Basalt fiber-reinforced concrete (BFRC), reinforced with chopped basalt fibers having lengths ranging from 12 to 30 mm and diameters ranging from 0.013 to 0.020 mm, is a sustainable construction material with enhanced strength and durability; however, its complex nonlinear behavior makes accurate prediction and optimal mix design challenging. This study proposes an integrated machine learning framework combining evolutionary optimization, multi-objective optimization, and explainable artificial intelligence (XAI) for BFRC strength prediction and mix design optimization. The proposed framework further incorporates a graphical user interface (GUI) deployment to enhance practical usability and support engineering decision-making. Random Forest, Gradient Boosting Regressor, and XGBoost models were optimized using Genetic Algorithms, Particle Swarm Optimization, and Differential Evolution, while NSGA-II was employed to identify optimal trade-offs between compressive strength and splitting tensile strength. SHAP analysis was applied to interpret the influence of key mix parameters on strength prediction. The optimized models achieved high prediction accuracy, with R² values of 0.88 for compressive strength and 0.95 for splitting tensile strength, demonstrating the effectiveness of the proposed framework. The developed GUI provides a practical decision-support tool for sustainable and performance-oriented BFRC mix design.

Basalt Fiber-Reinforced Concrete (BFRC) is a sustainable building material characterized by high strength, crack resistance, and durability. Because its mechanical properties involve intricate, nonlinear interactions among binder constituents, aggregates, water content, and fiber characteristics, traditional empirical mix design methods are often inefficient. Current literature largely focuses on single-property prediction using individual machine learning (ML) models, with limited integration of evolutionary optimization, Explainable AI (XAI), and practical deployment tools—leaving significant gaps in real-world applicability. To address this, this research presents a novel, integrated evolutionary ML framework for the accurate prediction and multi-objective optimization of BFRC Compressive Strength (CS) and Splitting Tensile Strength (STS). Three regression models—Random Forest, Gradient Boosting Regressor, and XGBoost—were developed and coupled with evolutionary optimization algorithms, including Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Differential Evolution (DE). The trade-offs between CS and STS were further evaluated through multi-objective optimization using the NSGA-II algorithm and Pareto front analysis. Additionally, SHapley Additive exPlanations (SHAP) were utilized to enhance model interpretability, revealing that cement content, water content, and fiber volume fraction are the predominant contributors to strength. The optimized models achieved high predictive accuracy, yielding $R^2$ values of 0.88 for CS and 0.95 for STS, alongside low prediction errors (Root Mean Square Error < 2.80 MPa and 0.29 MPa; Mean Absolute Error < 2.10 MPa and 0.19 MPa, respectively). Finally, a graphical user interface (GUI) was developed to enable real-time prediction and mix optimization, providing a practical decision-support tool for performance-driven and sustainable BFRC mix design.