Prediction of Compressive Strength and Modulus of Elasticity of Concrete using Machine Learning Models
Abstract
This paper presents machine learning (ML) models for high fidelity prediction of compressive strength and modulus of elasticity (MOE) of concrete in relation to primary attributes of its mixture design. Two comprehensive databases, consisting of over 1000 and 500 data-records consolidated from technical publications, were used for training and testing the ML models that included random forests (RF), support vector machine (SVM) and multilayer perceptron artificial neural network (MLP-ANN). The metrics used for evaluation of prediction performance included five different statistical parameters and composite performance index (CPI). Results show that the RF model consistently outperforms the other two ML models in terms of prediction accuracy. Overall, machine learning is a very powerful and efficient tool for prediction of concrete properties as well as for the optimization of its mixture design to meet a set of desired performance criteria.
Recommended Citation
T. Han et al., "Prediction of Compressive Strength and Modulus of Elasticity of Concrete using Machine Learning Models," Proceedings of Materials Science and Technology (2019, Portland, OR), pp. 604 - 611, Materials Science and Technology, Oct 2019.
The definitive version is available at https://doi.org/10.7449/2019/MST_2019_604_611
Meeting Name
Materials Science and Technology 2019, MS & T 2019 (2019: Sep. 29-Oct. 3, Portland, OR)
Department(s)
Civil, Architectural and Environmental Engineering
Second Department
Electrical and Computer Engineering
Third Department
Materials Science and Engineering
Research Center/Lab(s)
Re-Cast Tier1 University Transportation Center
Second Research Center/Lab
Center for Research in Energy and Environment (CREE)
Keywords and Phrases
Artificial neural network; Compressive strength; Machine learning; Modulus of elasticity; Random forest; Support vector machine
International Standard Book Number (ISBN)
978-087339770-4
Document Type
Article - Conference proceedings
Document Version
Citation
File Type
text
Language(s)
English
Rights
© 2019 Materials Science and Technology, All rights reserved.
Publication Date
01 Oct 2019
Comments
Computational tasks were conducted in the Materials Research Center of Missouri S&T. The first and fourth author would like to acknowledge funding provided by the Leonard Wood Institute (LWI). The second author would like to acknowledge funding provided by the RECAST University Transportation Center (at Missouri S&T) and Missouri Department of Transportation (MoDOT). The third and last authors would like to acknowledge funding provided by the National Science Foundation (NSF; CMMI: 1661609).