Mining Data from a Metallurgical Process by a Novel Neural Network Pruning Method

  • Henrik Saxén
  • Frank Pettersson
  • Matias Waller
Part of the Lecture Notes in Computer Science book series (LNCS, volume 4432)


Many metallurgical processes are complex and due to hostile environment it is difficult to carry out reliable measurement of their internal state, but the demands on high productivity and consideration of environmental issues require that the processes still be strictly controlled. Due to the complexity and non-ideality of the processes, it is often not feasible to develop mechanistic models. An alternative is to use neural networks as black-box models, built on historical process data. The selection of relevant inputs and appropriate network structure are still problematic issues. The present work addresses these two problems in the modeling of the hot metal silicon content in the blast furnace. An algorithm is applied to find relevant inputs and their time lags, as well as a proper network size, by pruning a large network. The resulting models exhibit good prediction capabilities and the inputs and time lags detected are in good agreement with practical metallurgical knowledge.


Blast Furnace Hide Node Silicon Content Feedforward Neural Network Relevant Input 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Cybenko, G.: Approximations by superpositions of sigmoidal function. Math. Contr. Sign. 2, 303–314 (1989)zbMATHCrossRefMathSciNetGoogle Scholar
  2. 2.
    Principe, J.C., Euliano, N.R., Lefebvre, W.C.: Neural and adaptive systems: Fundamentals through simulations. John Wiley & Sons, New York (1999)Google Scholar
  3. 3.
    Frean, M.: The Upstart Algorithm. A Method for Constructing and Training Feed-forward Neural Networks. Neural Computation 2, 198–209 (1991)CrossRefGoogle Scholar
  4. 4.
    Fahlman, S.E., Lebiere, C.: The Cascade-Correlation Learning Architecture. In: Touretzky, D.S. (ed.) Adv. Neural Inf. Proc. Syst. 2, pp. 524–532. Morgan Kaufmann, San Francisco (1990)Google Scholar
  5. 5.
    Le Chun, Y., Denker, J.S., Solla, S.A.: Optimal Brain Damage. In: Touretzky, D.S. (ed.) Adv. Neural Inf. Proc. Syst. 2, pp. 598–605. Morgan Kaufmann, San Francisco (1990)Google Scholar
  6. 6.
    Sridhar, D.V., Bartlett, E.B., Seagrave, R.C.: Information theoretic subset selection for neural networks. Comput. Chem. Engng. 22, 613–626 (1998)CrossRefGoogle Scholar
  7. 7.
    Saxén, H., Pettersson, F.: Method for the selection of inputs and structure of feedforward neural networks. Comput. Chem. Engng. 30, 1038–1045 (2006)CrossRefGoogle Scholar
  8. 8.
    Hinnelä, J., Saxén, H., Pettersson, F.: Modeling of the blast furnace burden distribution by evolving neural networks. Ind. Engng Chem. Res. 42, 2314–2323 (2003)CrossRefGoogle Scholar
  9. 9.
    Haykin, S.: Kalman filtering and neural networks. Wiley, New York (2001)Google Scholar
  10. 10.
    Omori, Y. (ed.): Blast Furnace Phenomena and Modelling. Elsevier, London (1987)Google Scholar
  11. 11.
    Phadke, M.S., Wu, S.M.: Identification of Multiinput - Multioutput Transfer Function and Noise Model of a Blast Furnace from Closed-Loop Data. IEEE Trans. Aut. Contr. 19, 944–951 (1974)CrossRefGoogle Scholar
  12. 12.
    Unbehauen, H., Diekmann, K.: Application of MIMO Identification to a Blast Furnace. In: IFAC Identification and System Parameter Estimation, pp. 180–185 (1982)Google Scholar
  13. 13.
    Saxén, H.: Short Term Prediction of Silicon Content in Pig Iron. Can. Met. Quart. 33, 319–326 (1994)Google Scholar
  14. 14.
    Saxén, H., Östermark, R.: State Realization with Exogenous Variables - A Test on Blast Furnace Data. Europ. J. Oper. Res. 89, 34–52 (1996)zbMATHGoogle Scholar
  15. 15.
    Chen, J.: A Predictive System for Blast Furnaces by Integrating a Neural Network with Qualitative Analysis. Engng. Appl. AI 14, 77–85 (2001)zbMATHGoogle Scholar
  16. 16.
    Waller, M., Saxén, H.: On the Development of Predictive Models with Applications to a Metallurgical Process. Ind. Eng. Chem. Res. 39, 982–988 (2000)CrossRefGoogle Scholar
  17. 17.
    Waller, M., Saxén, H.: Application of Nonlinear Time Series Analysis to the Prediction of Silicon Content of Pig Iron. ISIJ Int. 42, 316–318 (2002)CrossRefGoogle Scholar
  18. 18.
    Bhattacarya, T.: Prediction of silicon content in blast furnace hot metal using Partial Least Squares (PLS). ISIJ Int. 45, 1943–1945 (2005)CrossRefGoogle Scholar
  19. 19.
    Gao, C.H., Qian, J.X.: Time-dependent fractal characteristics on time series of silicon content in hot metal of blast furnace. ISIJ Int. 45, 1269–1271 (2005)CrossRefGoogle Scholar

Copyright information

© Springer Berlin Heidelberg 2007

Authors and Affiliations

  • Henrik Saxén
    • 1
  • Frank Pettersson
    • 1
  • Matias Waller
    • 2
  1. 1.Heat Engineering Lab., Åbo Akademi University, Biskopsg. 8, 20500 ÅboFinland
  2. 2.Åland Polytechnic, PB 1010, AX-22111 Mariehamn, ÅlandFinland

Personalised recommendations