基于场景约束的高压设备红外识别技术研究与应用

发布时间：2018-03-29 08:50

本文选题：红外在线监测　切入点：高压设备　出处：《安徽大学》2015年硕士论文

【摘要】：变电站场站内高压设备正常运行与否,事关人们日常生活和国家经济发展。红外检测技术能够无损的检测高压设备运行状态,被广泛应用在变电站故障检测中。不同类型高压设备故障的致热原理不同,其外在温度场表现也不同,因而不同类型高压设备的热故障诊断参数不一样,所以在对高压设备进行热故障诊断前需要确定待诊断设备类型以获取正确的诊断参数。论文关注变电站红外检测技术中的设备识别并进行相应热故障自动诊断问题,包括全景约束的在线红外热像仪视觉内高压设备的识别与局部场景约束的手持红外热像仪视觉内高压设备的识别,以实现智能的在线红外监测系统,提高监测效率。首先,研究了红外热像的相关应用技术,包括工业红外热像检测、红外弱小目标检测与红外热像目标识别。深入讨论了高压设备的热故障原因及分类,并研究了热故障诊断方法,提出了基于设备识别的高压设备热故障自动诊断方法,利用识别结果可获取相应的诊断参数,进而实现热故障的自动判断。其次,研究了全景约束的高压设备在线定位与识别方法。研究了在线红外热像仪的红外全景视觉构建方法,即构建高压设备全景温度场,具体包括温度场热像采集、温度场热像特征提取、基于特征点的热像配准、温度场融合。为解决全景视觉内任意高压设备的在线识别,提出了基于全景温度场的高压设备在线自动定位与识别方法。实验表明,在手动标注全景视觉内高压设备的基础上,该方法能够实现在线红外热像仪视觉内任意高压设备的识别,并能给出全景温度场中的定位信息,具有深刻的理论意义与现实意义。接着,研究了局部场景约束的高压设备分类识别技术。针对手持红外热像仪视觉内高压设备的识别问题,研究了现有的模式识别分类方法。鉴于变电站高压设备红外热像特性,提出了一种基于形态学操作及区域生长法的高压设备分割提取方法；研究了用以红外分类识别的梯度方向直方图特征(HOG特征)描述子以及常用的红外分类器,包括KNN与SVM。实验结果表明HOG特征与SVM分类器能有效分类识别局部场景约束的高压设备,识别精度较高。最后,介绍了高压设备识别在变电站在线红外监测系统中的应用背景,构建智能在线红外监测系统。利用软件工程设计思想,结合变电站实际环境,进行系统的需求分析、架构设计以及数据库设计,并开发出一套变电站红外智能在线监测系统。系统中有效验证了全景约束以及局部场景约束的高压设备自动识别与相应热故障自动诊断方法。
[Abstract]:The substation station high voltage equipment normal operation or not, is related to people's daily life and the development of the national economy. The infrared detection technology can detect high voltage equipment running state lossless, is widely used in substation fault detection. Different types of high voltage equipment fault caused by thermal performance in different principle, its temperature field is different, so the thermal fault diagnosis the parameters of different types of high voltage equipment is not the same, so in high-voltage equipment for thermal fault diagnosis before need to determine types of diagnostic equipment to obtain the correct diagnosis parameters. The corresponding thermal fault paper focuses on infrared detection technology in substation equipment identification and automatic diagnosis, identify constraints include a panoramic infrared online visual instrument internal pressure the equipment and the local scene constraints of handheld infrared camera vision in high voltage equipment, in order to realize the online monitoring of the intelligent infrared System, improve the monitoring efficiency. Firstly, the related application technology of infrared thermography, including infrared industrial detection, infrared dim target detection and infrared imaging target recognition are discussed. The thermal fault causes and classification of high voltage equipment, and to study the thermal fault diagnosis method, proposes the automatic diagnosis methods of high voltage equipment thermal equipment fault based on the identification of the identification results can obtain diagnostic parameters, and automatically determine the thermal fault. Secondly, study the panoramic constraint high-voltage equipment online positioning and recognition method of infrared panoramic vision. Construction method of online thermal infrared imager, namely the construction of high pressure equipment panoramic temperature field, including temperature field image acquisition the temperature field, thermal image features extraction, image registration based on feature points, the temperature field of fusion. In order to solve the panoramic vision arbitrary high pressure equipment in line identification, put forward The method of high voltage equipment online automatic positioning and identification of panorama based on temperature field. The results show that, based on manual annotation of panoramic vision in high voltage equipment, this method can realize on-line recognition of infrared thermal imager in any vision of high pressure equipment, and can locate the information given in the panoramic temperature field, has the profound theory significance and reality the significance of research. Then, the high voltage equipment identification and classification technology of local scene constraints. For the recognition problem of handheld infrared camera vision in high voltage equipment, the classification method of pattern recognition. In view of the existing substation high voltage equipment infrared characteristics, puts forward a kind of high voltage equipment and morphological operation based on region growing segmentation in order to study method; gradient direction recognition infrared histogram features (HOG feature) and infrared descriptor classifier, including KNN and SVM The experimental results show that the high voltage equipment. HOG features and SVM classifier can effectively classify the local scene constraints, high recognition accuracy. Finally, introduces the application background of high recognition in the substation equipment infrared online monitoring system, the construction of intelligent online infrared monitoring system. Using software engineering design, combined with the actual substation environment, analysis system demand, architecture design and database design, and developed a set of online monitoring system of substation infrared intelligent system. The effective validation of panoramic constraints and local scene constraints pressure equipment of the automatic identification and the corresponding thermal automatic fault diagnosis method.

【学位授予单位】：安徽大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM506;TN219

【相似文献】