Advancing Structural Safety: Genetic Programming Approaches to Steel Fiber-Reinforced Concrete (SFRC) Blast Response Prediction

Abstract

Steel Fiber-Reinforced Concrete (SFRC) has emerged as a preferred material for blast-resistant structures due to its exceptional mechanical properties and energy absorption capabilities. This study introduces a machine learning-based framework to predict the maximum displacement of SFRC structural members under blast loading. Using 107 experimental data points, split into 70% for training and 15% each for validation and testing, Gene Expression Programming (GEP) and Multi-Expression Programming (MEP) were applied. The GEP model exhibited superior predictive performance with R-values of 0.964 (training), 0.968 (validation), and 0.960 (testing), while the MEP model achieved reasonable accuracy with R-values of 0.922, 0.905, and 0.948, respectively. Additionally, parametric analysis revealed the influence of fiber properties on SFRC behavior. This approach not only simplifies predictive modeling but also enhances its reliability, offering valuable insights for optimizing SFRC design under extreme conditions and contributing to the advancement of resilient structural systems.