A predictive model of sinter chemical composition was developed to predict R (CaO/SiO2), TFe and SiO2 (CaO is calculated from R and SiO2). Based on the backpropagation neural network algorithm, it used the predictive result as a precondition. An expert system was designed to assist in controlling the sinter chemical composition by estimating the change of all the relative chemical components and providing the necessary adjustment. After the system commenced, the hit ratio of the predictive model was consistently over 90% and the goal of controlling chemical composition was achieved.
Get full access to this article
View all access options for this article.
References
1.
YeK. W.: Min. Metall. Eng., 1996, (Suppl. 2), 25.
2.
ShenB. X.: ‘Research on prediction of sinter chemical component’, Master’s thesis, Central South University, Changsha, China, 2002.
3.
FanX. H.WangH. D.: ‘Mathematical model and artificial intelligence of sintering process’; 2002, Changsha, Central South University Press.
4.
YanP. F.: ‘Analysis and application of artificial neural network model’; 1991, Hefei, Anhui Education Press.
