[MAP Logo]

Materials Algorithms Project
Program Library


Program MAP_STEEL_BAKE_HARDENING

  1. Provenance of code.
  2. Purpose of code.
  3. Specification.
  4. Description of program's operation.
  5. References.
  6. Parameter descriptions.
  7. Error indicators.
  8. Accuracy estimate.
  9. Any additional information.
  10. Example of code
  11. Auxiliary routines required.
  12. Keywords.
  13. Download source code.
  14. Links.

Provenance of Source Code

Sourav Das
Materials Science and Metallurgy,
Pembroke Street,
University of Cambridge,
Cambridge, CB2 3QZ.


The neural network program was produced by:
David MacKay,
Cavendish Laboratory,
University of Cambridge,
Madingley Road,
Cambridge, CB3 0HE, U.K.

E-mail: H.K.D.H. Bhadeshia

Added to MAP:August 2007.

Top | Next

Purpose

Neural Network model to understand the complexities of bake hardening effect.

Top | Next | Prev

Specification

Language: C & Fortran
Product form: Windows Executable and Unix source for compilation.

Complete program.

Top | Next | Prev

Description

The model is based on a neural network program called generate44, which was developed by David MacKay and is part of the bigback5 program. The model is constituted of a committee of several individual neural networks. It was trained on a set of experimental data for which the "outputs" are known, and creates a kind of non-linear, multi-parameter "regression" of the outputs versus the inputs. This "regression" has already been produced and the model is delivered ready to perform predictions for steels of any desired composition (within certain specified limits). The source code for the neural network program can be downloaded from David MacKay's website; the executable files only are available from MAP.

The program runs on a unix like operating system, Sun, Linux and Irix and on Windows systems which have DOS. The files for unix are separated compressed into a file called MAP_NN_BH.tar.gz the files for the windows systems are in the zip file MAP_NN_BH.zip  ;The archive file contains the following files which make the model:

README
A brief file with instructions for running the program.
labels.txt
A list of the input variables.
MINMAX
A text file containing the minimum and maximum limits of each input and output variable. This file is used to normalise and unnormalise the input and output data.
test.dat
An input text file containing the input variables used for predictions, together with an example set of data.
result_test.txt
Contains the results you should expect from the example set of data. To test the model is running properly on your computer, use the given 'test.dat' file to do predictions and compare the 'result' file with this file.
model.gen
This is a unix shell file containing the command steps required to run the module. It can be executed by typing sh model.gen  at the command prompt. These shell files run all the programs necessary for normalising the input data, executing the network for each model, unnormalising the output data and combining the results of each model to produce the final committee result.
spec.t1
Created by generate_spec, which contains information about the module and the number of data items being supplied. It is read by the program generate44.
.generate_spec (hidden)
This executable file creates a file called spec.t1, required by generate44.
.randomise (hidden)
This executable file creates a file called norm_test.in, which contains the normalised equivalent of the input data found in test.dat. It requires the MINMAX file
.generate44
This is the executable file for the neural network program. It reads the normalised input data file, norm_test.in (created by normalise) , and uses the weight files in subdirectory c, to find a value for the output. The results are written to the temporary output file _out.
.gencom
This executable file combines the predictions of the different models in the committee and calculates the combined error bar.
.treatout
This executable un-normalise the committee predictions and produces the file 'result'.
result
Contains the final un-normalised committee results for the predicted output.
SUBDIRECTORY c
_w*f
The weights files for the different models.
*.lu
Files containing information for calculating the size of the error bars for the different models.
_c*
Files containing information about the perceived significance value for each model.
_R*
Files containing values for the noise, test error and log predictive error for each model.
SUBDIRECTORY d
outran.x
A normalised output file which was created during the building of the model. It is accessed by generate44 via spec.t1.
SUBDIRECTORY outprdt
com.dat
The normalised output file containing the committee results. It is generated by .gencom.



Top | Next | Prev

References

  1. H. K. D. H. Bhadeshia, Neural Networks in Materials Science, ISIJ International 39 (1999) No. 10, 966 - 979.
  2. D.J.C. MacKay, Mathematical Modelling of Weld Phenomena 3 (1997), eds. H. Cerjak & H.K.D.H. Bhadeshia, Inst. of Materials, London, pp 359.
  3. D.J.C MacKay's website at https://wol.ra.phy.cam.ac.uk/mackay/README.html#Source_code

Top | Next | Prev

Parameters

Input parameters

The inputs to this model are the composition (wt%), annealing and aging temperatures (°C), annealing time (s) and aging time (min), and % prestrain.

Carbon - C_Wt%
wt. %.

Silicon - Si_Wt%
wt. %.

Phosphorus - P_Wt%
wt. %.

Sulpur - S_Wt%
wt. %.

Nitrogen - N_Wt%
wt. %.

Niobium - Nb_Wt%
wt. %.

Manganese - Mn_Wt%
wt. %.

Titanium - Ti_Wt%
wt. %.

Aluminium - Al_Wt%
wt. %.

Prestrain - Prestrain_%
%.

Prestrain Temperature - Prestrain_Temp_C
Centigrade.

Annealing Temperature - Anneal_Temp_C
Centigrade.

Aging Temperature - Aging_Temp_C
Centigrade.

Annealing Time - Annealing_Time_s
Second.

Aging Time - Aging_Time_mins
Minute.

Output parameters

The output is the amount of bake hardening with error bars that depend upon the position in the input space.

Bake hardening - Predicted
MPa.

Top | Next | Prev

Error Indicators

None.

Top | Next | Prev

Accuracy

Each prediction is accompanied with an estimated error that depends upon the position in the input space.

Top | Next | Prev

Further Comments

Top | Next | Prev

Example

1. Download the model

Download and uncompress the appropriate archive file file in a dedicated directory (for example: "neural").
On UNIX systems, this is done by:



On Microsoft systems you will first need to have the unzip or winzip programs installed.

2. Running the program (making predictions)

There are brief instructions in the README file inside each archive file. The unix download first needs to be compiled before you make predictions, this is done by the command:
sh install
Predictions are then made from the test.dat file using the command:
sh model.gen

On Microsoft systems the model is run by the command:
model

3. Program results

The results are written in the "Result" or "model_result.dat" file, as described in the README file.

Top | Next | Prev

Auxiliary Routines

Top | Next | Prev

Keywords

neural networks, ultra low carbon steel, bake hardening

Top | Next | Prev

Download

Download source code for Unix

Download Microsoft Windows Executable

Top | Prev

MAP originated from a joint project of the National Physical Laboratory and the University of Cambridge.