Abstract
MACHINE LEARNING ANALYSIS OF THE PULLOUT RESISTANCE BEHAVIOR ON A HELICAL PILE
*Shaymaa Alsamia and Edina Koch
ABSTRACT
In this study, we investigate the behavior of the pullout resistance of a helical pile using machine learning techniques. Specifically, we apply three different techniques - adaptive neuro-fuzzy inference system, random forest regression, and support vector regression - to the experimental results of a helical pile. We evaluate the performance of these techniques on both the training and test sets and compare their results. Our findings indicate that while the adaptive neuro-fuzzy inference system showed good performance on the training set, it had deficiencies when tested. The support vector technique showed better performance than the adaptive neuro-fuzzy inference system, but not as well as the random forest algorithm. Ultimately, the random forest machine learning regression outperformed other methods, delivering good predictions with acceptable error values. These results suggest that machine learning can be an effective tool for predicting the pullout resistance behavior of a helical pile embedded in the soil, which may have practical implications for the design and optimization of helical pile foundations.
[Full Text Article] [Download Certificate]