The implementation of a least square support vector regression model for predicting the ultimate bearing capacity of rock-socketed piles

The ultimate load-bearing capacity (Q_u) of rock-socketed piles, drilled shafts, or piers is contingent on various factors. Rock-socketed piles find widespread application in foundation engineering, particularly when it comes to supporting structures erected on rock formations. Determining the Q_u of these piles presents a challenging engineering endeavor, demanding a comprehensive grasp of geotechnical and structural principles, consideration of site-specific circumstances, and meticulous planning and execution during design and construction phases. Accurately predicting the Q_u of rock-socketed piles is of paramount importance for various reasons, including ensuring safety, optimizing costs, complying with regulations, and mitigating environmental impacts. Precise forecasts enable the optimization of foundation designs and significantly minimize potential risks inherent in construction projects. The primary aim of this research is to employ two machine learning-based models for the Q_u prediction process as innovative research. These models comprise Radial Basis Function and Least Squares Support Vector Regression. Additionally, two distinct meta-heuristic optimization techniques, namely the African Vultures Optimization Algorithm and the Pelican Optimization Algorithm, were chosen to fine-tune the base models, aiming to optimize their predictive performance. Various performance criteria, such as R², RMSE, MSE, MAPE, and the SI index, were utilized to assess the predictive performance of the models. These assessments illustrate that, when it comes to practical utility, the LSPO (LSSVR + POA) model emerges as the most superior predictor, attaining an impressive maximum R²_train of 0.994 and a minimal RMSE_train value of 762.93. Properly designed foundations based on accurate Q_u predictions can minimize the environmental impact of construction projects. This includes reducing the need for excessive excavation or altering the natural landscape.