Splitting Tensile Strength Prediction Using Machine Learning Based Optimization Algorithms

Session

Civil Engineering, Infrastructure and Environment

Description

The prediction of splitting tensile strength, a crucial mechanical attribute determining structural performance, is an integral part of assessing the feasibility of recycled aggregates for construction. Traditional techniques for evaluating the splitting tensile strength of recycled aggregates rely on advanced and time-consuming laboratory testing, which may be costly and inefficient for large scale applications. This work proposes machine learning-based algorithms for forecasting the performance of splitting tensile strength. In this research, 257 measurements were acquired from a previous study containing input variables affecting split tensile strength. Three methods were used to build different predictive models, i.e., support vector regression, XG boost, and random forest. The performance indices of various models were evaluated using metrics like MAE, RMSE, MAPE, and MASE to measure the models' accuracy and reliability. This research indicates that Random forest algorithms outperform other models with RMSE value of 1.76. The implementation of proposed models improves the reliability of predictions, allowing researchers to make informed decisions about incorporating recycled materials in sustainable construction practices, thereby contributing to the reduction of environmental impacts in the construction sector.

Keywords:

Recycled Aggregate Concrete (RAC), Machine Learning, Random Forest, XG Boost, Splitting Tensile Strength, Support Vector Regression

Proceedings Editor

Edmond Hajrizi

ISBN

978-9951-550-95-6

Location

UBT Lipjan, Kosovo

Start Date

28-10-2023 8:00 AM

End Date

29-10-2023 6:00 PM

DOI

10.33107/ubt-ic.2023.372

This document is currently not available here.

Share

COinS
 
Oct 28th, 8:00 AM Oct 29th, 6:00 PM

Splitting Tensile Strength Prediction Using Machine Learning Based Optimization Algorithms

UBT Lipjan, Kosovo

The prediction of splitting tensile strength, a crucial mechanical attribute determining structural performance, is an integral part of assessing the feasibility of recycled aggregates for construction. Traditional techniques for evaluating the splitting tensile strength of recycled aggregates rely on advanced and time-consuming laboratory testing, which may be costly and inefficient for large scale applications. This work proposes machine learning-based algorithms for forecasting the performance of splitting tensile strength. In this research, 257 measurements were acquired from a previous study containing input variables affecting split tensile strength. Three methods were used to build different predictive models, i.e., support vector regression, XG boost, and random forest. The performance indices of various models were evaluated using metrics like MAE, RMSE, MAPE, and MASE to measure the models' accuracy and reliability. This research indicates that Random forest algorithms outperform other models with RMSE value of 1.76. The implementation of proposed models improves the reliability of predictions, allowing researchers to make informed decisions about incorporating recycled materials in sustainable construction practices, thereby contributing to the reduction of environmental impacts in the construction sector.