Advertisement

Application of Neural Networks in Pulp Production

  • D. Obradovic
  • G. Deco
  • H. Furumoto
  • C. Fricke
Conference paper
Part of the Informatik aktuell book series (INFORMAT)

Abstract

This paper presents an application of neural networks in control of the pulp production process. The pulp is produced in a chemical reaction where wood is dissolved by the so called “cooking solution.” One of the main quality indicators of the produced pulp is the permanganate number. The latter strongly depends on the reaction time, temperature, and pressure profiles and is not available for measurement during the production process. The pressure and temperature evolution during the process is usually regulated by process automation system. Consequently, the process time is left as the only remaining variable that has major impact on the process outcome. If the reaction is stopped too soon or too late, the conesponding pulp quality is different from the desired one. Therefore, in order to achieve the desired permanganate number it is essential to make accurate prediction of the reaction time. A neural network model of the process which is used to predict the necessary time of reaction is described herein. This neural model was tested at Cellulose do Caima in Portugal where it proved significantly more accurate (30%) than the previously used analytical model. Furthermore, it was implemented in the commercially available automation system “Teleperm M” produced by Siemens AG.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Frieden B.,1983, “Probability, statistical, optics and data testing”, Springer series in Information Sciences, Springer-Verlag, Heidelberg, New York.zbMATHGoogle Scholar
  2. Rumelhart D., Hinton G. and Willians J., 1986, “Learning internal representations by error propagation”, Parallel Distributed Processing: Explorations in the Microstructure of Cognition. Vol. 1 Foundations. MIT Press, Cambridge, Mass.Google Scholar
  3. Kushner H. and Clark D., 1978, “Stochastic Approximation Methods for Constrained and Unconstrained Systems,”Springer-Verlag, New York.Google Scholar
  4. Ljung L., 1987, “System Identification: Theory for the User,” Prentice-Hall, Englewood Cliffs, New Jersey.zbMATHGoogle Scholar
  5. Furumoto H. and Gerdemann U., 1993, “Ausformen von Prozeßführungsrdumen mit Fuzzy Logik: am Beispiel der Zellstoffkocher in Cellulose do Calma, Portugal,” Siemens Symposium FUZZY 93, Munchen.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1993

Authors and Affiliations

  • D. Obradovic
    • 1
  • G. Deco
    • 1
  • H. Furumoto
    • 2
  • C. Fricke
    • 3
  1. 1.Corporate Research and DevelopmentSiemens AGGermany
  2. 2.Bereich Anlagetechnik, Zellstoff und PapierSiemens AGGermany
  3. 3.Siemens StockholmSiemens AGGermany

Personalised recommendations