Machine learning (ML) and deep learning (DL) in sustainable concrete construction: review, trend and gap analyses

Abstract

The pressing need for sustainability in the concrete construction industry necessitates innovative strategies that minimize environmental impact while maintaining performance and resilience. However, the critical review and bibliometric analysis of machine learning (ML) and deep learning (DL) for sustainable concrete technology are currently unexplored. This study conducts a bibliometric and thematic analysis of 502 publications to evaluate the contributions of ML and DL in the advancement of sustainable concrete materials. Research indicates that ML(84.5%) and DL (15.5%) applications enhance mix design optimization, particularly concerning supplementary cementitious materials, property prediction, structural health monitoring, and lifecycle sustainability assessment. AI-driven methodologies facilitate accurate forecasting, efficient material selection, and predictive maintenance, in accordance with the United Nations Sustainable Development Goals (UN-SDGs). The analysis identifies key challenges, including inadequate data quality, the absence of standardized datasets, issues with model interpretability, and the gap between experimental advancements and practical implementation. Overcoming these obstacles requires the implementation of explainable AI frameworks, hybrid physics-data models, and interdisciplinary collaboration among engineers, data scientists, and policymakers. This study synthesizes current trends, identifies gaps, and highlights opportunities to establish a framework for the integration of AI into concrete construction, promoting the potential transformation of sustainable materials.