基于Canny算子和神经网络复合绝缘子憎水性研究

发布时间：2018-03-28 12:12

本文选题：复合绝缘子　切入点：憎水性　出处：《华北水利水电大学》2017年硕士论文

【摘要】：随着我国社会的发展,电力系统建设的越来越完善,绝缘子作为电力系统的重要设备,也在不断地发展进步,由原来的陶瓷变为现在的复合绝缘子。但随着复合绝缘子运行时间的不断地增长和在环境等外力影响下,其憎水性水平会呈现出不同差别的降低,容易发生事故,目前还没有一套完善的复合绝缘子憎水性等级自动检测系统。本文分析了几种现有的复合绝缘子憎水性检测的方法,并进行了总结,在憎水性图像数字化处理的基础上,对Canny算子和BP神经网络在憎水性分级领域上的作用进行研究。(1)提出了RGB颜色分量最小值图像灰度化算法,将彩色复合绝缘子憎水性图像转化为灰度图像,增强了水珠(水迹)与背景之间灰度值的差值,使得水珠(水迹)的边缘更加清晰。(2)设计出一种改进的多参数h自适应平滑滤波算法处理灰度图像,该算法能减少因复合绝缘子憎水性图像不同而造成原自适应滤波算法的不通用性,使得算法的应用范围更加广泛。(3)提出了基于梯度图像的Canny算法来切割灰度图像,此算法使得到的二值图像边缘比较连续,并用形态学方法修正二值图像,提取二值图像的特征值,通过BP神经网络识别出复合绝缘子憎水性等级。本文是在Visual Studio 2010软件平台上进行复合绝缘子憎水性的仿真分析。通过仿真实验对各个算法进行研究,进而改进它们,使它们更好的满足复合绝缘子憎水性分级的需求。
[Abstract]:With the development of our society and the improvement of power system construction, insulators, as an important equipment of power system, are also developing and making progress. From the original ceramics to the composite insulators, but with the continuous increase of the running time of the composite insulators and the influence of external forces such as the environment, the hydrophobicity level of the composite insulators will show a different decrease, which is prone to accidents. There is not a perfect automatic testing system for hydrophobicity grade of composite insulators. This paper analyzes several existing methods of hydrophobicity detection of composite insulators, and summarizes them, based on the digital processing of hydrophobic images, The role of Canny operator and BP neural network in hydrophobicity classification is studied. (1) A grayscale algorithm for minimum image of RGB color component is proposed, and the hydrophobic image of color composite insulator is transformed into grayscale image. The difference between the grayscale value of water droplets (water trace) and background is enhanced, and the edge of water droplets (water trace) is clearer. A modified multi-parameter h adaptive smoothing algorithm is designed to deal with grayscale images. This algorithm can reduce the inapplicability of the original adaptive filtering algorithm caused by the different hydrophobic images of composite insulators, and make the application of the algorithm more extensive. (3) A gradient image based Canny algorithm is proposed to cut grayscale images. The algorithm makes the edge of the binary image more continuous, and uses morphological method to modify the binary image and extract the eigenvalue of the binary image. The hydrophobicity grade of composite insulators is identified by BP neural network. The simulation analysis of hydrophobicity of composite insulators is carried out on the Visual Studio 2010 software platform. So that they can better meet the hydrophobic classification of composite insulators.
【学位授予单位】：华北水利水电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TM216

【参考文献】