基于改进生成对抗网络的变电站接地网故障太赫兹层析成像  

Terahertz Tomography of Substation Grounding Grid Fault based on Improved Generative Adversarial Networks

作　　者：薛炳磊 兰峰 张盛晰 张春辉 李小帆 XUE Binlei;LAN Feng;ZHANG Shengxi;ZHANG Chunhui;LI Xiaofan(Economic&Technology Research Institute,State Grid Shandong Electric Power Company,Jinan 25000,China;Zhejiang Langsong Intelligent Power Equipment Co.Ltd.,Jiaxing 314500,China)

机构地区：[1]国网山东省电力公司经济技术研究院,山东济南250000 [2]浙江朗松智能电力设备有限公司,浙江嘉兴314500

出　　处：《红外技术》2025年第2期257-263,共7页Infrared Technology

基　　金：国网山东省电力公司科技项目资助(20625210004)。

摘　　要：由于城市融合型变电站接地网更换困难,一旦接地网发生故障会对变电站以及周边建筑的安全产生较大影响。针对变电站接地网的故障检测问题,本文研究了基于改进生成对抗网络的太赫兹层析成像检测技术。针对太赫兹图像分辨率低、噪声高的特性,提出了基于改进生成对抗网络的太赫兹层析图像检测技术,首先利用改进生成对抗网络提高图像的细节处理能力;其次利用跳层连接的方法,有效提高了图像的上下文信息参考能力,从而使得图像的细节性更强,对比度更高;最后将本文提出的方法与其他传统处理方法对比可知,本文所提出的算法得到的图像清晰度更高,更适合用于城市融合型变电站接地网的故障检测。Due to the difficulty in replacing the grounding grid of urban integrated substations,any grounding grid failure can have a significant impact on the safety of the substation and surrounding buildings.This paper studies the terahertz tomography detection technology based on an improved generative adversarial network for fault detection of substation grounding grids.According to the characteristics of low resolution and high noise of terahertz image,uses terahertz tomographic image detection technology based on improved generative adversarial network.Firstly,the improved generation countermeasure network to improve the detail processing ability of the image;Secondly,uses the layer hopping connection method to effectively improve the context information reference ability of the image,so as to make the image more detailed and more specific.Finally,the method proposed in this paper is compared with other traditional processing methods,this method is more suitable for substation grounding grid fault detection.

关 键 词：对抗网络 变电站接地网 太赫兹层析成像 故障检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]