基于红外测温数据的水电站设备缺陷检测方法  

Defect Detection Method for Hydropower Station Equipment Based on Infrared Temperature Measurement Data

作　　者：周鹍 郭俊鑫 罗杰 李云红[2] 李丽敏[2] 苏雪平 侯乐乐 ZHOU Kun;GUO Junxin;LUO Jie;LI Yunhong;LI Limin;SU Xueping;HOU Lee(CHN Energy Xinjiang Jilintai Hydropower Development Co.,Ltd.,Yili 835000,China;School of Electronics and Information,Xi’an Polytechnic University,Xi’an 710048,China)

机构地区：[1]国家能源集团新疆吉林台水电开发有限公司,新疆伊犁835000 [2]西安工程大学电子信息学院,陕西西安710048

出　　处：《红外技术》2024年第11期1308-1314,共7页Infrared Technology

基　　金：国家自然科学基金(62203344);陕西省科技厅自然科学基础研究重点项目(2022JZ-35);陕西高校青年科技创新团队项目。

摘　　要：针对目前水电站设备缺陷检测精度低、效率慢、花费时间长和人力成本大等问题,提出了一种基于注意力机制的红外测温数据的水电站设备缺陷检测算法。首先在YOLO(You Only Look Once)v5的特征提取网络中引入新的注意力模块CA(Coordinate Attention),以便更好地捕捉输入中的关键信息,提高模型性能和准确度。其次使用Ghost Convolution和Bottleneck Transformer替换YOLOv5中的普通卷积和C3模块,减少模型的参数量和计算量。最后,将缺陷检测算法和PyQt5界面相结合,实现可视化检测。实验结果表明,改进后的YOLOv5模型的mAP@0.5提升了6.3%,P和R分别提升了6.6%和5.9%,满足水电站设备缺陷的检测需求且具有更高的检测精度。To address the challenges of low accuracy,low efficiency,and significant time and labor costs of the current defect detection process for hydropower station equipment,a defect detection algorithm for hydropower station equipment based on the attention mechanism of infrared temperature measurement data is proposed.First,a new attention module,Coordinate Attention(CA),is introduced into the YOLOv5s feature extraction network to better capture key information in the input and improve the model performance and accuracy.Second,we replaced the ordinary convolution and C3 modules in YOLOv5 with Ghost Revolution and Bottleneck Transformer to reduce the model's parameters and computational complexity.Finally,the defect detection algorithm is combined with the PyQt5 interface to achieve visual image detection and output detection results.The experimental results show that the improved YOLOv5 model's mAP@0.5 is increased by 6.3%,and P and R are increased by 6.6%and 5.9%,respectively,meeting the detection needs of hydropower station equipment defects and achieving a higher detection accuracy.

关 键 词：红外测温 深度学习 YOLOv5 缺陷检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]