扫 描 看 全 文
王杰，张志芬，秦锐，等.基于机器学习的不锈钢薄板MIG焊焊穿缺陷识别［J］.电焊机，2023，53（9）：70-77.
WANG Jie, ZHANG Zhifen, QIN Rui, et al.Identification of MIG Welding Burn-through Defects in Stainless Steel Sheet Based on Machine Learning[J].Electric Welding Machine, 2023, 53(9): 70-77.
随着焊接工艺在工业环境中的广泛应用，对焊接缺陷的研究受到越来越多的关注，焊穿缺陷作为焊接过程中最严重的缺陷之一，必须采取相应的措施对其进行监测以及控制。提出了一种采用被动视觉传感技术，利用工业CCD相机结合减光、滤光系统消除焊接过程中强弧光干扰，获得MIG焊熔池图像的方法。设计正交试验并从图像的多维特征出发，提取熔池图像4个种类对应的10个维度特征作为机器学习的输入。在进行焊穿缺陷的标签标定时，基于所提出的特征提取对正交试验中获得的熔池图像数据进行预处理，通过反向搜索实现对无效数据的剔除和有效数据的正确归类，同时发现了数据在空间分布的规律。最后探究了不同维度输入特征下三种分类器的识别效果，结果表明：所选择的特征在三种分类器上均获得了不低于97%的分类精度，为后续开发基于视觉在线焊穿缺陷检测系统提供了技术支撑和理论依据。
With the wide application of welding technology in industrial environment, more and more scholars pay attention to the study of welding defects. As one of the most serious defects in the welding process, the corresponding measures must be taken to monitor and control the burn-through defects. In this paper, a passive visual sensing technique is proposed to eliminate the strong arc light interference in the welding process by using industrial CCD camera combined with dimming and filtering system, then obtain MIG welding pool image. Based on the multi-dimensional features of the image, ten dimensional features corresponding to the four categories of molten pool image are extracted as the input of machine learning. During the label calibration of normal and burn-through defects, the image data of molten pool obtained from orthogonal test were preprocessed based on the proposed feature extraction, the invalid data were eliminated and the valid data were correctly classified by reverse search. Meanwhile, the spatial distribution of data was found. Finally, the recognition effects of the three classifiers under the input features of different dimensions were explored, and the results showed that the selected features obtained classification accuracy of no less than 97% on the three classifiers, which provided technical support and theoretical basis for the subsequent development of visual online burn-through defects detection system.
MIG焊视觉传感熔池图像处理机器学习缺陷识别
MIG weldingvision sensingmolten pool image processingmachine learningdefect identification
李健. 基于激光视觉的镀锌钢板MIG焊接质量监测［D］. 江西：南昌大学， 2020.
LI J. MIG Welding Quality Monitoring of Galvanized Steel Sheet Based on Laser Vision［D］. Jiangxi：Nanchang University，2020.
Cui L， Pan F， He D， et al. Joining 6 mm thick dissimilar steel/Al alloy butt joints using two keyhole laser-MIG hybrid welding processes［J］. Journal of Laser Applications， 2018， 30（3）： 32418.
石玕，樊丁，吴巍. 铝合金MIG焊熔池图像的形态学处理方法［J］. 焊接学报， 2005（03）： 37-40.
SHI Y，FAN D，WU W. Morphology processing image of aluminum alloy metal inert gas welding pool［J］. Transactions of The China Welding Institution， 2005（03）： 37-40.
石玗，吴巍，樊丁，等. 铝合金MIG焊正面熔池图像视觉传感与处理［J］. 中国机械工程， 2005（02）： 84-87.
SHI Y，WU W，FAN D，et al. Vision-based Image Taking and Processing of Top-side Molten Pool in Aluminum Alloy MIG Welding Process［J］. China Mechanical Engineering， 2005（02）： 84-87.
王建军，林涛，陈善本. 铝合金TIG焊熔池图像的获取与处理［J］. 机械工程学报， 2003（05）： 125-129.
WANG J J，LIN T，CHEN S B. Image Taking and Processing of Molten Pool Characters During Aluminum Alloy TIG Welding［J］. Chinese Journal of Mechanical Engineering， 2003（05）： 125-129.
张帆，泮佳俊，刘腾，等. 基于机器视觉的焊缝识别研究现状与发展趋势［J］. 电焊机，2022，52（7）：24-33.
ZHANG F， PAN J J， LIU T，et al. Research status and Development trend of welding seam recognition based on Machine Vision［J］.Electric Welding Machine，2022，52（7）：24-33.
王一珺. 图像特征提取算法研究及应用［D］. 江苏：南京农业大学， 2015.
WANG Y J. MIG Welding Quality Monitoring of Galvanized Steel Sheet Based on Laser Vision［D］. Jiangsu： Nanjing Agricultural University， 2015.
Cai W， Wang J， Zhou Q， et al. Equipment and Machine Learning in Welding Monitoring： A Short Review［C］//ACM， 2019.
薛龙， 曹楷顺， 黄军芬， 等. 基于深度学习的焊缝定位与缺陷识别［J］. 电焊机， 2021， 51（9）：31-35.
XUE L， CAO K S， HUANG J F， et al. Weld location and defect identification based on deep learning［J］. Electric Welding Machine， 2021， 51（9）：31-35.
Zhang S， Li X， Zong M， et al. Efficient kNN Classification With Different Numbers of Nearest Neighbors ［J］. IEEE Transactions on Neural Networks and Learning Systems， 2018， 29（5）： 1774-1785.
解晓童. 基于主被动视觉传感的机器人焊接过程焊缝跟踪与熔池监测研究［D］. 上海：上海交通大学，2020.
XIE X T. Insitu Studies of Full Field Strainand Resdual Stress of Welding Heataffect Zone Using Digital Image Correlation［D］. Shanghai： Shanghai JiaoTong University，2020.
龙鹏，马前进，薛家祥. 基于正交试验的铝合金薄板脉冲MIG焊参数优化方法研究［J］. 电焊机， 2013， 43（4）： 55-58.
LONG P，MA Q J，XUE J X. Research on the parameter optimization method of pulsed MIG welding for sheet aluminum alloy based on orthogonal test［J］. Electric Welding Machine， 2013， 43（4）： 55-58.
董金枋，汤大赟，吴頔，等. 强噪声下的激光视觉焊缝跟踪图像处理研究现状及展望［J］. 电焊机， 2022，52（12）：1-16.
DONG J F， TANG D Y， WU D， et al.Research Status and Prospect of Laser Vision Weld Seam Tracking Image Processing Under Strong Noise［J］. Electric Welding Machine， 2022， 52（12）： 1-16.
洪宇翔，杨明轩，都东，等. 铝合金爬坡TIG焊熔池失稳状态的视觉检测［J］. 焊接学报， 2021， 42（10）： 8-13.
HONG Y X，YANG M X，DU D，et al. Unstable state vision detection of molten pool during aluminum alloy climbing-TIG welding［J］. Transactions of The China Welding Institution， 2021， 42（10）： 8-13.
于英飞，朱志明，孙博文，等. 焊接电弧图像的边缘检测及其批处理算法［J］. 焊接学报， 2018， 39（11）： 17-21.
YU Y F，ZHU Z M， SUN B W，et al. Edges detection and batch algorithm for welding arc images［J］. Transactions of The China Welding Institution， 2018， 39（11）： 17-21.
编辑部网址：http：//www.71dhj.comhttp://www.71dhj.com
相关作者
相关机构
公网安备11010802024621