基于图像融合的输电线路覆冰重量估计研究

Research on Transmission Line Ice Weight EstimationUsing Image Fusion

摘要:

输电线路覆冰是电力系统运行中常见且危害性极大的气象灾害，其附加荷载会导致导线弧垂增加、绝缘性能下降甚至杆塔损毁；基于此，对输电线路覆冰重量估计问题进行了研究。针对无人机巡检图像因拍摄角度差异导致的几何畸变以及低能见度条件下图像质量下降的情况，采用透视变换进行视角校正，并利用改进小波变换融合可见光与红外图像，以提升图像清晰度、对比度和细节表现。结合SURF特征点检测、FLANN匹配与加权平均策略，实现了输电线路全景图像的无缝拼接。在拼接图像中提取导线弧垂参数，并结合抛物线力学模型与导线设计参数，反演计算线路覆冰重量。经实验测试，该方法在多日实测数据中与拉力传感器测值的平均误差小于6%，在准确性、鲁棒性和计算效率方面均优于传统方法，能够满足电力系统覆冰监测与防灾减灾的工程应用需求。

Abstract:

Icing on transmission lines is a common and highly hazardous meteorological disaster in power systems. The additional load increases conductor sag, reduces insulation performance, and may even cause tower damage. To address this, research on icing weight estimation was conducted. Perspective transformation was applied to correct geometric distortion in UAV inspection images, and an improved wavelet transform was used to fuse visible and infrared images, enhancing clarity, contrast, and detail. SURF feature detection, FLANN matching, and weighted averaging were integrated to achieve seamless panoramic stitching of transmission line images. Conductor sag parameters were extracted from the stitched images, and icing weight was inversely calculated using a parabolic mechanical model combined with conductor design parameters. Experimental results show that the method achieves an average error of less than 6% compared with tension sensor measurements, demonstrating superior accuracy, robustness, and computational efficiency over traditional methods, and meeting engineering requirements for icing monitoring and disaster prevention in power systems.