Computers and Concrete

Techno-Press (테크노프레스)
권호 표지
DOI QR코드

DOI QR Code

A sensitivity analysis of machine learning models on fire-induced spalling of concrete: Revealing the impact of data manipulation on accuracy and explainability

  • Mohammad K. al-Bashiti (School of Civil and Environmental Engineering & Earth Sciences (SCEEES), Clemson University) ;
  • M.Z. Naser (School of Civil and Environmental Engineering & Earth Sciences (SCEEES), Clemson University)
  • Received : 2023.09.07
  • Accepted : 2023.11.14
  • Published : 2024.04.25
⟨ Previous Next ⟩

Abstract

Using an extensive database, a sensitivity analysis across fifteen machine learning (ML) classifiers was conducted to evaluate the impact of various data manipulation techniques, evaluation metrics, and explainability tools. The results of this sensitivity analysis reveal that the examined models can achieve an accuracy ranging from 72-93% in predicting the fire-induced spalling of concrete and denote the light gradient boosting machine, extreme gradient boosting, and random forest algorithms as the best-performing models. Among such models, the six key factors influencing spalling were maximum exposure temperature, heating rate, compressive strength of concrete, moisture content, silica fume content, and the quantity of polypropylene fiber. Our analysis also documents some conflicting results observed with the deep learning model. As such, this study highlights the necessity of selecting suitable models and carefully evaluating the presence of possible outcome biases.

Keywords

References

  1. Abo Sabah, S.H., Zainal, N.L., Muhamad Bunnori, N., Megat Johari, M.A. and Hassan, M.H. (2019), "Interfacial behavior between normal substrate and green ultra-high-performance fiber-reinforced concrete under elevated temperatures", Struct. Concrete, 20(6), 1896-1908. https://doi.org/10.1002/suco.201900152.
  2. Chen, T. and Guestrin, C. (2016), "XGBoost: A scalable tree boosting system", Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, August.
  3. Cortes, C. and Vapnik, V. (1995), "Support-vector networks", Mach. Learn., 20(3), 273-297. https://doi.org/10.1007/BF00994018.
  4. Cox, D.R. (1958), "The regression analysis of binary sequences", J. Royal Stat. Soc.: Series B Methodol., 20(2), 215-232. https://doi.org/10.1111/j.2517-6161.1958.tb00292.x.
  5. Domingos, P. and Pazzani, M. (1996), "Beyond independence: Conditions for the optimality of the simple bayesian classifer", Proceedings of the 13th International Conference on Machine Learning, Bari, Italy, July.
  6. Duda, R.O., Hart, P.E. and Stork, D.G. (2001), Pattern Classification, 2nd Edition, John Wiley & Sons, Hoboken, NJ, USA.
  7. Dwaikat, M.B. and Kodur, V.K.R. (2010), "Fire induced spalling in high strength concrete beams", Fire Technol., 46(1), 251-274. https://doi.org/10.1007/s10694-009-0088-6.
  8. Fisher, R.A. (1936), "The use of multiple measurements in taxonomic problems", Ann. Eugen., 7(2), 179-188. https://doi.org/10.1111/j.1469-1809.1936.tb02137.x.
  9. Fix, E. (1985), Discriminatory Analysis: Nonparametric Discrimination, Consistency Properties, USAF School of Aviation Medicine, Dayton, OH, USA.
  10. Freund, Y. and Schapire, R.E. (1997), "A decision-theoretic generalization of on-line learning and an application to boosting", J. Comput. Syst. Sci., 55(1), 119-139. https://doi.org/10.1006/jcss.1997.1504.
  11. Geurts, P., Ernst, D. and Wehenkel, L. (2006), "Extremely randomized trees", Mach. Learn., 63(1), 3-42. https://doi.org/10.1007/s10994-006-6226-1.
  12. Ibrahimbegovic, A., Boulkertous, A., Davenne, L., Muhasilovic, M. and Pokrklic, A. (2010), "On modeling of fire resistance tests on concrete and reinforced-concrete structures", Comput. Concrete, 7(4), 285. https://doi.org/10.12989/cac.2010.7.4.285.
  13. Kanema, M., Pliya, P., Noumowe, A. and Gallias, J.L. (2011), "Spalling, thermal, and hydrous behavior of ordinary and high-strength concrete subjected to elevated temperature", Am. Soc. Civil Eng., 23(7), 921-930. https://doi.org/10.1061/(ASCE)MT.1943-5533.0000272.
  14. Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W. and Liu, T.Y. (2017), "LightGBM: A highly efficient gradient boosting decision tree", Adv. Neural Inf. Process. Syst., 30, 1-9.
  15. Khoury, G.A. (2000), "Effect of fire on concrete and concrete structures", Prog. Struct. Eng. Mater., 2(4), 429-447. https://doi.org/10.1002/pse.51.
  16. Khoury, G.A. (2015), "Passive fire protection of concrete structures", ICE Proc. Struct. Build., 161(3), 135-145. http://doi.org/10.1680/stbu.2008.161.3.135.
  17. Liu, J.C. and Zhang, Z. (2020), "A machine learning approach to predict explosive spalling of heated concrete", Arch. Civil Mech. Eng., 20(4), 1-25. https://doi.org/10.1007/s43452-020-00135-w.
  18. McCulloch, W.S. and Pitts, W. (1943), "A logical calculus of the ideas immanent in nervous activity", Bull. Math. Biol., 5, 115-133. https://doi.org/10.1007/BF02478259.
  19. McKinney, J. and Ali, F. (2014), "Artificial neural networks for the spalling classification & failure prediction times of high strength concrete colunms", J. Struct. Fire Eng., 5(3), 203-214. https://doi.org/10.1260/2040-2317.5.3.203.
  20. Mindeguia, J.C., Carre, H., Pimienta, P. and La Borderie, C. (2015), "Experimental discussion on the mechanisms behind the fire spalling of concrete", Fire Mater., 39(7), 619-635. https://doi.org/10.1002/fam.2254.
  21. Naser, M.Z. (2023), Machine Learning for Civil and Environmental Engineers: A Practical Approach to Data-Driven Analysis, Explainability, and Causality, John Wiley & Sons, Hoboken, NJ, USA.
  22. Naser, M.Z. and Kodur, V.K. (2022), "Explainable machine learning using real, synthetic and augmented fire tests to predict fire resistance and spalling of RC columns", Eng. Struct., 253, 113824. https://doi.org/10.1016/j.engstruct.2021.113824.
  23. Naser, M.Z., al-Bashiti, M.K. and Naser, A.Z. (2023), "SPINEX: Similarity-based predictions and explainable neighbors exploration for regression and classification tasks in machine learning", arXiv preprint arXiv:2306.01029. https://doi.org/10.48550/arXiv.2306.01029.
  24. Ozawa, M. and Morimoto, H. (2014), "Effects of various fibres on high-temperature spalling in high-performance concrete", Constr. Build. Mater., 71, 83-92. https://doi.org/10.1016/j.conbuildmat.2014.07.068.
  25. Panev, Y., Kotsovinos, P., Deeny, S. and Flint, G. (2021), "The use of machine learning for the prediction of fire resistance of composite shallow floor systems", Fire Technol., 57(6), 3079-3100. https://doi.org/10.1007/s10694-021-01108-y.
  26. Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A.V. and Gulin, A. (2018), "CatBoost: Unbiased boosting with categorical features", Adv. Neural Inf. Process. Syst., 31, 1.
  27. Quinlan, J.R. (1986), "Induction of decision trees", Mach. Learn., 1(1), 81-106. https://doi.org/10.1007/BF00116251.
  28. Saberian, M., Shi, L., Sidiq, A., Li, J., Setunge, S. and Li, C.Q. (2019), "Recycled concrete aggregate mixed with crumb rubber under elevated temperature", Constr. Build. Mater., 222, 119-129. https://doi.org/10.1016/j.conbuildmat.2019.06.133.
  29. Seitllari, A. and Naser, M.Z. (2019), "Leveraging artificial intelligence to assess explosive spalling in fire-exposed RC columns", Comput. Concrete, 24(3), 271-282. https://doi.org/10.12989/cac.2019.24.3.271.
  30. Scikit-learn (2023), sklearn.model_selection.GridSearchCV - scikit-learn 1.3.0 documentation; https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html
  31. Tapeh, A.T.G. and Naser, M.Z. (2022), "Artificial intelligence, machine learning, and deep learning in structural engineering: A scientometrics review of trends and best practices", Arch. Comput. Method. Eng., 30(1), 115-159. https://doi.org/10.1007/s11831-022-09793-w.
  32. Thai, H.T. (2022), "Machine learning for structural engineering: A state-of-the-art review", Struct., 38, 448-491. https://doi.org/10.1016/j.istruc.2022.02.003.
  33. LightGBM (2023), Welcome to LightGBM's documentation! - LightGBM 3.3.5 documentation; https://lightgbm.readthedocs.io/en/v3.3.5/index.html
  34. Zhang, H.L. and Davie, C.T. (2013), "A numerical investigation of the influence of pore pressures and thermally induced stresses for spalling of concrete exposed to elevated temperatures", Fire Saf. J., 59, 102-110. https://doi.org/10.1016/j.firesaf.2013.03.019.
  35. Zhao, J., Zheng, J.J., Peng, G.F. and van Breugel, K. (2014), "A meso-level investigation into the explosive spalling mechanism of high-performance concrete under fire exposure", Cement Concrete Res., 65, 64-75. https://doi.org/10.1016/j.cemconres.2014.07.010.
  36. Breiman, L. (2001), "Random forests", Mach. Learn., 45(1), 5-32. https://doi.org/10.1023/A:1010933404324.