Machine learning approaches for real-time prediction of compressive strength in self-compacting concrete

Abstract

Self-compacting concrete (SCC) has transformed civil engineering by efficiently filling formwork without mechanical consolidation, enhancing construction efficiency, and durability, and reducing labor needs. Accurate prediction of compressive strength (C_S), a crucial mechanical property, is essential for optimal results. The complex nature of SCC mixtures has led researchers to explore modern days tool like machine learning and artificial intelligence. This study assesses six machine learning techniques (MLTs) by coupling long-established AI algorithms like artificial neural network (ANN), adaptive neuro-fuzzy inference system (ANFIS), and extreme learning machine (ELM) with nature-inspired optimization algorithms like moth flame optimization algorithm (MOFA) and wild horse optimizer (WHO). Addressing gaps in input parameter consistency, dataset standardization, and model comparison, the results demonstrate high accuracy in C_S prediction for all six models, with ELM tuned with MFOA consistently outperforming others in various metrics. Visual representations validate model effectiveness, suggesting potential benefits such as improved quality control, reduced costs, and enhanced safety. This research contributes to MLT applications in construction materials, highlighting ELM–MOFA as a preferred model for C_S prediction in SCC.