Answers are grounded in this paper and cite their sources.
Ultra-high-performance concrete (UHPC) has been increasingly used in advanced infrastructure because of its high strength and reliable durability. The performance and construction quality of UHPC are governed by the coupled effects of mixture composition and manufacturing methods, which complicates multi-objective design and structural assessment. Compared with conventional laboratory testing and empirical methods, artificial intelligence (AI) and data-driven approaches provide efficient tools for optimising UHPC mixture design and supporting decision-making throughout its production and construction. This paper reviews recent advances in AI-assisted UHPC material design, structural performance optimisation, and intelligent construction. At the material level, data-driven methods are reviewed for low-carbon mixture design, fresh property prediction, and durability assessment, highlighting the role of AI in multi-objective optimisation of UHPC. At the structural level, AI-assisted approaches for the interfacial behaviour of UHPC composites are discussed, including UHPC-normal concrete, UHPC-fibre-reinforced polymer, and UHPC-steel, with emphasis on data-driven modelling strategies. AI-assisted evaluations of structural performance and load-bearing capacity of UHPC members are also presented. Emerging applications, including automated design, quality inspection, and structural monitoring, are examined to illustrate the trend toward intelligent UHPC construction. Eventually, the evolving role of AI in UHPC design and construction is critically discussed to identify current limitations and future research directions. Although existing studies have applied AI to UHPC mixture optimisation and structural behaviour analysis, practical implementation remains constrained by fragmented datasets and insufficient validation. Future research should develop standardised databases and physics-informed models to support closed-loop workflows in UHPC engineering practice.
No takes yet. Share an insight, caveat, or question.
Xia et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: