基于卷积神经网络的CO2焊接熔池图像状态识别方法

覃科, 刘晓刚, 丁立新

覃科, 刘晓刚, 丁立新. 基于卷积神经网络的CO2焊接熔池图像状态识别方法[J]. 焊接, 2017, (6): 21-26.
引用本文: 覃科, 刘晓刚, 丁立新. 基于卷积神经网络的CO2焊接熔池图像状态识别方法[J]. 焊接, 2017, (6): 21-26.
Qin Ke, Liu Xiaogang, Ding Lixin. Recognition of molten pool morphology in CO2 welding based on convolution neural network[J]. WELDING & JOINING, 2017, (6): 21-26.
Citation: Qin Ke, Liu Xiaogang, Ding Lixin. Recognition of molten pool morphology in CO2 welding based on convolution neural network[J]. WELDING & JOINING, 2017, (6): 21-26.

基于卷积神经网络的CO2焊接熔池图像状态识别方法

Recognition of molten pool morphology in CO2 welding based on convolution neural network

  • 摘要: 为了通过熔池图像对焊接状态进行判断,将卷积神经网络引入到CO2焊接熔池图像状态识别中,提出了一种CO2焊接熔池状态识别卷积神经网络CNN-M。该网络使用简单预处理的熔池图像作为输入向量,避免了人工提取图像特征的主观性对识别率的不良影响。同时,CNN-M采用了ReLU激活函数、随机Dropout及SVM分类器来降低样本集稀少可能导致的网络过拟合现象。试验结果表明,和人工提取熔池特征状态作为输入向量的BP神经网络相比,CNN-M在识别率及识别速度方面均体现出了更好的性能,其良好的泛化能力能够满足在线熔池状态监控的要求。
    Abstract: A kind of convolution neutral network CNN-M was proposed to recognize the morphology of molten pool in CO2 welding. A simple pretreatment of the molten pool image was adopted as the input vector in CNN-M, which avoids the adverse effects brought by the subjectivity of artificial extraction of molten pool image features. In order to decrease the possibility of network over-fitting caused by the sparse training data set, several methods including ReLU activation function, random Dropout and SVM classifier were used in CNN-M. The experimental results show that CNN-M has better performance in recognition rate and recognition speed than the BP neural network does, which adopt the characteristic values of the molten pool as the input vector. The performance of CNN-M is able to meet the requirement of the on-line molten pool monitoring.
  • [1] Zhang Y,Gao X,Katayama S. Weld appearance prediction with BP neural network improved by genetic algorithm during disk laser welding[J]. Journal of Manufacturing Systems,2015,34: 53-59.
    [2] Gao Xiangdong,Zhang Yanxi. Prediction model of weld width during high-power disk laser welding of 304 austenitic stainless steel[J]. International Journal of Precision,Engineering and Manufacturing,2014,15(3): 399-405.
    [3] 林 俊. 电弧焊熔透状态视觉检测模型研究[D].广州:广东工业大学硕士学位论文,2015.
    [4] 刘建伟,刘 媛,罗雄麟. 深度学习研究进展[J]. 计算机应用研究,2014,31(7): 1921-1942.
    [5] Krizhevsky A,Sutskever I,Hinton GE. ImageNet classification with deep convolutional neural networks[C]. Advances in Neural Information Processing Systems, Curran Associates RedHook,NY,USA ,2012: 1097-1105.
    [6] 夏胜全,区智明,孙晓明. 一种降低CO2气体保护焊接能量输入的方法[J]. 焊接,2012(3): 32-35.
    [7] 向远鹏,曹 彪. 熔滴短路过渡频率对CO2焊接过程稳定性的影响[J]. 焊接,2010(12): 33-37.
    [8] 齐志龙,李 科,孙佳男,等. CO2气体保护焊短路过渡熔滴尺寸的研究[J]. 焊接,2016(11): 34-37.
    [9] 刘晓刚,谢存禧,张昌年,等. 基于弧光反射的焊缝图像获取及图像处理[J]. 焊接学报,2008,29(4): 73-76.
    [10] 杨 钊,陶大鹏,张树业,等. 大数据下的基于深度神经网的相似汉字识别[J]. 通信学报,2014,9(35): 184-189.
    [11] Khemchandani R, Pal A. Multi-category laplacian least squares twin support vector machine[J]. Applied Intelligence,2016: 1-17.
    [12] Hinton G E,Srivastava N,Krizhevsky A,et al. Improving neural networks by preventing co-adaptation of feature detectors[J]. Computer Science,2012,3(4): 212-223.
    [13] Mendenhall J,Meiler J. Improving quantitative structure-activity relationship models using Artificial Neural Networks trained with dropout[J]. Journal of computer-aided molecular design,2016,30: 1-13.
    [14] Bouvrie J. Notes on convolution neural networks[R]. MIT CBCL,2006.
计量
  • 文章访问数: 
  • HTML全文浏览量:  0
  • PDF下载量: 
  • 被引次数: 0
出版历程
  • 收稿日期:  2017-01-20
  • 发布日期:  2017-06-24

目录

    /

    返回文章
    返回