Design Hybrid ARO-SVR Analysis to Predict the Pile Bearing Capacity

Abstract

The assessment of the load-bearing capacity of piles holds significant importance in the design of pile foundations. This paper presents hybridized support vector regression (SVR) models by utilizing the Artificial Rabbit Optimization (ARO) and the Black Widow Optimization Algorithm (BWOA) to predict the bearing capacity of concrete piles. A repository comprising 472 reports on static load tests conducted on driven piles was employed for the study. The dataset was allocated into three parts: the training set (70%), the validation set ( 15% ), and the testing set ( 15% ). Multiple criteria for assessing quality were utilized to evaluate the effectiveness of the models. The first rank belonged to the SVR model integrated with the ARO algorithm, where it could gain the higher value of R^2 in all of training (R^2 = 0.9876), validating (R^2 = 0.9778), and testing sections (R^2 = 0.9874), and the lowest value of RMSE in all the training ( RMSE = 39.393 ), validating ( RMSE = 53.727 ) and testing sections ( RMSE = 38.082). The findings indicate that the suggested model is highly appropriate for predicting the capacity of concrete piles.

Keywords

• concrete piles
• bearing capacity
• estimation
• regression analysis
• artificial rabbit optimization