第16卷第3期 智能系统学报 Vol.16 No.3 2021年5月 CAAI Transactions on Intelligent Systems May 2021 D0L:10.11992tis.202005013 融合迁移学习的AlexNet神经网络不锈钢焊缝缺陷分类 陈立潮，闫耀东，张睿，傅留虎2，曹建芳3 (1,太原科技大学计算机科学与技术学院，山西太原030024,2.山西省机电设计研究院机械产品质量监督检 验站，山西太原030009：3.忻州师范学院计算机科学与技术系，山西忻州034000) 摘要：针对不锈钢焊缝缺陷特征提取存在主观单一性和客观不充分性等问题，提出一种融合迁移学习的 AlexNet卷积神经网络模型，用于不锈钢焊缝缺陷的自动分类。首先，由于不锈钢焊缝缺陷数据较为缺乏，通过采 用迁移学习对网络前3层冻结，减少网络对输入数据量的要求：对后2层卷积层提取的特征信息批量归一化 (batch normalization,BN),以加快网络的收敛速度；并使用带泄露线性整流(leaky rectified linear unit,. LeakyReLU)函数对抑制神经元进行激活，从而提高模型的鲁棒性和特征提取能力。结果表明，该模型最终达到 了9512%的准确率，相比原结构识别精度提高了9.8%。验证了改进后方法能够对裂纹、气孔、夹渣、未熔合和 未焊透5类不锈钢焊缝缺陷实现高精度分类。相比现有方法，其识别面更广，精度更高，具有一定的工程实践意义。 关键词：不锈钢焊缝缺陷分类；卷积神经网络；图像预处理；AlexNet模型；迁移学习：数据增强；焊缝数据集；深 度学习 中图分类号：TP391.4文献标志码：A文章编号：1673-4785(2021)03-0537-07 中文引用格式：陈立潮，闫耀东，张睿，等.融合迁移学习的AlxNεt神经网络不锈钢焊缝缺陷分类J.智能系统学报，2021， 16(3):537-543. 英文引用格式：CHEN Lichao,YAN Yaodong,.ZHANG Rui,etal.Welding defect classification of stainless steel based on AlexNet neural network combined with transfer learning[Jl.CAAI transactions on intelligent systems,2021,16(3):537-543. Welding defect classification of stainless steel based on AlexNet neural network combined with transfer learning CHEN Lichao',YAN Yaodong',ZHANG Rui',FU Liuhu',CAO Jianfang3 (1.School of Computer Science and Technology,Taiyuan University of Science and Technology,Taiyuan 030024,China;2.Mech- anical Product Quality Supervision and Inspection Station,Shanxi Mechanical and Electrical Design&Research Institute,Taiyuan 030009,China:3.Department of Computer Science and Technology,Xinzhou Teachers University,Xinzhou 034000,China) Abstract:In order to solve the problems of subjectivity and objectivity in feature extraction of stainless steel weld de- fects,an AlexNet convolutional neural network model based on transfer learning is proposed for automatic classifica- tion of stainless steel weld defects.First,due to the lack of stainless steel weld defect data,the first three layers of the network are frozen by transfer learning,which reduces the requirement of the network on the input data.In order to speed up the convergence of the network,the batch normalization(BN)of the feature information extracted from the two latter layers of convolution is carried out.The LeakyReLU function is used to activate the features in the negative interval so as to improve the robustness of the model and the ability of feature extraction.The results show that the ac- curacy of the model is 95.12%,and the recognition accuracy is 9.8%higher than that of the original structure.It has been verified that the improved method can classify five kinds of stainless steel weld defects such as crack,blowhole, slag inclusion,incomplete fusion,and incomplete penetration with high precision.Compared to the existing methods, this method has a wider recognition area,higher accuracy,and certain engineering significance. Keywords:classification of weld defects in stainless steel;convolutional neural network,image preprocessing;AlexNet model;the migration study;data enhancement;weld data set;deep learning 收稿日期：2020-05-10. 不锈钢材料因具有良好的抗氧化性、抗腐蚀 基金项目：先进控制与装备智能化山西省重点实验室开放课 题(ACEI202002):山西省高等学校科技创新项目 性和易焊接等特点逐渐被推广到储存输送、加工 (2019L0653):山西省应用基础研究项目 201801D221179). 生产、机械制造等行业。不锈钢工件是利用焊接 通信作者：张容.E-mail:zhangrui(@yust.edu.cn. 来实现两两结合，由于焊接存在局部不可见性，DOI: 10.11992/tis.202005013 融合迁移学习的 AlexNet 神经网络不锈钢焊缝缺陷分类 陈立潮1 ，闫耀东1 ，张睿1 ，傅留虎2 ，曹建芳1,3 （1. 太原科技大学 计算机科学与技术学院，山西 太原 030024; 2. 山西省机电设计研究院 机械产品质量监督检 验站，山西 太原 030009; 3. 忻州师范学院 计算机科学与技术系，山西 忻州 034000） 摘 要 ：针对不锈钢焊缝缺陷特征提取存在主观单一性和客观不充分性等问题，提出一种融合迁移学习的 AlexNet 卷积神经网络模型，用于不锈钢焊缝缺陷的自动分类。首先，由于不锈钢焊缝缺陷数据较为缺乏，通过采 用迁移学习对网络前 3 层冻结，减少网络对输入数据量的要求；对后 2 层卷积层提取的特征信息批量归一化 (batch normalization, BN)，以加快网络的收敛速度；并使用带泄露线性整流 (leaky rectified linear unit, LeakyReLU) 函数对抑制神经元进行激活，从而提高模型的鲁棒性和特征提取能力。结果表明，该模型最终达到 了 95.12% 的准确率, 相比原结构识别精度提高了 9.8%。验证了改进后方法能够对裂纹、气孔、夹渣、未熔合和 未焊透 5 类不锈钢焊缝缺陷实现高精度分类。相比现有方法，其识别面更广，精度更高，具有一定的工程实践意义。 关键词：不锈钢焊缝缺陷分类；卷积神经网络；图像预处理；AlexNet 模型；迁移学习；数据增强；焊缝数据集；深 度学习 中图分类号：TP391.4 文献标志码：A 文章编号：1673−4785(2021)03−0537−07 中文引用格式：陈立潮, 闫耀东, 张睿, 等. 融合迁移学习的 AlexNet 神经网络不锈钢焊缝缺陷分类[J]. 智能系统学报, 2021, 16(3): 537–543. 英文引用格式：CHEN Lichao, YAN Yaodong, ZHANG Rui, et al. Welding defect classification of stainless steel based on AlexNet neural network combined with transfer learning[J]. CAAI transactions on intelligent systems, 2021, 16(3): 537–543. Welding defect classification of stainless steel based on AlexNet neural network combined with transfer learning CHEN Lichao1 ，YAN Yaodong1 ，ZHANG Rui1 ，FU Liuhu2 ，CAO Jianfang1,3 (1. School of Computer Science and Technology, Taiyuan University of Science and Technology, Taiyuan 030024, China; 2. Mech￾anical Product Quality Supervision and Inspection Station, Shanxi Mechanical and Electrical Design & Research Institute, Taiyuan 030009, China; 3. Department of Computer Science and Technology, Xinzhou Teachers University, Xinzhou 034000, China) Abstract: In order to solve the problems of subjectivity and objectivity in feature extraction of stainless steel weld de￾fects, an AlexNet convolutional neural network model based on transfer learning is proposed for automatic classifica￾tion of stainless steel weld defects. First, due to the lack of stainless steel weld defect data, the first three layers of the network are frozen by transfer learning, which reduces the requirement of the network on the input data. In order to speed up the convergence of the network, the batch normalization (BN) of the feature information extracted from the two latter layers of convolution is carried out. The LeakyReLU function is used to activate the features in the negative interval so as to improve the robustness of the model and the ability of feature extraction. The results show that the ac￾curacy of the model is 95.12%, and the recognition accuracy is 9.8% higher than that of the original structure. It has been verified that the improved method can classify five kinds of stainless steel weld defects such as crack, blowhole, slag inclusion, incomplete fusion, and incomplete penetration with high precision. Compared to the existing methods, this method has a wider recognition area, higher accuracy, and certain engineering significance. Keywords: classification of weld defects in stainless steel; convolutional neural network; image preprocessing; AlexNet model; the migration study; data enhancement; weld data set; deep learning 不锈钢材料因具有良好的抗氧化性、抗腐蚀 性和易焊接等特点逐渐被推广到储存输送、加工 生产、机械制造等行业。不锈钢工件是利用焊接 来实现两两结合，由于焊接存在局部不可见性， 收稿日期：2020−05−10. 基金项目：先进控制与装备智能化山西省重点实验室开放课 题 (ACEI202002)；山西省高等学校科技创新项目 (2019L0653) ；山西省应用基础研究项 目 (201801D221179). 通信作者：张睿. E-mail：zhangrui@tyust.edu.cn. 第 16 卷第 3 期 智 能 系 统 学 报 Vol.16 No.3 2021 年 5 月 CAAI Transactions on Intelligent Systems May 2021