Dr Shgh Woei Ooi, November 2006, steveooi@postech.ac.kr
Computational Metallurgy Group,
Graduate Institute of Ferrous Metallurgy (GIFT),
Pohang University of Science and Technology (POSTECH),
Republic of Korea
To predict the Ac3 temperature of steel as a function of the chemical composition and the heating rate.
Language: | executable for Windows environment |
Product form: | Source code |
MAP_STEEL_AC3 uses neural network analysis to predict the temperature at the completion of austenite formation.
The neural net was trained using 394 of a database of 788 examples constructed using information from Refs [1-7]. The remaining 394 examples were used as `new' experiments to test the trained network. This database is in the file `neural_dataset'. The data are available in MAP_DATA_AUSTENMAT.
An excel file (Input and Copy.xls) containing the chemical composition and heating rate table is included. The user can edit the excel file, copy and paste special (in unformatted text) the composition and heating rate to test.dat file. The user needs to save the file before proceeding. To start the prediction (in the Windows operating system) double click on the file model.exe (the executable program).
Inputs in sequence C, Si, Mn, S, P, Cu, Ni, Cr, Mo, Nb, V, Ti, Al, B, W, As, Sn, Zr, Co, N, O (all in wt%) followed by heating rate in K/s. The following example has three sets of inputs.
0.09 0.17 0.97 0 0 0.019 0.09 9.2 0.96 0.045 0.18 0 0 0 0.97 0 0 0 1.87 0.052 0 5 0.09 0.17 0.97 0 0 0.019 0.09 9.2 0.96 0.045 0.18 0 0 0 0.97 0 0 0 1.87 0.052 0 10 0.09 0.17 0.97 0 0 0.019 0.09 9.2 0.96 0.045 0.18 0 0 0 0.97 0 0 0 1.87 0.052 0 20
The output is the Ac3 temperature in centrigrade, followed by a one sigma modelling uncertainty, Ac3-uncertainty and Ac3+uncertainty. The outputs listed below correspond to the inputs described above.
Ac3 / C Error / C Ac3-Error Ac3+Error 899.464417 19.997770 879.466675 919.462158 911.400269 19.649918 891.750366 931.050171 933.057861 20.260756 912.797119 953.318604
None.
neural network, austenite formation
Download source code (Windows)
MAP originated from a joint project of the National Physical Laboratory and the University of Cambridge.