不同机器学习模型对钢板缺陷分类的性能比较Comparison on different machine learning models′ steel plate defect classification performance
刘莉琳;谭荣;高翔;
摘要(Abstract)：
为了评估机器学习技术在钢板缺陷分类中的应用,该研究基于CART决策树、RF、MLPNN和CNN建立了4种不同的机器学习分类模型,对UCI机器学习库的钢板缺陷数据集进行分类,通过混淆矩阵、准确率等不同指标评估了4种模型对7种常见缺陷的分类性能。其中,CNN模型在训练集和测试集上的准确率分别达到了98.67%和97.41%,取得了卓越的分类性能。此外,RF模型相对CART模型可以更好地处理过拟合问题。实验结果表明,神经网络尤其是CNN模型对钢板缺陷分类问题具有更好的性能。
关键词(KeyWords)： 钢板缺陷分类;机器学习;CART;RF;MLPNN;CNN
基金项目(Foundation): 国家重点研发计划(2016YFB0700504);; 陕西省科技计划项目(2018GY-048)
作者(Author): 刘莉琳;谭荣;高翔;
Email:
DOI: 10.16652/j.issn.1004-373x.2021.01.022
参考文献(References)：
- [1]兰红，方治屿.3维灰度矩阵的钢板缺陷图像识别[J].中国图象图形学报，2019(6):859-869.
- [2]董秀.基于全局-局部SVM的钢板表面缺陷分类[J].信息技术，2019(6):5-8.
- [3] SIMI?D,SVIR?EVI?V,SIMI?S. An approach of steel plates fault diagnosis in multiple classes decision making[J]. Hybrid artificial intelligence systems,2014,8480:86-97.
- [4] ZHANG Yu,WANG Shen,WANG Qing,et al. Profile imaging of actual defects in steel plate based on electromagnetic ultrasonic SH guided wave scattering[J]. Insight:non-destructive testing and condition monitoring,2017,59(7):383-388.
- [5]王斌.中厚钢板分层缺陷的力学性能影响分析[J].物理测试，2015,33(4):46-48.
- [6] FULKERSON B. Machine learning,neural and statistical classification[J]. Technometrics,1995,37(4):459.
- [7] ZHAO Y J,YAN Y H,SONG K C. Vision-based automatic detection of steel surface defects in the cold rolling process:considering the influence of industrial liquids and surface textures[J]. International journal of advanced manufacturing technology,2016,16(5):1-14.
- [8]徐海燕，郭京.基于HMT模型的带钢表面缺陷检测研究[J].计算机应用与软件，2015(3):325-330.
- [9]胡慧君，李元香，刘茂福，等.基于机器学习的带钢表面缺陷分类方法研究[J].计算机工程与设计，2014(2):620-624.
- [10]李国和，王峰，郑阳，等.基于决策树生成及剪枝的数据集优化及其应用[J].计算机工程与设计，2018(1):205-211.
- [11] AHMAD M W,MOURSHED M,REZGUI Y. Trees vs neurons:comparison between random forest and ANN for high-resolution prediction of building energy consumption[J]. Energy and buildings,2017,147:77-89.
- [12] LU M,SADIQ S,FEASTER D J,et al. Estimating individual treatment effect in observational data using random forest methods[J]. Journal of computational and graphical statistics,2018,27(1):209-219.
- [13] MASNATA A,SUNSERI M. Neural network classification of flaws detected by ultrasonic means[J]. NDT&E international,1996,29(2):87-93.
- [14] YU K C,SALZMANN M. Second-order convolutional neural networks[J]. Clinical immunology&immunopathology,2017,66(3):230-238.
- [15]徐永林，王香蒙，李鑫，等.基于高通量计算及机器学习的新材料带隙预测[J].中国科学：技术科学，2019(1):44-54.