Combination of thermodynamic knowledge and multilayer feedforward neural networks for accurate prediction of MS temperature in steels

被引:28
作者
Lu, Qi [1 ,2 ]
Liu, Shilong [2 ]
Li, Wei [1 ,2 ]
Jin, Xuejun [1 ,2 ]
机构
[1] Shanghai Jiao Tong Univ, Shanghai Key Lab Mat Laser Proc & Modificat, Shanghai 200240, Peoples R China
[2] Shanghai Jiao Tong Univ, Sch Mat Sci & Engn, Inst Adv Steels & Mat, Shanghai 200240, Peoples R China
基金
中国国家自然科学基金;
关键词
Multilayer feedforward neural networks; M-S temperature; Kernel principal component analysis; Crude estimation property; Thermodynamic; MARTENSITE START TEMPERATURE; EMPIRICAL FORMULAS; KINETICS; PRECIPITATION; CARBONITRIDE; MODEL;
D O I
10.1016/j.matdes.2020.108696
中图分类号
T [工业技术];
学科分类号
08 ;
摘要
The combination of multilayer feedforward neural networks (MLFFNN) and thermodynamic knowledge has excellent advantages to study the complicated phenomena in material science. In the present study, a thermodynamic knowledge-based MLFFNN has been developed to predict the martensite start temperature (M-S) of steels that accounts for the variations in critical driving force (Delta G(c)) and austenitization temperature (T-gamma). This is achieved by integrating two crudely estimated properties (Delta G(c) and T-gamma) in the feature space. Instead of using the original dataset directly, the feature space was reshaped by kernel principal component analysis (KPCA). The genetic algorithm was implemented to find a suitable hyperparameters set that was able to induce a model with high predictive capability. The resulting neural network performs well in the present dataset, and the RMSE in the unseen datasets is 21.52 K. Benchmarking of the MLFFNN predictions against JmatPro-V8 calculations also shows a significant improvement in predictive accuracy. Results indicate KPCA and the integration of crude estimation property (CEP) boost the predictive accuracy of MLFFNN. Besides, the present study also demonstrates the CEP strategy is general enough to be employed in several well-known machine learning methods including support vector regression, decision tree and Gaussian process regression. (C) 2020 The Authors. Published by Elsevier Ltd.
引用
收藏
页数:11
相关论文
共 57 条
[1]  
ANDREWS KW, 1965, J IRON STEEL I, V203, P721
[2]  
[Anonymous], 2007, IN 2007 IEEE C COMP
[3]  
Atkins M., 1980, ATLAS CONTINUOUS COO
[4]  
Boyer A.G., 1977, ATLAS ISOTHERMAL TRA
[5]   A comparison of PCA, KPCA and ICA for dimensionality reduction in support vector machine [J].
Cao, LJ ;
Chua, KS ;
Chong, WK ;
Lee, HP ;
Gu, QM .
NEUROCOMPUTING, 2003, 55 (1-2) :321-336
[6]   Analysis of effect of alloying elements on martensite start temperature of steels [J].
Capdevila, C ;
Caballero, FG ;
de Andrés, CG .
MATERIALS SCIENCE AND TECHNOLOGY, 2003, 19 (05) :581-586
[7]   Determination of Ms temperature in steels:: A Bayesian neural network model [J].
Capdevila, C ;
Caballero, FG ;
De Andrés, CG .
ISIJ INTERNATIONAL, 2002, 42 (08) :894-902
[8]   Predicting glass transition temperatures using neural networks [J].
Cassar, Daniel R. ;
de Carvalho, Andre C. P. L. F. ;
Zanotto, Edgar D. .
ACTA MATERIALIA, 2018, 159 :249-256
[9]   Material structure-property linkages using three-dimensional convolutional neural networks [J].
Cecen, Ahmet ;
Dai, Hanjun ;
Yabansu, Yuksel C. ;
Kalidindi, Surya R. ;
Song, Le .
ACTA MATERIALIA, 2018, 146 :76-84
[10]   The role of occam's razor in knowledge discovery [J].
Domingos, P .
DATA MINING AND KNOWLEDGE DISCOVERY, 1999, 3 (04) :409-425