Anode Temperature Classification of Liquid Metal in an Electric Arc Furnace using K-Nearest Neighbors

• Tiago Ferreira Pontifícia Universidade Católica do Rio de Janeiro
• Karla Figueiredo Universidade do Estado do Rio de Janeiro
• Marley Vellasco Pontifícia Universidade Católica do Rio de Janeiro

DOI: https://doi.org/10.5753/eniac.2019.9316

Abstract

Metallurgical and steel industries adopt arc electric furnaces in their processes for the refining of the chemical composition of liquid metals. Therefore, it is necessary to control the anode temperature of these reactors, so that it does not get into excessively high or low ranges, which leads to the loss of the metallic bath or the early wear of the furnace refractory. Given a pilot database of attributes that involve this process, Data Mining and Machine Learning techniques were applied to preprocess the data and create models that solve the anode temperature classification problem at low, normal and high temperatures. The Machine Learning model used is the KNN (k- Nearest Neighbors) algorithm which, based on distance between test data point and training data points, classifies the output variable by majority vote. Three different approaches to creating KNN models are used and the results of each are presented and discussed.

Keywords: k-Nearest Neighbors, Principal Component Analysis, Electrical Arc Furnace, Machine Learning, Classification, Anode Temperature

References

ABM, (1978) “Fusão em Fornos Elétricos - Princípios de Funcionamentos de Fornos Elétricos: Forno a Arco Direto, a Arco Indireto, a Resistência; Forno a Indução e Forno Especiais Processo Duplex e Triplex”. Associação Brasileira de MetaisFundição, 10ª edição.

Bergman, K., Kjellberg, B. (2001) “Arc Furnace Techonology Applied to Smelting Applications”. INFACON IX – 9ª Inter. Ferroalloys Cong., Quebec, Canadá.

Blachnik M., Mączka K., Wieczorek T. (2010) “A Model for Temperature Prediction of Melted Steel in the Electric Arc Furnace (EAF)”. Artifical Intelligence and Soft Computing. ICAISC 2010. vol 6114. Springer, Berlim, Heidelberg.

Chen, C., Liu, Ying, Kumar, M., Qin, J. (2018) “Energy Consumption Modelling Using Deep Learning Technique — A Case Study for EAF”. Elsevier. 51º CIRP Conferência em Sistemas de Manufatura. p. 1063 - 1068.

Da Costa e Silva, A. (2013) Refino dos Aços - Capítulo 7: Forno Elétrico a Arco.

Fernandéz, J. M. M., Cabal, V.A., Montequin, V.R., Balsera, J.V. (2008) “Online estimation of electric arc furnace tap temperature by using fuzzy neural networks”. Elsevier. Engineering Applications of Artificial Intelligence. p. 1001 – 1012.

Goodfellow, I., Bengio, Y., and Courville, A. (2016) Deep learning. MIT press.

Jackson, J. (1991) A User’s Guide to Principal Components, Wiley-Interscience Publication, John Wiley & Sons, Inc.

Haykin, S. (1998) Neural Networks: A Comprehensive Foundation, Prentice Hall, 2ª Edição.

Jolliffe, I. (2002) Principal Component Analysis, Springer.

Jones, R. (2011) “Smelting Applications of DC Arc Furnaces. Fray Inter. Symposium on Metals and Materials Processing in a Clean Environment”. Cancun, México.

Mourão, M. (2019) “PMT 2420 – Metalurgia Geral”, USP, São Paulo.

O’Brien, N. and Ford, M. (2018) “Practical Aspects of DC Furnace Design”. INFACON VIII, Pequim, China.

Parsian, M. (2015) Data Algorithms – Recipes for Scaling Up with Hadoop and Spark, O’Reilly Media.

Rogers, C. and Cramer, C. (1989) “Advanced Process Control for Electric Arc Furnaces”. Carnegie Mellon Research Institute, Pittsburg, Pennsylvania, EUA.

Sampaio, J. and Almeida, S. (2005) “Calcário e dolomito In: Rochas & minerais industriais: usos e especificação”, RJ: CETEM/MCT, Parte II. Cap. 15. p.327-350.