Interpretación y desempeño en modelos predictivos de la resistencia del hormigón

Autores/as

DOI:

https://doi.org/10.47842/juts.v8i1.85

Palabras clave:

resistencia del hormigón, análisis de componentes principales, equilibrio precisión–interpretabilidad, random forest, modelización predictiva

Resumen

La modelización predictiva de la resistencia del hormigón plantea un desafío fundamental: equilibrar la precisión obtenida mediante modelos no lineales complejos y la transparencia ofrecida por los enfoques lineales. Este estudio aborda dicha problemática mediante el análisis de una base de datos que contiene 1.030 mezclas de hormigón. En primer lugar, se aplicó el análisis de componentes principales (ACP) para extraer factores latentes interpretables. Posteriormente, se desarrollaron y compararon tres modelos: (1) regresión por componentes principales utilizando cinco componentes que explican el 87,3% de la varianza, (2) regresión lineal múltiple con las variables originales, y (3) un modelo no lineal Random Forest. Los resultados muestran que el modelo Random Forest alcanza la mayor precisión (R² = 0,88, RMSE = 5,52 MPa), cuantificando la relevancia de las interacciones no lineales. El modelo de regresión por componentes principales (R² = 0,48, RMSE = 11,58 MPa), aunque menos preciso, aporta una comprensión sobre la estructura de los datos, una reducción dimensional significativa y mitiga la multicolinealidad presente en el modelo de regresión lineal múltiple (R² = 0,63, RMSE = 9,80 MPa). Los resultados indican la necesidad de incorporar técnicas de interpretabilidad para profundizar en la comprensión de las variables determinantes de la resistencia del hormigón.

Descargas

Los datos de descargas todavía no están disponibles.

Biografía del autor/a

Sandro Martini, Universidade São Judas Tadeu

Universidade São Judas Tadeu, Brasil

Lara Kühl Teles, Instituto Tecnológico de Aeronáutica

Bacharel em Física pela Universidade de São Paulo (1994), Mestre (1997), Doutora (2001) e Pós-Doutora (2004) em Ciências. Realizou estágio durante o seu doutoramento na Friedrich-Schiller Universität de Jena, Alemanha. Em 2005 ingressou no Instituto Tecnológico de Aeronáutica (ITA), sendo atualmente professora titular da instituição. Ocupou diversos cargos de coordenação e de diretoria. É consultora ad hoc de diversas agências de fomento e realiza arbitragem para diversos periódicos científicos internacionais indexados. Líder no Grupo de Materiais Semicondutores e Nanotecnologia do ITA (GMSN - www.gmsn.ita.br). Possui 87 artigos científicos publicados em periódicos internacionais e mais de 3300 citações.

Citas

ABRAMS, D. A. Design of concrete mixtures. Chicago: Structural Materials Research Laboratory, Lewis Institute, Bulletin 1, 1919. 20 p. Disponível https://archive.org/details/designofconcrete00abrarich. Acesso em: 6 nov. 2025.

AHMAD, A.; AHMAD, W.; ASLAM, F.; JOYKLAD, P. Compressive strength prediction of fly ash-based geopolymer concrete via advanced machine learning techniques. Case Studies in Construction Materials, v. 16, e00840, 2022. ISSN 2214-5095. DOI: https://doi.org/10.1016/j.cscm.2021.e00840 DOI: https://doi.org/10.1016/j.cscm.2021.e00840

AHMAD, M. R.; FERNÁNDEZ-JIMENEZ, A.; CHEN, B.; LENG, Z.; DAI, J.-G. Low-carbon cementitious materials: Scale-up potential, environmental impact and barriers. Construction and Building Materials, v. 455, art. 139087, 2024. DOI: https://doi.org/10.1016/j.conbuildmat.2024.139087. DOI: https://doi.org/10.1016/j.conbuildmat.2024.139087

ASTERIS, Panagiotis G.; SKENTOU, Athanasia D.; BARDHAN, Abidhan; SAMUI, Pijush; PILAKOUTAS, Kypros. Predicting concrete compressive strength using hybrid ensembling of surrogate machine learning models. Cement and Concrete Research, v. 145, art. 106449, 2021. ISSN 0008-8846. DOI: 10.1016/j.cemconres.2021.106449. DOI: https://doi.org/10.1016/j.cemconres.2021.106449. DOI: https://doi.org/10.1016/j.cemconres.2021.106449

CHI, L. et al. Machine learning prediction of compressive strength of concrete with resistivity modification. Materials Today Communications, v. 36, art. 106470, 2023. DOI: https://doi.org/10.1016/j.mtcomm.2023.106470. DOI: https://doi.org/10.1016/j.mtcomm.2023.106470

ELHISHI, S.; ELASHRY, A. M.; EL-METWALLY, S. Unboxing machine learning models for concrete strength prediction using XAI. Scientific Reports, v. 13, art. 19892, 2023. DOI: https://doi.org/10.1038/s41598-023-47169-7. DOI: https://doi.org/10.1038/s41598-023-47169-7

JACKSON, D. A. Stopping rules in principal components analysis: a comparison of heuristical and statistical approaches. Ecology, v. 74, n. 8, p. 2204-2214, 1993. DOI: https://doi.org/10.2307/1939574. DOI: https://doi.org/10.2307/1939574

JIANG, Y.; ZHANG, J.; ZUO, W.; XU, G.; YUAN, C.; WANG, L.; DU, Z.; LU, Y.; SHE, W. Prediction of time-dependent concrete mechanical properties based on advanced deep learning models considering complex variables. Case Studies in Construction Materials, v. 21, art. e03629, 2024. DOI: https://doi.org/10.1016/j.cscm.2024.e03629. DOI: https://doi.org/10.1016/j.cscm.2024.e03629

JOLLIFFE, I. T. Principal Component Analysis. 2. ed. New York: Springer Science+Business Media, 2002. DOI: https://doi.org/10.1007/b98835. DOI: https://doi.org/10.1007/b98835

JUENGER, M. C. G.; WINNEFELD, F.; PROVIS, J. L.; IDEKER, J. H. Advances in alternative cementitious binders. Cement and Concrete Research, v. 41, n. 12, p. 1232-1243, 2011. DOI: https://doi.org/10.1016/j.cemconres.2010.11.012. DOI: https://doi.org/10.1016/j.cemconres.2010.11.012

KHODAPARASTI, M.; ALIJAMAAT, A.; POURAMINIAN, M. Prediction of the concrete compressive strength using improved random forest algorithm. Journal of Building Pathology and Rehabilitation, v. 8, art. 92, 2023. DOI: https://doi.org/10.1007/s41024-023-00337-8. DOI: https://doi.org/10.1007/s41024-023-00337-8

KOBAKA, J. Principal Component Analysis as a Statistical Tool for Concrete Mix Design. Materials, v. 14, n. 10, art. 2668, 2021. DOI: https://doi.org/10.3390/ma14102668. DOI: https://doi.org/10.3390/ma14102668

MEHTA, P. K. Concrete technology for sustainable development. Concrete International, Detroit, v. 21, n. 11, p. 47-53, nov. 1999.

MONTEIRO, H.; MOURA, B.; SOARES, N. Advancements in nano-enabled cement and concrete: Innovative properties and environmental implications. Journal of Building Engineering, v. 56, art. 104736, 2022. DOI: https://doi.org/10.1016/j.jobe.2022.104736. DOI: https://doi.org/10.1016/j.jobe.2022.104736

MURDOCH, W. J.; SINGH, C.; KUMBIER, K.; ABBASI-ASL, R.; YU, B. Definitions, methods, and applications in interpretable machine learning. Proceedings of the National Academy of Sciences of the United States of America, v. 116, n. 44, p. 22071-22080, 2019. DOI: https://doi.org/10.1073/pnas.1900654116. DOI: https://doi.org/10.1073/pnas.1900654116

NIKOOPAYAN TAK, M. S.; FENG, Y.; MAHGOUB, M. Advanced Machine Learning Techniques for Predicting Concrete Compressive Strength. Infrastructures, v. 10, n. 2, art. 26, 2025. DOI: https://doi.org/10.3390/infrastructures10020026. DOI: https://doi.org/10.3390/infrastructures10020026

RIBEIRO, M. T.; SINGH, S.; GUESTRIN, C. "Why Should I Trust You?": Explaining the Predictions of Any Classifier. In: PROCEEDINGS OF THE 22ND ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY & DATA MINING (KDD '16), 2016, San Francisco. Anais... New York: ACM, 2016. p. 1135-1144. DOI: https://doi.org/10.1145/2939672.2939778. DOI: https://doi.org/10.1145/2939672.2939778

SNELLINGS, R.; SURANENI, P.; SKIBSTED, J. Future and emerging supplementary cementitious materials. Cement and Concrete Research, v. 171, art. 107199, 2023. DOI: https://doi.org/10.1016/j.cemconres.2023.107199. DOI: https://doi.org/10.1016/j.cemconres.2023.107199

UNIVERSITY OF CALIFORNIA, IRVINE. UCI Machine Learning Repository. Irvine: School of Information and Computer Sciences, 1987-. Disponível em: http://archive.ics.uci.edu/ . Acesso em: 6 nov. 2025.

WAN, Z.; XU, Y.; ŠAVIJA, B. On the use of machine learning models for prediction of compressive strength of concrete: influence of dimensionality reduction on the model performance. Materials, v. 14, n. 4, art. 713, 2021. DOI: https://doi.org/10.3390/ma14040713. DOI: https://doi.org/10.3390/ma14040713

YEH, I. C. Modeling of strength of high-performance concrete using artificial neural networks. Cement and Concrete Research, v. 28, n. 12, p. 1797-1808, 1998. DOI: https://doi.org/10.1016/S0008-8846(98)00165-3. DOI: https://doi.org/10.1016/S0008-8846(98)00165-3

ZHANG, Y.; REN, W.; CHEN, Y.; MI, Y.; LEI, J.; SUN, L. Predicting the compressive strength of high-performance concrete using an interpretable machine learning model. Scientific Reports, v. 14, art. 28346, 2024. DOI: https://doi.org/10.1038/s41598-024-79502-z. DOI: https://doi.org/10.1038/s41598-024-79502-z

ZHU, J.; FANG, S.; YANG, Z.; QIN, Y.; CHEN, H. Prediction of concrete strength based on random forest and gradient boosting machine. In: IEEE 3RD INTERNATIONAL CONFERENCE ON POWER, ELECTRONICS AND COMPUTER APPLICATIONS (ICPECA), 2023, Shenyang, China. Anais... [S. l.]: IEEE, 2023. p. 306-312. DOI: https://doi.org/10.1109/ICPECA56706.2023.10075839. DOI: https://doi.org/10.1109/ICPECA56706.2023.10075839

Publicado

2026-07-06

Cómo citar

MARTINI, S.; TELES, L. K. Interpretación y desempeño en modelos predictivos de la resistencia del hormigón. Journal of Urban Technology and Sustainability, [S. l.], v. 8, n. 1, p. 1–19, 2026. DOI: 10.47842/juts.v8i1.85. Disponível em: https://journaluts.emnuvens.com.br/journaluts/article/view/85. Acesso em: 13 sep. 2026.

Número

Sección

Artigos