TY - JOUR
T1 - Predicting the compressive strength of green concretes using Harris hawks optimization-based data-driven methods
AU - Mohammadi Golafshani, Emadaldin
AU - Arashpour, Mehrdad
AU - Behnood, Ali
N1 - Funding Information:
This work was partly funded by the Monash Data Futures Institute (MDFI) grant scheme on “AI and Data Science for Monash Global Challenges". Any opinions, findings, conclusions, and recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of MDFI.
Publisher Copyright:
© 2021 Elsevier Ltd
PY - 2022/2/7
Y1 - 2022/2/7
N2 - This study develops a practical tool for measuring the compressive strength (CS) of concrete using data-driven techniques. To develop predictive models for the CS of concretes containing supplementary cementitious materials, multi-layer neural network (MLNN) and radial basis function neural network (RBFNN) are extended by the Harris hawks optimization (HHO) algorithm. A comprehensive dataset was used containing information on 1374 unique concrete mixture proportions, curing age, and CS. The MLNN was trained using classic, HHO, and hybridized algorithms, while classic and HHO algorithms were deployed for training the RBFNN. Based on statistical metrics, the MLNN model developed by the hybridized training algorithm has the best performance among all developed models. The results indicate that cement, coarse and fine aggregate are the top three variables influencing the CS, while slag, fly ash, and concrete age are the least effective variables.
AB - This study develops a practical tool for measuring the compressive strength (CS) of concrete using data-driven techniques. To develop predictive models for the CS of concretes containing supplementary cementitious materials, multi-layer neural network (MLNN) and radial basis function neural network (RBFNN) are extended by the Harris hawks optimization (HHO) algorithm. A comprehensive dataset was used containing information on 1374 unique concrete mixture proportions, curing age, and CS. The MLNN was trained using classic, HHO, and hybridized algorithms, while classic and HHO algorithms were deployed for training the RBFNN. Based on statistical metrics, the MLNN model developed by the hybridized training algorithm has the best performance among all developed models. The results indicate that cement, coarse and fine aggregate are the top three variables influencing the CS, while slag, fly ash, and concrete age are the least effective variables.
KW - Concrete compressive strength
KW - Harris hawk optimization
KW - Multi-layer neural network
KW - Radial basis function neural network
UR - https://www.scopus.com/pages/publications/85120746484
U2 - 10.1016/j.conbuildmat.2021.125944
DO - 10.1016/j.conbuildmat.2021.125944
M3 - Article
AN - SCOPUS:85120746484
SN - 0950-0618
VL - 318
JO - Construction and Building Materials
JF - Construction and Building Materials
M1 - 125944
ER -