Coke Index Analysis Based on Principal Component Analysis and Decision Tree Mining

Abstract:

Current analysis in the relations between blast furnace production index and coke index is still using the traditional statistical analysis method，but it involves too many coke quality evaluation indexes and there are some overlap between the indexes. According to this situation, this paper puts forward a new method based on principal component analysis and decision tree mining to analyze the relations between blast furnace production index and coke index . The materials of blast furnace production mainly include ore, coke and coal, in which the coke quality index have the biggest influence on the blast furnace production index. It has profound meaning to analyze the relation between coke index and blast furnace production index to evaluate Coke quality indicators reasonably[1] and improve the blast furnace production index. Current analysis in the relations between blast furnace production index and coke index is still using the traditional statistical analysis method[2]，but it involves too many coke quality evaluation indexes and there are some overlap between the indexes. According to this situation, this paper puts forward a new method based on principal component analysis and decision-tree-based data-mining to analyze the relations between blast furnace production index and coke index. On the one hand this method can get few representative indexes from so many evaluation indexes by principal component analysis; on the other hand, decision-tree-based data-mining on the coke representative index based on the principal component analysis can get accurately quantitative relation between blast furnace production index and coke index.

• Periodical: Applied Mechanics and Materials (Volumes 48-49)
• Pages: 318-322
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.48-49.318
• Online since: February 2011
• Authors: Hong Wei Guo, Bu Xin Su, Jian Chang, Jian Liang Zhang, Wei Chao Cao
• Keywords: Blast Furnace, Coke, Data Mining (DM), Decision Tree, Principal Component Analysis (PCA)

References

[1] ShiYong Zhou, in: Discussion of positive effect of the simulation of current blast furnace coke index [J]. Iron & Steel，2000, 25(2)：1~3.

[2] QingGui Zhao, Keng Wu, in: Influence of coke quality on high PCI BF [J]. Metal World, 200 (1): 9~12.

[3] ShiYong Zhou, in: Inquiry on coke quality indexes [J]. Ironmaking, 2002, 19(6): 22~25.

[4] Liang He, in: Principal Components Analysis in SPSS [J]. Journal of Shanxi Agricultural University: Social Science Edition, 2007, 6(5): 20~21.

[5] MARGARETH Dunham, in: Data Mining Introductory and Advanced Topics [M] 2005: 79-88.

[6] Breiman L, Friedman J, Olshen R, in: Classification and regression trees[M], California: Wadsworth Belement, (1984).

[7] ScottC D, W illettRM, Nowak R D, in: Classification or regression Trees[J]. IEEE Acoustics Speech and Signal Processing, 2003, 4(6): 153-156.

[8] BittencourtH R, Clarke R T. in: Use of classification and regression trees (CART) to classify remotely-sensed digital images[J]. IEEE Geoscience and Remote Sensing Symposium, 2003, 7(6): 3751-37.

DOI: 10.1109/igarss.2003.1295258