A Machine Learning-based Structural Load Estimation Model for Shear-Critical RC Beams and Slabs - Manuscript Draft

Aug 2, 2026

Upload by Finn

This manuscript develops a machine learning model to estimate load levels for shear-critical reinforced concrete (RC) beams and slabs by leveraging multifractal features extracted from characteristic crack patterns. Multifractal analysis is performed on a dataset of 508 crack images, using features derived from singularity and generalized dimension spectra as predictors. Among four evaluated learning models, extreme gradient boosting achieves the strongest performance, with a predicted-to-true ratio mean of 1.04 and coefficient of variation of 0.27. Shapley additive explanations identify fractal and information-related dimensions, correlation dimension, and a singularity-spectrum area feature as key drivers, supporting automated, inspector-oriented damage assessment decisions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages84

Reading Time3 h 32 min

AccessFree to download

Abstract

Data and multifractal feature extraction

Machine learning model development and evaluation

Feature importance with SHAP

Practical implications for RC damage assessment

Construction and Building Materials A Machine Learning-based Structural Load Estimation Model for Shear-Critical RC Beams and Slabs using Multifractal Analysis --Manuscript Draft-- | Manuscript Number: | CONBUILDMAT-D-22-09224R1 | | --- | --- | | Article Type: | Research Paper | | Keywords: | Multifractal analysis; load-level assessment; beams and slabs; Machine Learning; score analysis | | Corresponding Author: | Jack Banahene Osei, PhD
Kwame Nkrumah University of Science and Technology
Kumasi, GHANA | | First Author: | Jack Banahene Osei, PhD | | Order of Authors: | Jack Banahene Osei, PhD | | | Mark Adom-Asamoah, PhD | | | Jones Owusu Twumasi, PhD | | | Peter Andras, PhD | | | Hexin Zhang | | Abstract: | This paper presents a machine learning model for load-level estimation for shearcritical reinforced concrete (RC) beams and slabs using multifractal features of their characteristic crack patterns to automate and provide well-informed decisions for RC damage assessment. Multifractal analysis was conducted on a database of 508 images, of which critical features were extracted from the singularity and generalized dimension spectra. These features are used as predictors for the load-level estimation model. The extreme gradient boosting algorithm yielded the best performance among the four machine learning models considered. The mean of the predicted-to-true ratio for the developed model was 1.04 with a coefficient of variation of 0.27. Upon applying Shapley additive explanations, the fractal dimension, information dimension, correlation dimension and the area under the left branch of the singularity spectrum were the critical features influencing load-level estimation. The proposed model can be useful to RC building inspectors. | | Suggested Reviewers: | Salvatore Salamone, PhD
Professor, The University of Texas at Austin
[salamone@utexas.edu](salamone@utexas.edu)
He is an expert in employing fractal analysis for damage assessment of reinforced concrete members | | | Gregory Miller, PhD
Professor, University of Washington
[gmiller@u.washington.edu](gmiller@u.washington.edu)
His research revolves around the application of fractal analysis for damage assessment of reinforced concrete members | | | Shirley Dyke, PhD Professor, Purdue University[sdyke@purdue.edu](sdyke@purdue.edu)
His research works revolve around the application of computer-vision techniques for bridge inspection | | | Christian Koch, PhD
Professor, University of Nottingham
[christian.koch@nottingham.ac.uk](christian.koch@nottingham.ac.uk)
He has considerable knowledge on the application of computer vision techniques for concrete and civil infrastructure | | | Koichi Maekawa, PhD
Professor, The University of Tokyo | Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation | | [maekawa@concrete.t.u-tokyo.ac.jp](maekawa@concrete.t.u-tokyo.ac.jp)
He has used several machine learning algorithms for estimating in-service fatigue life assessment of road bridge decks | | --- | --- | Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation Cover Letter Department of Civil Engineering, Jack Banahene Osei, Kwame Nkrumah University of Science and Technology, Tel: +2330550766883 Kumasi, Ghana Email: [j](jobanahene.coe@knust.edu.gh)[obanahene.coe@knust.edu.gh](jobanahene.coe@knust.edu.gh) Dear Editors Editors-in-Chief Construction and Building Materials Manuscript: A Machine Learning-based Structural Load Estimation Model for Shear-Critical RC Beamsand Slabs using Multifractal Analysis. Authors: Jack Banahene Osei, Mark Adom-Asamoah, Jones Owusu-Twumasi, Peter Andras and Hexin Zhang I have pleasure in submitting to you our paper entitled “A Machine Learning-based Structural Load Estimation Model for Shear-Critical RC Beams and Slabs using Multifractal Analysis” for review and possible publication in the Journal of Construction and Building

FAQs

What problem does the model address for RC beams and slabs?

The model estimates load levels for shear-critical reinforced concrete (RC) beams and slabs using multifractal features from crack patterns, enabling more informed decisions for damage assessment.

How are the predictors for load-level estimation obtained?

Multifractal analysis is conducted on 508 crack images, and critical features are extracted from singularity and generalized dimension spectra to serve as predictors.

Which machine learning method performed best and what were the results?

Extreme gradient boosting produced the best performance among four models. The predicted-to-true ratio averaged 1.04 with a coefficient of variation of 0.27.

Aug 2, 2026

Upload by Finn