Machine Learning no processamento do nióbio: aplicação da regressão Random Forest para análise de dados experimentais
Machine Learning in niobium processing: application of Random Forest regression for experimental data analysis
João Gabriel Niquini Cordeiro; Cássia Ribeiro Souza; Janúbia Cristina Bragança da Silva Amaral; Sônia Denise Ferreira Rocha; Pedro Henrique Alves Campos
Resumo
Com os avanços tecnológicos na área computacional, o Machine Learning (ML), ou em português “Aprendizado de Máquina”, emergiu como ferramenta poderosa para aplicações na ciência de dados, destacando-se a sua capacidade de fazer previsões baseadas em análises avançadas de dados por meio da identificação de padrões e relações implícitas. Nesse sentido, o objetivo deste trabalho é avaliar o seu desempenho de previsão no campo experimental a partir de dados hidrometalúrgicos referentes ao nióbio. Nesta perspectiva, foram feitos dois estudos de casos distintos: o primeiro a respeito do equilíbrio do ácido nióbico em soluções aquosas, e o segundo envolvendo a técnica de sub-molten salt e lixiviação alcalina de concentrados de pirocloro. O ML foi aplicado através do regressor Random Forest (RF). Os resultados mostram que para os dois estudos de caso, o erro gerado pelo modelo de RF foi inferior ao previsto pelas metodologias tradicionais, o que corrobora o benefício do uso de técnicas de ML em análises experimentais.
Palavras-chave
Abstract
With technological advancements in computational fields, Machine Learning (ML) has emerged as a powerful tool for data science applications, particularly for its ability to make predictions based on advanced data analyses through the identification of patterns and implicit relationships. This study aims to evaluate the predictive performance of ML in the experimental context using hydrometallurgical data related to niobium. Two distinct case studies were conducted: the first on niobic acid equilibrium in aqueous solutions, and the second on the sub-molten salt method and leaching of pyrochlore concentrates. ML was applied using the Random Forest (RF) regressor. The results demonstrate that, in both case studies, the error generated by the RF model was lower than that predicted by traditional methodologies, supporting the advantages of ML techniques in experimental analyses.
Keywords
Referências
1 Kubat M. An introduction to machine learning. 1st ed. Berlin: Springer; 2017. http://doi.org/10.1007/978-3-319- 63913-0.
2 McCoy J, Auret L. Machine learning applications in minerals processing: a review. Minerals Engineering. 2019;132:95-109. http://doi.org/10.1016/j.mineng.2018.12.004.
3 Ahmadzadeh F, Lundberg J. Remaining useful life prediction of grinding mill liners using an artificial neural network. Minerals Engineering. 2013;53:1-8. http://doi.org/10.1016/j.mineng.2013.05.026.
4 Karimi M, Dehghani A, Nezamalhosseini A, Talebi S. Prediction of hydrocyclone performance using artificial neural networks. Journal of the Southern African Institute of Mining and Metallurgy. 2010 [acesso em 16 jan. 2025];110(5):207-212. Disponível em: http://www.scielo.org.za/scielo.php?script=sci_arttext&p id=S2225-62532010000500003
5 Jahedsaravani A, Marhaban MH, Massinaei M. Prediction of the metallurgical performances of a batch flotation system by image analysis and neural networks. Minerals Engineering. 2014;69:137-145. http://doi.org/10.1016/j. mineng.2014.08.003.
6 Breiman L, Friedman J, Olshen RA, Stone CJ. Classification and regression trees. 1st ed. London: Chapman & Hall/ CRC; 1984. http://doi.org/10.1201/9781315139470.
7 Breiman L. Random Forests. Machine Learning. 2001;45(1):5-32. http://doi.org/10.1023/A:1010933404324. 8 Kettaneh N, Berglund A, Wold S. PCA and PLS with very large data sets. Computational Statistics & Data Analysis. 2005;48(1):69-85. http://doi.org/10.1016/j.csda.2003.11.027.
9 Cortes C, Vapnik V. Support-vector networks. Machine Learning. 1995;20(3):273-297. http://doi.org/10.1007/ BF00994018.
10 LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;521(7553):436-444. http://doi.org/10.1038/nature14539.
11 Schmidhuber J. Deep learning in neural networks: an overview. Neural Networks. 2015;61:85-117. http://doi. org/10.1016/j.neunet.2014.09.003.
12 Mohanty S, Nanda R, Panigrahi P. Introduction to data science: a Python approach to concepts, techniques and applications. 1st ed. New York: Springer; 2019.
13 Deblonde GJP, Chagnes A, Bélair S, Cote G. Solubility of niobium(V) and tantalum(V) under mild alkaline conditions. Hydrometallurgy. 2015;156:99-106. http://doi.org/10.1016/j.hydromet.2015.05.015.
14 Shikika A, Sethurajan M, Muvundja F, Mugumaoderha MC, Gaydardzhiev S. A review on extractive metallurgy of tantalum and niobium. Hydrometallurgy. 2020;198:105496. http://doi.org/10.1016/j.hydromet.2020.105496.
15 Wang X, Zheng S, Xu H, Zhang Y. Leaching of niobium and tantalum from a low-grade ore using a KOH roastwater leach system. Hydrometallurgy. 2009;98(3–4):219-223. http://doi.org/10.1016/j.hydromet.2009.05.002.
16 Ayanda OS, Adekola FA. A review of niobium-tantalum separation in hydrometallurgy. Journal of Minerals & Materials Characterization & Engineering. 2011;10(3):245-256. http://doi.org/10.4236/jmmce.2011.103016.
17 Montgomery DC, Rumsby GL. Estatística aplicada e probabilidade para engenheiros. 6. ed. Rio de Janeiro: LTC; 2014.
Submetido em:
16/01/2025
Aceito em:
29/05/2025
Facebook
Google+
Twitter
LinkedIn
Mendeley
StumbleUpon
CiteULike
Reddit
Email