Machine learning algorithms on self-healing concrete

Abstract

Self-healing concrete (SHC) has emerged as a groundbreaking material in sustainable construction, addressing critical challenges such as crack formation and durability limitations in traditional concrete. Recent research highlights the use of advanced techniques, including bacterial and fungal agents, machine learning (ML) models, and innovative material compositions, to enhance the performance and self-healing capabilities of SHC. Studies have focused on leveraging bacteria-induced calcium carbonate precipitation, particularly using organisms like Bacillus subtilis and Trichoderma reesei, to autonomously seal cracks and improve compressive strength. Machine learning techniques such as Adaptive Boosting (AB), Gradient Boosting (GB), and Random Forest (RF) have been employed to optimize crack repair rate predictions and material design, with models achieving exceptional accuracy metrics (e.g., R² > 0.98). Visualization tools like heatmaps and radar charts reveal insights into aggregate balance, strength recovery, and model performance across evaluation criteria. Despite these advancements, the adoption of SHC faces challenges, including standardization of methods, cost constraints, and scalability for large-scale applications. This review provides a comprehensive understanding of SHC’s potential, emphasizing its role in creating durable, efficient, and environmentally friendly materials for modern infrastructure.