Fuzzy modeling of predictionM _s temperature for martensitic stainless steel

Abstract

A method of fuzzy modeling based on fuzzy clustering and Kalman filtering was proposed for predicting M_s temperature from chemical composition for martensitic stainless steel. The membership degree of each sample was calculated by the fizzy clustering algorithm. Kalman filtering was used to identify the consequent parameters. Only Grade 95 steel are available for training and validation, and the fuzzy model is valid for the following element concentration ranges (wt%): 0.01<C<0.7; 0<Si<1.0; 0.10<Mn<1.25; 11.5<Cr<17.5; 0<Ni<2.5; 0<Mo<1.0. Compared with that of several empirical models reported, the accuracy of the fuzzy model was almost 5 times higher than that of the best empirical model. Furthermore, the compositional dependences of Ms were successfully determined and compared with those of the empirical formulae. It was found that the specific element dependences were a function of the overall composition, something could not easily be found using conventional statistics.

Keywords

fuzzy modeling / prediction model / Ms temperature / alloying element / martensitic stainless steel

References

[1] Payson P, Savage C H. Martensite Reactions in Alloy Steels. Trans ASM., 1944, 33: 261-280.

[2] Carapella L A. Computing A" orM _s (Transformation Temperature on Quenching) form Analysis. Metals Progress, 1944, 46: 108-108.

[3] Rowland E S, Lyle S L. The Application ofM _s Points Case Depth Measurement. Trans, ASM, 1946, 37: 27-47.

[4] Grange R A, Stewart H M. The Temperature Range of Martensite Formation. Trans. AIME, 1946, 167: 467-501.

[5] Steven W, Haynes A G. The Temperature Formation of Martensite and Bainite in Low-alloy Steels Some Effects of Chemical Composition. JISI, 1956, 183(4): 349-359.

[6] Andrews K W. Empirical Formulae for Calculation of Some Transformation Temperatures. JISI, 1965, 203(6): 721-727.

[7] Flnkler H, Schirra M. Transformation Behavior of the High Temperature Martensitic Steels with 8%–14% Chromium. Steel Research, 1996, 67(8): 328-342.

[8] Li Z C, Liu P Y, Wang M G. Computation ofM _s Temperature in Carbon Equivalence Method. Journal of Liaoning Technical University (Natural Science), 1998, 17(3): 293-295.

[9] Liu C, Zhao Z B, Derek O. A New Empirical Formula for the Calculation ofM _s Temperatures in Pure Iron and Super-low Alloy. Materials Processing Technology, 2001, 113: 556-562.

[10] Nakagawa H, Miyazaki T. Experimental Examination of the Validity of the Model for PredictingM _s Temperature for Martensitic Stainless Steels. Tetsu-to-Hagana, 1999, 85(3): 49-54.

[11] Kung C Y, Rayment J J. An Examination of the Validity of Existing Empirical Formulae for the Calculation of MsTemperature. Metall. Trans., 1982, 130A(2): 328-331.

[12] Zhong L, Liu L S, Yuan J L. The Application of Neural Network in Lifetime Prediction of Concrete. Journal of Wuhan University of Technology—Materials Science Edition, 2002, 17(1): 79-81.

[13] Qian H C, Xia B C, Li S Z, Wang F G. Fuzzy Neural Network Modeling of Material Properties. Journal of Materials Processing Technology, 2002, 122: 196-200.

[14] Su J H, Dong Q M, Liu P, Li H J, Kang B X. The Rapidly Solidified Aging Copper Alloy by BP Neural Network. Journal of Wuhan University of Technology—Materials Science Edition, 2003, 18(4): 50-54.

[15] Wang S T. Application Programming of Fuzzy System and Fuzzy Neural Network, 1998 Shanghai: Shanghai Scientific and Technical Literature Press. 6-7.

[16] Al-Saba Tawfuq, El-Amin Ibrahim. Artificial Neural Networks as Applied To long-time Demand Forecasting. Artifical Intelligence in Engineering, 1999, 13: 189-197.

[17] Takagi T., Sugeno. Fuzzy Identification of System and its Application to Modeling and Control. IEEE Trans. Syst. Man Cybernet., 1985, 15(1): 116-132.

[18] Lin H G, Fu D Z. Austenite Transformation Curve of Steels, 1988 Beijing: China Mechanical Industry Press. 435-874.

[19] Yang K G. Isothermal Transformation Curve of Steels, 1981 Harbin: China Heilongjiang People Press. 89-370.