摘要
为确保输电线路稳定运行,研究基于分级树状级联分类算法的输电线路日夜一体可视化方法,提升输电线路日夜一体可视化监测效果。通过800万像素日夜摄像头实时采集输电线路图像;利用Hough变换算法在输电线路图像内提取输电线路;利用软级联加性核化支持向量机算法构建输电线路监测简单分类器;结合郊狼优化算法,优化简单分类器,建立输电线路日夜一体监测的简单分类器;利用决策树级联简单分类器得到分级树状级联分类器,以提取的输电线路为数据样本,输入分级树状级联分类器,输出输电线路日夜一体监测结果;在Google Earth软件内导入监测结果,实现输电线路日夜一体监测结果的可视化显示。试验结果表明:该方法可有效实时采集白天与黑夜的输电线路图像,并有效提取输电线路,能够有效实现输电线路日夜一体可视化监测;对于不同输电线路运行情况,该方法可视化监测的可信度较高,效果较优。
To ensure the stable operation of transmission lines,a day and night integrated visualization method for transmission lines based on hierarchical tree cascade classification algorithm is studied to improve the effectiveness of day and night integrated visualization monitoring of transmission lines.First,real-time transmission line images is collected through an 8 million pixels day and night camera.Using the Hough transform algorithm,transmission lines is extracted from the transmission line image.Then a simple classifier for transmission line monitoring is constructed using the soft cascade additive kernel support vector machine algorithm.Combining the Coyote optimization algorithm,a simple classifier is optimized and a simple classifier for integrated day and night monitoring of transmission lines is established.Second,a simple classifier uses a decision tree to obtain a hierarchical tree like cascade classifier.Using the extracted transmission lines as data samples,input a hierarchical tree cascade classifier and output the integrated monitoring results of the transmission lines day and night.Finally,monitoring results are imported into Google Earth software to achieve visual display of transmission line day and night monitoring results.The experiment results prove that this method can effectively collect real-time images of transmission lines during the day and night,and effectively extract transmission lines,which can effectively achieve day and night integrated visual monitoring of transmission lines.For different transmission line operation situations,this method has high credibility in visual monitoring,indicating that the visual monitoring effect is better.
作者
刘礼良
王小军
赵雁荣
袁登堂
李家豪
LIU Liliang;WANG Xiaojun;ZHAO Yanrong;YUAN Dengtang;LI Jiahao(State Grid Kezhou Power Supply Company,Artux 845350,China)
出处
《甘肃科学学报》
2025年第6期146-152,共7页
Journal of Gansu Sciences
关键词
分级树状
级联分类算法
输电线路
日夜一体
可视化监测
支持向量机
Hierarchical tree
Cascade classification algorithm
Transmission line
Day and night
Visual monitoring
Support vector machine
