Four novel evolutionary computational models to estimate piles' bearing capacity

Abstract

Accurate pile-bearing capacity prediction is crucial for ensuring the stability and safety of deep foundations, particularly for tall buildings. This study investigates the use of four hybrid evolutionary computational models - Whale Optimization Algorithm (WOA), Artificial Bee Colony (ABC), Ant Colony Optimization (ACO), and Ant Lion Optimizer (ALO) - to enhance prediction accuracy. These models were evaluated for training and testing datasets based on their population sizes and performance metrics, such as the coefficient of determination (R2) and root mean square error (RMSE). The WOA model demonstrated the highest accuracy, achieving an R2 of 0.979 (training) and 0.968 (testing), along with RMSE values of 0.079 and 0.11, respectively. The ALO model followed closely, with an R2 of 0.989 (training) and 0.968 (testing), though it showed a higher RMSE in testing at 0.235. ABC and ACO, with R2 values ranging between 0.883 and 0.958, displayed lower accuracy than WOA and ALO. The models were ranked based on their performance, with WOA obtaining the highest total rank, followed by ALO, while ABC and ACO shared a similar total rank. These findings highlight the potential of hybrid evolutionary models for improving pile-bearing capacity predictions, which is vital for geotechnical engineering applications.