Comparative Assessment Using Machine Learning Algorithms for Ultimate Bond Strength Estimations - Research

Aug 5, 2026

Upload by Levi

Corrosion-driven degradation significantly reduces bond strength, creating challenges for infrastructure repair and maintenance planning. This study compares machine learning approaches—SVR, XG Boost, and random forest—for predicting ultimate bond behavior between corroded reinforcement and concrete. A dataset of 218 corroded samples compiled from prior studies provides input and output variables for model training and evaluation. Model accuracy is assessed using MAE, RMSE, MAPE, and MASE. Random forest achieves the most reliable performance, with RMSE of 1.26, supporting more efficient assessment of corroded reinforced concrete structures.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

Abstract

1 Introduction

Corrosion effects on bond strength

Limits of traditional prediction methods

Machine learning approaches in civil engineering

Research objective and methods

University for Business and Technology in Kosovo UBT Knowledge Center | UBT International Conference | 2023 UBT International Conference | | --- | --- | | Oct 28th, 8:00 AM-Oct 29th, 6:00 PM
Comparative Assessment Using Machine Learning Algorithms for Ultimate Bond Strength Estimations
Lukesh Veloso de Parida
Department of Civil Engineering, Shiv Nadar Institution of Eminence, Uttar Pradesh, Dadri, India, [Leonardo.kuhn@estudante.ufla.br](Leonardo.kuhn@estudante.ufla.br)
Sumedha Melo de Moharana
Federal University of Lavras, fabricio.fontenelle@estudante. ufla. br
Sourav Kumar Giri
Federal University of Lavras, [tulio.guimaraes2@estudante.ufla.br](tulio.guimaraes2@estudante.ufla.br)
Follow this and additional works at: [https://knowledgecenter.ubt-uni.net/conference](https://knowledgecenter.ubt-uni.net/conference)
Part of the Engineering Commons | | Recommended Citation Parida, Lukesh Veloso de; Moharana, Sumedha Melo de; and Giri, Sourav Kumar, "Comparative Assessment Using Machine Learning Algorithms for Ultimate Bond Strength Estimations" (2023) . UBT International Conference. 36. [https://knowledgecenter.ubt-uni.net/conference/IC/civil/36](https://knowledgecenter.ubt-uni.net/conference/IC/civil/36) This Event is brought to you for free and open access by the Publication and Journals at UBT Knowledge Center. It has been accepted for inclusion in UBT International Conference by an authorized administrator of UBT Knowledge Center. For more information, please contact [knowledge.center@ubt-uni.net](knowledge.center@ubt-uni.net). Comparative Assessment Using Machine Learning Algorithms for Ultimate Bond Strength Estimations Lukesh Parida 1, Sumedha Moharana 1, Sourav Kumar Giri2 1Department of Civil Engineering, Shiv Nadar Institution of Eminence, Uttar Pradesh, Dadri, India, 201314 2School of Computer Engineering, KIIT Deemed to be University, Bhubaneswar, Patia, India,756001 [Presenting Author/Corresponding Author:lp617@snu.edu.in](Presenting Author/Corresponding Author:lp617@snu.edu.in) Abstract. Corrosion-induced bond strength reduction is a critical problem in infrastructure maintenance and repair. This study investigates several machine learning techniques, i.e., SVR, XG Boost, and random forest, for predicting the ultimate bond behavior between corroded reinforcement and concrete. In this study author employed 218 datasets of corroded samples collected from past studies containing input and output parameters used for predicting the models. The model's performance was evaluated and compared using various performance metrics, i.e., MAE, RMSE, MAPE, and MASE. The results show that random forest algorithms can reliably estimate ultimate bond strength with an RMSE value of 1.26 over SVR and XG Boost models. This research helps inefficient structural evaluations and maintenance planning for corroded reinforced concrete buildings. Keywords: Corrosion, Bond Strength, Random Forest, Support Vector Regression, XG Boost, Machine Learning 1 Introduction The durability and robustness of reinforced concrete structures are critical for infrastructure safety and lifespan. Reinforced concrete, commonly employed in construction, is structurally dependent on the bond strength between the reinforcing bars and concrete matrix [1-3] . Environmental conditions such as moisture and strong chemicals can cause reinforcement corrosion over time, decreasing bond strength [4] . This corrosion-induced deterioration offers a severe barrier to civil infrastructure maintenance and serviceability. In previous studies, the bond strength exhibits complicated patterns of changing corrosion levels [5] . For assessing and rehabilitating degraded buildings, accurate prediction of the maximum bond strength is critical. Traditional bond strength prediction methods rely on empirical formulations based on laboratory testing and limited in situ data, frequently failing to represent the complex and num

FAQs

What problem does the study address?

The study addresses corrosion-induced reductions in bond strength between reinforcement and concrete, which hinder infrastructure maintenance and repair.

Which machine learning algorithms are compared?

The study compares SVR, XG Boost, and random forest for predicting ultimate bond behavior.

How is model performance evaluated?

Performance is evaluated using MAE, RMSE, MAPE, and MASE to compare predictive quality across models.

Aug 5, 2026

Upload by Levi