A method for detecting cause-effects in data from complex processes
When models are developed to aid the decision making in the operation of industrial processes, lack of understanding of the underlying mechanisms can make a first-principles modeling approach infeasible. An alternative is to develop a black-box model on the basis of historical data, and neural networks can be used for this purpose to cope with nonlinearities. Since numerous factors may influence the variables to be modeled, and all potential inputs cannot be considered, one may instead solely focus on occasions where the (input or output) variables exhibit larger changes. The paper describes a modeling method by which historical data can be interpreted with respect to changes in key variables, yielding a model that is well suited for analysis of how changes in the input variables affect the outputs.
KeywordsBlast Furnace Feedforward Neural Network Coke Property Coke Strength Coke Strength After Reaction
Unable to display preview. Download preview PDF.
- Omori, Y. (ed.) (1987) Blast Furnace Phenomena and Modelling, The Iron and Steel Institute of Japan, Elsevier, LondonGoogle Scholar
- Petterson, F. and Saxén, H. (2003) A hybrid algorithm for weight and connectivity optimization in feedforward neural networks. In: Pearson, D. et al. (eds) Artificial Neural Nets and Genetic Algorithms. Springer, Wien, pp. 47–52Google Scholar
- Raipala, K. (2003) On Hearth Phenomena and Hot Metal Carbon Content in Blast Furnace, Doctoral Thesis, Helsinki University of Technology, HelsinkiGoogle Scholar
- Husslage, W. (2004) Dynamic distributions — Sulphur transfer and flow in a high temperature packed coke bed, Doctoral Thesis, Delft University of Technology, Febodruk, Enschede.Google Scholar