金属管件表面微尺度磕碰伤的卷积神经网络模型构建  

Construction of convolutional neural network model for micro-scale bump on metal pipe fittings

作　　者：刘子豪 陶国好 薛峰 鹿业波 杨俊 Liu Zihao;Tao Guohao;Xue Feng;Lu Yebo;Yang Jun(College of Mechanical Engineering,Tianjin University,Tianjin 300072,China;College of Artificial Intelligence,Jiaxing University,Jiaxing,Zhejiang 314100,China;School of Computer Science and Technology,Zhejiang Sci-Tech University,Hangzhou,Zhejiang 310018,China;Zhejiang Master Hydraulic Fittings Co.,Ltd.,Jiaxing,Zhejiang 316002,China;School of Mechanical Engineering,Jiaxing University,Jiaxing,Zhejiang 314100,China)

机构地区：[1]天津大学机械工程学院,天津300072 [2]嘉兴大学人工智能学院,浙江嘉兴314100 [3]浙江理工大学计算机科学与技术学院,浙江杭州310018 [4]浙江迈思特液压管件股份有限公司,浙江嘉兴316002 [5]嘉兴大学机械工程学院,浙江嘉兴314100

出　　处：《光电工程》2025年第3期55-68,共14页Opto-Electronic Engineering

基　　金：国家自然科学基金面上项目(62374074);浙江省“尖兵领雁”研发攻关计划(2024C04028);嘉兴市公益性研究计划项目(2024AY10059,2024AD10045,2024AY40010);校企合作项目(00523144);海盐重点研发计划项目(2024ZD03);嘉兴大学勤慎骨干学者人才项目(CD70623008)。

摘　　要：金属管件表面微小缺陷低检出率是工业零部件检测中面临的关键问题。针对此问题,本文构造一种改进YOLOv9-MM模型,以提高小目标检测的准确性。设计了一种针对精密金属管件的图像实时采集系统,采用环形光源结合远心镜头,可实现管件表面的全角度覆盖,消除缺失区域导致的漏检问题;引入浅层网络的特征图,结合Dysample上采样模块,实现深度特征的动态融合;通过改进损失函数,提高小目标检测的准确率。结果表明,所提方法平均检测精度达到70.2%,检测速度达到90f/s。所提方法在应用中展现出一定的可行性。The low detection rate of tiny defects on the surface of metal pipe fittings is a key issue confronting industrial component inspection.In aiming at this problem,an improved YOLOv9-MM model was constructed to improve the accuracy of small target detection.A real-time image acquisition system for precision metal pipe fittings was designed.By using an annular light source combined with a telecentric lens,the surface of pipe fittings can be snapped by the CCD camera and covered at all angles to eliminate the problem of missing areas.The feature map extracted methods of shallow network were introduced,and the upper sampling module of Dysample was combined to realize the dynamic fusion of depth features.By improving the loss function,the precision of small target detection is greatly improved.The results show that the proposed method has an average detection accuracy of 70.2%and a detection speed of 90 f/s.The proposed method shows some feasibility in the actual application.

关 键 词：金属管件 微小缺陷检测 深度学习 YOLOv9 检测系统 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]