Prediction of mechanical properties of steel fibre-reinforced self-compacting concrete by machine learning algorithms

Cui, Tianyi, Kulasegaram, Sivakumar ORCID: https://orcid.org/0000-0002-9841-1339 and Li, Haijiang ORCID: https://orcid.org/0000-0001-6326-8133 2023. Prediction of mechanical properties of steel fibre-reinforced self-compacting concrete by machine learning algorithms. Presented at: International RILEM Conference on Synergising expertise towards sustainability and robustness of CBMs and concrete structures, 14-16 June 2023. SynerCrete 2023, RILEM Bookseries. , vol.44 Cham, Switzerland: Springer, pp. 703-711. 10.1007/978-3-031-33187-9_65

Preview

PDF - Published Version Download (743kB) | Preview

Abstract

With the development of big data processing technology and the continuous improvement of computer operation ability, machine learning has achieved remarkable results in recent years. Applying machine learning to solve engineering problems is gaining more attention from researchers. Steel fibre-reinforced self-compacting concrete (SFRSCC) is a new type of composite material prepared by combining the advantages of the high fluidity of self-compacting concrete (SCC) and the high toughness of steel fibre-reinforced concrete. However, the performance of SFRSCC is influenced by many factors such as water-binder ratio, mineral powder content and steel fibre content. This study aims to predict the mechanical properties of SFRSCC mixes based on datasets collected from the literature. In the presented work, the machine learning algorithms are employed to investigate the effect of SCC compositions and steel fibre on the performance of SFRSCC. The models used for the prediction are support vector regression (SVR) and artificial neural network (ANN). In both models, input variables are set to be water to binder ratio, sand to aggregate ratio, maximum size of coarse aggregate, amount of other mix components (e.g., superplasticizers, limestone powder, fly ash), volume fraction and aspect ratio of steel fibre, and curing age. The output variables are flexural strength and compressive strength of SFRSCC specimens. The performances of machine learning models are evaluated by comparing the predicted results with experimental results obtained from the literature. Furthermore, a comparative study is performed to select the best-proposed model with better accuracy.

Item Type:	Conference or Workshop Item (Paper)
Date Type:	Published Online
Status:	Published
Schools:	Engineering
Publisher:	Springer
ISBN:	978-3-031-33186-2
Date of First Compliant Deposit:	21 June 2023
Date of Acceptance:	26 February 2023
Last Modified:	23 Jul 2023 01:30
URI:	https://orca.cardiff.ac.uk/id/eprint/160486

Actions (repository staff only)

Edit Item