GBDT-based multivariate structural stress data analysis for predicting the sinking speed of an open caisson foundation

Abstract

ABSTRACT Open caisson foundations are often used in large-span bridge construction because of their advantageous, and an open caisson foundation gradually sinks to a predetermined position via earth excavation. Excessive sinking speed may result in various construction risks (e.g. inclination and structural damage of the open caisson). To prevent these risks, it is important to analyse the sinking situation of the foundation and predict the sinking speed during the sinking process. A sinking speed prediction model is proposed based on the gradient boosting decision tree (GBDT) algorithm, and the model can extract the data features of the structural stress monitoring data to predict the sinking speed of open caissons. Taking the supersized open caisson foundation in the Changtai Yangtze River Bridge Project as a case study, the proposed model was validated by using the monitoring data of this foundation. The validation results of this project indicate that the proposed model has high prediction accuracy, short time consumption, a good prediction effect and high practicability. Based on the model’s prediction, an earth excavation scheme can be flexibly adjusted to prevent the potential construction risks of open caissons caused by excessive sinking speed.