Empirical Method for Settlement Prediction of Single Piles Using Higher Order Neural Network and Particle Swarm Optimization

摘要

The settlement analysis of axially loaded is a challenging task due to the complicated nature of load transfer mechanism, the nonlinear behaviour of soil and the modification of soil properties as a result of pile installation. In this paper, an empirical model for predicting load - settlement relationship of a single pile under axial load has been developed using a hybrid PSO/higher-order neural network algorithm. The aim of developing the model is to reduce the current level of difficulty in predicting the deformation of piles under axial loading, taking into account the complex behaviour of soil-pile interface and the effects of pile installation on the soil behaviour. A database of static pile loading tests is used to train and validate the model. The inputs to the model include the variation of SPT blow counts and soil type along the pile embedment, type of pile installation and geometric parameters and elastic modulus of the pile. Based on the comparisons with load transfer method and field measurements, the prediction of the new model outperforms the former and in good agreement with the latter.