高压输电线路视频检测及故障诊断方法研究
文献类型：学位论文
| 作者 | 何思远
|
| 学位类别 | 博士 |
| 答辩日期 | 2015-05-27 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 中国科学院沈阳自动化研究所 |
| 导师 | 唐延东 |
| 关键词 | 旋翼无人机 巡检机器人 目标识别 输电线路检测 故障诊断 |
| 其他题名 | Power Transmission Line Detection and Diagnosis Based on Computer Vision |
| 学位专业 | 模式识别与智能系统 |
| 中文摘要 | 随着经济持续快速发展对电力需求的不断增加，高压输电线路的分布越来越广，迅速增长的输电线路给其巡检维护工作带来了巨大的压力。目前，高压输电线路巡检方式主要是采用人工巡检方式，即依靠巡检工作人员徒步行走或者借助直升飞机沿输电线路进行巡检，并以纸质介质方式记录巡检情况。这种巡检方式存在劳动强度大，工作效率低，可靠性差等缺点，而且还存在一定的危险性，已经不能适应当前高压输电线路巡检工作的需要。因此，亟需研究出一种高效、先进、科学、安全的现代化高压输电线路巡检方法。我国在“十二五”期间提出大力发展智能电网的计划，并将智能化巡检列为其中重点发展的项目。利用旋翼无人机和巡检机器人进行输电线路智能化巡检是近几年国内外研究的热点之一，并且随着计算机图像处理技术、模式识别技术以及机器人技术的发展，将其在输电线路巡检中应用已经成为一种必然趋势。 在上述背景下我们开展了本文的研究工作。本文以国家电网公司科技项目“无人飞行器巡检技术研究”为依托，针对输电线路设备检测及故障诊断方法进行研究，构建了输电线路巡检系统，针对输电线、输电塔、绝缘子、防震锤、相位标志牌的检测问题，以及输电线温度异常、输电线断股及异物悬挂、绝缘子自爆及局部放电的故障诊断问题，提出了基于视频图像的相关自动识别算法。本文的研究工作为高压输电线路的巡检提供了解决方案，对于提高其自动化水平，降低工作强度具有重要的应用价值。本文的具体研究工作如下： （1）介绍了高压输电线路视频检测及故障诊断方法的研究背景及意义，分析了旋翼无人机和巡检机器人输电线路巡检的现状，描述了其工作过程，分析了目前输电线路视频检测及故障诊断中已经取得的成果和主要存在的问题，给出了构建的输电线路巡检系统所能巡检的具体故障。 （2）提出了视频图像中输电线和输电塔的检测方法。通过分析航拍视频图像中输电线和输电塔的特征，提出了基于Hessian矩阵的高压输电线检测方法和基于Line Segment Detector的输电塔检测方法。通过旋翼无人机实际拍摄的视频，验证了所提方法的有效性。该方法也为无人机巡检时的导航，以及输电线和输电塔附近设备的识别与检测奠定了基础。 （3）提出了基于图像的绝缘子、防震锤和相位标志牌的检测方法。通过分析航拍图像中绝缘子、防震锤和相位标志牌的特征，提出了基于不变矩特征和级联AdaBoost分类器的绝缘子检测方法，基于小波特征和Affinity Propagation聚类的防震锤检测方法，以及基于区域一致性和目标显著性的相位标志牌检测方法。这些提出的方法可以有效识别航拍图像中绝缘子、防震锤和相位标志牌等目标，为后续的故障检测奠定了基础。 （4）提出了输电线和绝缘子的故障诊断方法。通过分析拍摄的红外图像和紫外图像，提出了输电线异常发热和绝缘子局部放电的故障诊断方法。针对输电线断股及异物悬挂和绝缘子自爆，提出了基于矩阵迹的输电线断股及异物悬挂故障诊断方法，以及基于距离计算的绝缘子自爆故障诊断方法。通过实际拍摄的视频图像，验证了这些方法的有效性。 （5）构建了输电线路巡检系统软件。通过分析巡检工作的实际需求，给出了巡检系统软件的具体功能，并通过实验测试结果验证了软件系统的有效性。 |
| 索取号 | TM726.1/H32/2015 |
| 英文摘要 | With the rapid development of economy, more electric power is needed to meet the practical demand. The distribution of high-voltage power transmission line is more and more widely, and it brings great pressure to power transmission line inspection. Nowadays, manual inspection is still the main method of power transmission line inspection, that is, the inspector inspects the power transmission line on walk or helicopter, and keeps the record on the paper. However, this way has the disadvantages of high labor strength, low efficiency and poor reliability, it also has certain risks for inspector, and it cannot meet the demand of power transmission line inspection. So it is necessary to develop a modern method of power transmission line inspection. The Chinese government puts forward the develop plan of smart grid during the Twelfth Five Year Plan period, and the smart inspection of power transmission line is the key construction program. In recent years, the use of UAV (unmanned aerial vehicle) and inspection robot for power transmission line inspection is a hot technology, and with the development of image processing and pattern recognition, it will become an inevitable trend to use them to inspect the power transmission line. This dissertation is carried out based on above objectives. The research of this dissertation is supported by the State Grid Corporation of China item unmanned aerial vehicle inspection technology. It mainly focuses on power transmission line detection and diagnosis, designing the inspection scheme, and constructing the inspection software system. Based on the video/images taken by UAV and other robots, its main tasks include the automatic detection of power transmission line, power transmission tower, insulator, vibration damper and power phase sign, and the automatic fault diagnosis of power transmission line temperature rising, broken strand, extra matter and insulator spontaneous explosion and partial discharge. The research of this dissertation provides the solution for power transmission line inspection, improves its automatic level, reduces the labor intensity. The main contents of this dissertation are as follows. 1. We introduce the background and significance of power transmission line detection and diagnosis, analyze the current situation and working process of UAV and inspection robot, achievements and problems of power transmission line detection and diagnosis based on video/images, and introduce the faults that the constructed inspection system can diagnose. 2. We propose the detection methods of power transmission line and tower. According to the features of power transmission line and tower from the aerial images, the methods of power transmission line detection based on Hessian matrix and power transmission tower detection based on LSD (line segment detector) are put forward. The results of experiments demonstrate that the proposed methods have better validity, and it lays the foundation for UAV navigation and the equipments detection and fault diagnosis which are near the power transmission line and tower. 3. We propose the detection methods of insulator, vibration damper and power phase sign. According to the features of insulator, vibration damper and power phase sign from the aerial images, the methods of insulator detection based on moments invariant features and cascade AdaBoost classifier, vibration damper detection based on wavelet eigenvalue and AP (affinity propagation) clustering algorithm and power phase sign detection based on region conformance and saliency are put forward. The results of experiments demonstrate that the proposed methods can recognize the insulator, vibration damper and power phase sign, and it lays the foundation for fault diagnosis. 4. We propose the fault diagnosis methods of power transmission line and insulator. The methods include power transmission line temperature rising diagnosis based on infrared image, power transmission line broken strand and extra matter diagnosis based on matrix trace, insulator spontaneous explosion diagnosis based on distance and insulator partial discharge diagnosis based on ultraviolet image. The results of experiments demonstrate that the proposed methods have better validity. 5. We construct the power transmission line detection system software. According to practical requirement of power transmission line inspection, we design the functions of system software, and realize it. After the testing, the software runs well, and has better validity. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 页码 | 106页 |
| 源URL | [http://ir.sia.ac.cn/handle/173321/16789]  |
| 专题 | 沈阳自动化研究所_机器人学研究室 |
| 推荐引用方式 GB/T 7714 | 何思远. 高压输电线路视频检测及故障诊断方法研究[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2015. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。