Prediction of the Dynamic Soil-Pile Interaction under Coupled Vibration using Artificial Neural Network Approach

摘要

Pile foundations resulting from dynamic excitation is a complex phenomenoninvolving complex pile-soil interaction. Pile foundations behave in a nonlinearfashion at large displacements because of the soil nonlinearity at high strain and pileseparation. This paper attempts to study the dynamic soil-pile interaction on thenonlinear response under coupled vibration by both experimental and numericalstudy. This paper presents the small-scale field test results of single piles and 2 × 2group piles subjected to different excitation intensities. Resonant frequencies andamplitudes are determined for both horizontal and rocking motion from the observedresponse curves. Artificial neural network (ANN) models are developed based onfield test results for the prediction of dynamic behaviour of piles under coupledmotion. Importance of different strategies in developing robust ANN models areexplored and discussed. Different ANN models are developed using evolutionarylearning algorithm and Bayesian regularization algorithm. Various statisticalperformance criteria and its importance to compare the developed ANN models arediscussed. Different sensitivity analyses are made to identify the important inputparameters for amplitude and resonance determination.