Obtaining Suitable Outcomes by Combining the Radial Basis Function with A Novel Optimizer for Predicting Unconfined Compressive Strength

This study introduces a novel approach for predicting the Unconfined Compressive Strength (UCS) of soil materials used in embankment fills for transportation construction by employing Radial Basis Function (RBF) modeling frameworks. Traditionally, UCS determination has relied on time-consuming, costly laboratory tests, such as the Proctor test (ASTM D698/AASHTO). In contrast, the proposed method establishes reliable and accurate correlations between UCS and a range of fundamental soil properties using data-driven RBF models. A diverse dataset covering various soil types and stabilization conditions was rigorously used for model training, validation, and testing based on previously published experimental results. This comprehensive evaluation improves prediction robustness and demonstrates the potential of the proposed approach as an efficient alternative to conventional laboratory-based UCS assessment methods in transportation geotechnical engineering. To improve UCS prediction accuracy, this study hybridized the Gorilla Troops Optimizer (GTO) and Fox Optimization (FOX) algorithms with the RBF model, resulting in two hybrid frameworks, RBGT and RBFO, which were further analyzed alongside the standalone RBF model. Among the developed models, the RBFO framework demonstrated superior predictive performance, achieving an R² of 0.994 and an RMSE of 95.10 during training. Overall, the results indicate that RBFO outperformed both RBF and RBGT in terms of prediction accuracy and generalization capability. This advancement in UCS prediction holds promise for improving soil stabilization precision and supporting more efficient design practices in transportation infrastructure.