基于ViBe检测的输电线路防外力破坏预警模型

ViBe Detection-based Early Warning Model for Prevention of External Damage of Transmission Lines

根据近年电网的数据统计,机械车辆碰线、漂浮物缠绕和人为破坏等外力因素已成为输电线路跳闸事故的主要原因。为了有效应对此类故障隐患,设计构建BP神经网络输电线路防外力破坏预警模型,对安装于线路终端实时获取的监测图像进行预警等级分类。考虑输电线路通道内危险物通常为移动目标,本文对常用目标识别方法支持向量机(SVM)分类器引入ViBe移动侦测算法确定目标区域,提高模型的图像识别速度以满足线路预警的实时性要求。为解决传统危险预警判断依据单一导致准确率不高的问题,提出将危险物大小、危险物类型、目标最大作业高度和输电线路电压等级共同作为BP网络模型输入,通过引入多个影响因子提高模型可靠性。收集各地区电力公司发布的输电线路外力破坏案例数据作为训练样本,利用tensorflow建立输电线路防外力破坏BP神经网络预警模型,通过现场图片试验验证,方法预警准确,具有很高的应用价值。

According to statistical data about power grids in recent years,external factors such as short by mechanical vehicles,twining by floating objects and man-made damage have become the main cause of line tripping on transmission lines.In order to deal with these hidden accidents,an early warning model using BP neural network was designed and built up for prevention of external damages on the transmission line,and warning grade classification was completed for supervisory graphs obtained at line terminals on a real-time base.Considering that dangerous objects were usually moving targets in the channels of transmission lines,ViBe motion detection algorithm was introduced into the support vector machine(SVM)classifier of the common target recognition method to determine the object area and raise the speed of image identification of the model,so as to meet the real-time requirement of the early warning of the line.To solve the problem of low accuracy rate attributable to unitary judgment criterion for traditional early warning,it was proposed that the size and type of the dangerous object,maximal working altitude of the object and voltage grade of the transmission line should all be used as input to the BP network model,and several factors of influence should be introduced to improve model reliability.Data issued by electric power companies in various areas about external damage cases on transmission lines were collected as training examples.Tensorflow was used to establish the BP neural network early warning model for prevention of damages of transmission lines by external force.Pictures taken on the site verified that the proposed approach could achieve accurate early warning and had high application value.