Cite this article as:

Research Article

Prediction of Charpy V-notch impact energy of low carbon steel by using shallow neural network and deep learning

+ Author Affiliations
  • Received: 17 July 2020Revised: 6 August 2020Accepted: 10 August 2020Available online: 14 August 2020
  • In the present work, the impact energy prediction model of low carbon steel was investigated based on the industrial data. A three layer neural network, extreme learning machine and deep neural network were compared with different activation functions, structure parameters and training functions. To determine the optimal hyper-parameters of deep neural network, Bayesian optimization was applied. The model with best performance was applied to investigate importance degree of process parameter variables on impact energy of low carbon steel. The results show that deep neural network obtains better prediction results than that of shallow neural network due to the multiple hidden layers improving the learning ability of the model. Among all the models, the Bayesian optimization deep neural network achieves the highest correlation coefficient of 0.9536, lowest mean absolute relative error of 0.0843 and lowest root mean square error of 17.34 J for predicting the impact energy of low carbon steel. Among all the variables, the main factors affecting the impact energy of low carbon steel with final thickness of 7.5 mm are the thickness of the original slab, the thickness of intermediate slab and rough rolling exit temperature on the specific hot rolling production line.
  • 加载中
  •  

  • [1] Shahab Shahsavar,Mostafa Ketabchi, and Saeed Bagherzadeh, Fabrication of robust aluminum–carbon nanotube composites using ultrasonic assembly and rolling process, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-020-1969-4
    [2] Qiu-wei Xing,Jiang Ma, and Yong Zhang, Phase thermal stability and mechanical properties analyses of (Cr,Fe,V)–(Ta,W) multiple-based elemental system using a compositional gradient film, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-020-2063-7
    [3] Xiao-man Tian,Shen-xu Bao, and Yi-min Zhang, Adsorption properties of V(IV) on the resin-activated carbon composite electrodes in capacitive deionization, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-020-2100-6
    [4] Hai-zhen Wang,Yun-dong Zhao,Yue-hui Ma, and Zhi-yong Gao, Effect of low-energy proton on the microstructure, martensitic transformation and mechanical properties of irradiated Ni-rich TiNi alloy thin films, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-019-1893-7
    [5] Fan Zeng,Xue-jiao Bai,Cheng-liang Hu,Min-jun Tang, and Zhen Zhao, Effect of plastic strain and forming temperature on magnetic properties of low-carbon steel, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-019-1905-7
    [6] Man Liu,Guang Xu,Jun-yu Tian,Qing Yuan, and Xin Chen, Effect of austempering time on microstructure and properties of a low-carbon bainite steel, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-019-1881-y
    [7] Zheng-hua Deng,Hai-qing Yin,Xue Jiang,Cong Zhang,Guo-fei Zhang,Bin Xu,Guo-qiang Yang,Tong Zhang,Mao Wu, and Xuan-hui Qu, Machine-learning-assisted prediction of the mechanical properties of Cu–Al alloy, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-019-1894-6
    [8] Azhar Iqbal, Long Chen, Yong Chen, Yu-xian Gao, Fang Chen, and  Dao-cong Li, Erratum to: Lithium-ion full cell with high energy density using nickel-rich LiNi0.8Co0.1Mn0.1O2 cathode and SiO-C composite anode, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-019-1732-x
    [9] Li-ying Huang, Kuai-she Wang, Wen Wang, Kai Zhao, Jie Yuan, Ke Qiao, Bing Zhang, and  Jun Cai, Mechanical and corrosion properties of low-carbon steel prepared by friction stir processing, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-019-1725-9
    [10] Jalil Pazhoohan, Hossein Beiki, and  Morteza Esfandyari, Experimental investigation and adaptive neural fuzzy inference system prediction of copper recovery from flotation tailings by acid leaching in a batch agitated tank, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-019-1762-4
    [11] Vinayak Dakre, D. R. Peshwe, S. U. Pathak, and  Ajay Likhite, Effect of austenitization temperature on microstructure and mechanical properties of low-carbon-equivalent carbidic austempered ductile iron, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-018-1625-4
    [12] Azhar Iqbal, Long Chen, Yong Chen, Yu-xian Gao, Fang Chen, and  Dao-cong Li, Lithium-ion full cell with high energy density using nickel-rich LiNi0.8Co0.1Mn0.1O2 cathode and SiO-C composite anode, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-018-1702-8
    [13] Dan Wu, Fu-ming Wang, Jin Cheng, and  Chang-rong Li, Effect of Nb and V on the continuous cooling transformation of undercooled austenite in Cr–Mo–V steel for brake discs, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-018-1638-z
    [14] Jing Guo, Shu-sen Cheng, Han-jie Guo, and  Ya-guang Mei, Novel mechanism for the modification of Al2O3-based inclusions in ultra-low carbon Al-killed steel considering the effects of magnesium and calcium, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-018-1571-1
    [15] Ze-an Zhou, Wan-tang Fu, Zhe Zhu, Bin Li, Zhong-ping Shi, and  Shu-hua Sun, Excellent mechanical properties and resistance to cavitation erosion for an ultra-low carbon CrMnN stainless steel through quenching and partitioning treatment, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-018-1601-z
    [16] Sayed Ghafar Hashemi B. Eghbali, Analysis of the formation conditions and characteristics of interphase and random vanadium precipitation in a low-carbon steel during isothermal heat treatment, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-018-1577-8
    [17] Babak Shahriari, Reza Vafaei, Ehsan Mohammad Sharifi, and  Khosro Farmanesh, Aging behavior of a copper-bearing high-strength low-carbon steel, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-018-1588-5
    [18] Hong-xiang Li, Xin-yu Nie, Zan-bing He, Kang-ning Zhao, Qiang Du, Ji-shan Zhang, and  Lin-zhong Zhuang, Interfacial microstructure and mechanical properties of Ti-6Al-4V/Al7050 joints fabricated using the insert molding method, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-017-1534-y
    [19] Bo-lin He, Lei Xiong, Ming-ming Jiang, Ying-xia Yu, and  Li Li, Surface grain refinement mechanism of SMA490BW steel cross joints by ultrasonic impact treatment, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-017-1421-6
    [20] Jian-long Guo, Yan-ping Bao, and  Min Wang, Cleanliness of Ti-bearing Al-killed ultra-low-carbon steel during different heating processes, Int. J. Miner. Metall. Mater., https://doi.org/10.1007/s12613-017-1529-8
  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Share Article

Article Metrics

Article views(344) PDF downloads(8) Cited by()

Proportional views

Prediction of Charpy V-notch impact energy of low carbon steel by using shallow neural network and deep learning

  • Corresponding author:

    Jian Yang    E-mail: yang_jian@t.shu.edu.cn

  • 1. State Key Laboratory of Advanced Special Steel, School of Materials Science and Engineering, Shanghai University, Shanghai 200444, China
  • 2. State Key Laboratory of Rolling and Automation, Northeastern University, Shenyang 110819, Liaoning, China

Abstract: In the present work, the impact energy prediction model of low carbon steel was investigated based on the industrial data. A three layer neural network, extreme learning machine and deep neural network were compared with different activation functions, structure parameters and training functions. To determine the optimal hyper-parameters of deep neural network, Bayesian optimization was applied. The model with best performance was applied to investigate importance degree of process parameter variables on impact energy of low carbon steel. The results show that deep neural network obtains better prediction results than that of shallow neural network due to the multiple hidden layers improving the learning ability of the model. Among all the models, the Bayesian optimization deep neural network achieves the highest correlation coefficient of 0.9536, lowest mean absolute relative error of 0.0843 and lowest root mean square error of 17.34 J for predicting the impact energy of low carbon steel. Among all the variables, the main factors affecting the impact energy of low carbon steel with final thickness of 7.5 mm are the thickness of the original slab, the thickness of intermediate slab and rough rolling exit temperature on the specific hot rolling production line.

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return