lower threshold would lead to the loss of highly overlapped targets,but higher threshold would lead to more false detection.Therefore,a dynamic non maximum suppression method was proposed and was used to YOLOv4 target detection.By using the method,the dynamic threshold was determined according to the statistical characteristics of the detection frame around the target,the accuracy of boundary box selection was improved,and the probability of false detection and missing detection in the detection of high overlap damper was reduced.In order to further improve the detection accuracy of damper,the piecewise linear function was used as the activation function to overcome the problem that Leaky ReLU function in YOLOv4 algorithm did not deal with negative value very well and the function curve was not smooth,the nonlinear expression ability of the model was enhanced.The results show that by using the improved YOLOv4 damper target detection method,the overlapping dampers can be detected well,and it has higher detection accuracy.

关 键 词：防振锤检测 深度学习 YOLOv4算法 非极大值抑制 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]