Blast furnace analysis with neural networks
Nowadays blast furnace operation is supervised by extensive measurements and controlled accordingly. Characteristic indications concerning process quality are given by the analysis of the radial temperature profile in the upper part of the furnace. Optimising this temperature distribution would lead to considerable savings of input material. To achieve an optimisation, quantitative relations between furnace parameters are needed. As those relationships are unknown, a process model can be provided using neural networks and fuzzy methods. In this paper we show the application of fuzzy clustering and neural networks to classify temperature profiles and to build a model of the interdependence betwenn process operation parameters and the resulting temperature profiles. These investigations have been carried out in a plant of a German steel producer.
Keywordsblast furnace process analysis fuzzy clustering multilayer perceptron identification of temperature profiles
Unable to display preview. Download preview PDF.
- [Bezdek 1981]Bezdek, J. C.: Pattern Recognition with Fuzzy Objective Function Algorithms. Plenum Press, New York London, 1981.Google Scholar
- [Bulsari, Saxen 1995]Bulsari, A., Saxen, B.: Classification of blast furnace probe temperatures using neural networks. Steel Research 66, 1995, No. 6, pp. 231–236.Google Scholar
- [Omori 1987]Omori, Y. (ed.): Blast Furnace Phenomena and Modelling. The Iron and Steel Institute of Japan. Elsevier, London, 1987.Google Scholar
- [Windham 1981]Windham, M.P.: Cluster Validity for Fuzzy Clustering Algorithms. Fuzzy Sets and Systems 5, 1981, pp. 177–185.Google Scholar