主要电力设备故障图像特征及识别方法研究改教程.docxVIP

摘要 摘要内容 伴随着我国电网规模的日益加大,各类变电设备的运作状态是促使其安全高效运行的最为主要的因素之一。对于各类变电设备的在线状态监测系统的推广越来越发普及。研究基于图像特征的电力设备自动故障识别具有重要意义。 本文对各类主要电力设备，研究各类变电设备故障识别分类及相应故障的图像特征，以及基于红外与紫外图像特征的故障识别方法。对于紫外放电成像技术图像的处理与特征提取，本文从紫外成像技术的基本原理出发，在讲解紫外放电图片特性的基础上，对紫外放电图像使用灰度化预处理，以及应用中值滤波等方法对图像进行降噪。并通过canny算子边缘检测计算紫外光斑面积判断是否发生放电故障。针对红外故障图像，本文在红外成像原理的基础上，对红外图像进行超像素分割及HSV空间颜色提取，对应用卷积神经网络对红外故障图像故障区域检测进行理论上的研究。 关键词： 红外成像 紫外成像 图像处理  ABSTRACT With the increasing scale of Chinas power grid, the operation of various types of substation equipment is one of the most important factors to promote the safe and efficient operation. The popularization of the on-line condition monitoring system for all kinds of transformer equipment is becoming more and more popular. Research on image feature based automatic fault recognition of power equipment is of great significance. In this paper, various types of main power equipment, the study of various types of substation equipment fault identification and classification of image features, as well as infrared and ultraviolet image features based on fault identification method. For ultraviolet discharge imaging technique to image processing and feature extraction, this paper from the basic principle of UV imaging technology of on the explanation of the ultraviolet discharge picture characteristics based and discharge on the UV image using grayscale preprocessing and application of median filtering method of image in noise reduction. And through the Canny operator edge detection to determine whether the area of the UV spot to determine whether the discharge fault. Aiming at the fault infrared image, this paper on the basis of the principle of infrared imaging, the infrared image were super pixel segmentation and HSV color space extraction, the convolutional neural network is applied to fault section detection of infrared fault image of theoretical research. KEY WORDS:  目录  第一章 1.1课题背景 伴随着我国国民经济水平的前进与发展