This paper proposes a machine-learning-based methodology to automatically classify different types of steel weld defects,including lack of the fusion,porosity,slag inclusion,and the qualified(no defects)cases.This met...This paper proposes a machine-learning-based methodology to automatically classify different types of steel weld defects,including lack of the fusion,porosity,slag inclusion,and the qualified(no defects)cases.This methodology solves the shortcomings of existing detection methods,such as expensive equipment,complicated operation and inability to detect internal defects.The study first collected percussed data from welded steel members with or without weld defects.Then,three methods,the Mel frequency cepstral coefficients,short-time Fourier transform(STFT),and continuous wavelet transform were implemented and compared to explore the most appropriate features for classification of weld statuses.Classic and convolutional neural network-enhanced algorithms were used to classify,the extracted features.Furthermore,experiments were designed and performed to validate the proposed method.Results showed that STFT achieved higher accuracies(up to 96.63%on average)in the weld status classification.The convolutional neural network-enhanced support vector machine(SVM)outperformed six other algorithms with an average accuracy of 95.8%.In addition,random forest and SVM were efficient approaches with a balanced trade-off between the accuracies and the computational efforts.展开更多
The cause of the premature failure of 304 stainless steel tube heat exchanger was investigated.The unique skeleton structure inside the leakage point reveals that this is a new damage mechanism caused by a δ+γ two-p...The cause of the premature failure of 304 stainless steel tube heat exchanger was investigated.The unique skeleton structure inside the leakage point reveals that this is a new damage mechanism caused by a δ+γ two-phase structure and crevice corrosion.The three-dimensional structure of the leakage point was demonstrated using X-ray diffraction topography.The results of scanning electron microscope examination show the microstructure of the weld to be columnar and dendritic.It is found by electron probe microscope analysis and transmission electron microscopy that columnar dendrites consisted of γ-dendrite and an amount of δ-ferrite phases at the dendrite trunk.Simulated corrosion test results confirmed that the corrosion medium was the chloride ion.Crevice corrosion of chloride ions occurred at weld defects on the inner wall thus forming a concentration cell.Grains of columnar dendrites were then corroded by chloride ions and δ-ferrite phases on the grain boundaries were retained,which formed the particular skeleton corrosion structure.As a result,it led to leakage when the corrosion of weld occurred from the inner wall to the outer wall.展开更多
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...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。展开更多
基金support of Shanghai Pinlan Data Technology Co.,Ltd.,and Open Fund of Shanghai Key Laboratory of Engineering Structure Safety,SRIBS(No.2021-KF-06).
文摘This paper proposes a machine-learning-based methodology to automatically classify different types of steel weld defects,including lack of the fusion,porosity,slag inclusion,and the qualified(no defects)cases.This methodology solves the shortcomings of existing detection methods,such as expensive equipment,complicated operation and inability to detect internal defects.The study first collected percussed data from welded steel members with or without weld defects.Then,three methods,the Mel frequency cepstral coefficients,short-time Fourier transform(STFT),and continuous wavelet transform were implemented and compared to explore the most appropriate features for classification of weld statuses.Classic and convolutional neural network-enhanced algorithms were used to classify,the extracted features.Furthermore,experiments were designed and performed to validate the proposed method.Results showed that STFT achieved higher accuracies(up to 96.63%on average)in the weld status classification.The convolutional neural network-enhanced support vector machine(SVM)outperformed six other algorithms with an average accuracy of 95.8%.In addition,random forest and SVM were efficient approaches with a balanced trade-off between the accuracies and the computational efforts.
文摘The cause of the premature failure of 304 stainless steel tube heat exchanger was investigated.The unique skeleton structure inside the leakage point reveals that this is a new damage mechanism caused by a δ+γ two-phase structure and crevice corrosion.The three-dimensional structure of the leakage point was demonstrated using X-ray diffraction topography.The results of scanning electron microscope examination show the microstructure of the weld to be columnar and dendritic.It is found by electron probe microscope analysis and transmission electron microscopy that columnar dendrites consisted of γ-dendrite and an amount of δ-ferrite phases at the dendrite trunk.Simulated corrosion test results confirmed that the corrosion medium was the chloride ion.Crevice corrosion of chloride ions occurred at weld defects on the inner wall thus forming a concentration cell.Grains of columnar dendrites were then corroded by chloride ions and δ-ferrite phases on the grain boundaries were retained,which formed the particular skeleton corrosion structure.As a result,it led to leakage when the corrosion of weld occurred from the inner wall to the outer wall.
文摘利用超声波水浸聚焦入射法对1 mm厚的SUS304奥氏体不锈钢板点焊接头进行超声C扫描成像检测.分析了不同焊接工艺参数下的C扫描图像特征,甄别了飞溅、焊穿等典型焊接缺陷,并提取其对应的A扫描信号.基于C扫描图像对焊核直径进行了测量,并与焊核切口端面尺寸进行了比较.结果表明,基于超声波水浸聚焦入射法得到的C扫描图像,能有效观测焊核内部形貌特征.焊接电流超过8 k A,电极力小于2 700 N时,超声波C扫描图像中清晰反映出飞溅、焊穿等缺陷,其对应区域的A扫描信号与正常熔核区波形特征有明显差异;借助超声C扫描图像测得的焊核直径为4.39~5.25 mm.
文摘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。
