An ensemble machine learning approach for prediction and optimization of modulus of elasticity of recycled aggregate concrete. (30th May 2020)

• Record Type: Journal Article
• Authors: Han, Taihao; Siddique, Ashfia; Khayat, Kamal; Huang, Jie; Kumar, Aditya
• Is Part Of: Construction & building materials. Volume 244(2020)
• Publication Date: 2020-05-30
• Subjects: Recycled concrete aggregate (RCA) -- Modulus of elasticity (MOE) -- Ensemble machine learning -- Random forests -- And voting; Building materials -- Periodicals; 624.18
• DOI: 10.1016/j.conbuildmat.2020.118271
• Languages: English
• ISSNs: 0950-0618

Abstract

Highlights: This study presents the first application of an ensemble machine learning (ML) model to predict the modulus of elasticity (MOE) of recycled aggregate concrete. The ensemble ML model – comprising of random forests (RF) and support vector machine (SVM) – produces accurate predictions of concretes' MOE (RMSE of ≈3.0 GPa). Prediction performance of the ensemble ML model is consistently superior than several standalone ML models. The ensemble ML model is able to develop optimal mixture designs for RCA concretes that satisfy imposed target MOE. Abstract: This paper presents an ensemble machine learning (ML) model for prediction of modulus of elasticity (MOE) of concrete formulated using recycled concrete aggregate (RCA), in relation to features of its mixture design (e.g., physiochemical characteristics of RCA). The ensemble ML model's prediction performance was compared with five commonly-used ML models. It is shown that the ensemble ML model unfailingly produces more accurate predictions compared to standalone models. To demonstrate the ability of the ensemble ML model to go beyond MOE predictions, the model was used to develop optimal mixture designs for RCA concretes that satisfy imposed target MOE.