基于DCGAN的绝缘子图像生成方法  

Image Generation Method of Insulator Based on DCGAN

作　　者：杨彦利[1] 矫红岩 YANG Yan-li;JIAO Hong-yan(School of Electronics and Information Engineering,Tiangong University,Tianjin 300387,China)

机构地区：[1]天津工业大学电子与信息工程学院,天津300387

出　　处：《自动化与仪表》2021年第1期5-9,共5页Automation & Instrumentation

基　　金：天津市应用基础与前沿技术研究计划项目(19JCYBJC16400)。

摘　　要：绝缘子是输电线路的重要组成部分,利用深度学习等人工智能方法可以对绝缘子进行智能检测,但需要大容量的样本库,而绝缘子故障图像比较稀缺,可借助计算机生成故障绝缘子图像。针对绝缘子图像的特点对DCGAN网络模型进行改进,实现故障绝缘子图像的计算机生成。改进后DCGAN模型网络生成的绝缘子图像多样性好、稳定性高。对比其他的图像生成模型,结果表明,所生成的绝缘子图像质量更好,为生成大量故障绝缘子图像提供了技术途径。Insulators are important part of transmission line.Artificial intelligence methods such as deep learning can be used to detect insulators intelligently,but a large sample library is needed,and insulator fault images are scarce.Therefore,the image of faulty insulator can be generated by computer.The DCGAN network model is used to generate the fault insulator images and it is improved according to the characteristics of the fault insulator image.The insulator images generated by the improved DCGAN model network has good diversity and high stability.The insulator images generated by the improved DCGAN model network has good diversity and high stability.Compared with other image generation models,the results show that the generated insulator images are of better quality,which provides a technical approach for generating a large number of faulty insulator images.

关 键 词：绝缘子故障图像 图像生成 生成对抗网络 DCGAN 深度学习 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TM216[自动化与计算机技术—控制科学与工程]