电气设备红外图像故障辨识方法研究  

Research on Infrared Image Fault Identification Method for Electrical Equipment

作　　者：田庆 杨天驰 Tian Qing;Yang Tianchi(Jiangsu Rugao Secondary Professional School,Nantong Jiangsu 226500,China;State Grid Jiangsu Electric Power Co.,Ltd.,Rugao Power Supply Branch,Nantong Jiangsu 226500,China)

机构地区：[1]江苏省如皋中等专业学校,江苏南通226500 [2]国网江苏省电力有限公司如皋市供电分公司,江苏南通226500

出　　处：《信息与电脑》2025年第6期8-10,共3页Information & Computer

摘　　要：为了提升红外图像的分辨率和电气设备故障检测的精确度,减少人工消耗,文章提出了一种基于超分辨率图像重建与在线硬样本挖掘的电气设备红外图像故障辨识方法。首先,利用对抗生成网络对低分辨率红外图像进行4倍超分辨率放大;其次,利用在线硬样本挖掘方法优化检测算法训练过程,强化困难样本训练;最后,利用目标检测算法对电气设备红外图像进行故障辨识。实验结果表明,文章提出的算法能够有效提高故障检测的辨识成功率,相较于主流目标检测算法,在精度上有显著提升。To enhance the resolution of infrared images,improve the accuracy of electrical equipment fault detection,and reduce manual labor,the paper proposes a method for fault identification of electrical equipment infrared images based on super-resolution image reconstruction and online hard sample mining.Firstly,the low-resolution infrared images are magnified by 4 times using an adversarial generative network.Secondly,the online hard sample mining method is utilized to optimize the training process of the detection algorithm and strengthen the training of difficult samples.Finally,the target detection algorithm is used to identify faults in electrical equipment infrared images.Experimental results show that the proposed algorithm can effectively improve the success rate of fault detection and has higher accuracy compared with mainstream target detection algorithms.

关 键 词：图像故障辨识 超分辨率重建 在线硬样本挖掘 目标检测算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]