Developing a Machine Learning Model for Predicting the Settlement of Bored Piles

Abstract

Analyzing the behavior and determining the settlement of bored piles is of significant importance in construction practice. Traditional experimental methods are time-consuming and expensive, while theoretical methods often yield less reliable results. This research focuses on developing a machine learning model based on an artificial neural network, which is trained and deployed to predict pile top settlement using EXCEL software. The data used to train the model consist of results from static pile load tests conducted in Vietnam and around the world. The findings indicate that the prediction model is highly accurate in predicting pile settlement. Compared to empirical formulas, the artificial neural network model demonstrates superior performance in determining pile top settlement. Additionally, the research proposes an empirical formula that simulates the artificial neural network in EXCEL, enabling the quick estimation of pile top settlement using only a few specific parameters for the pile and soil.