水道港口2023，Vol.44Issue(5)：747-753,810,8.
基于深度神经网络的群桩局部冲刷深度预测
Local scour depth prediction of pile group based on deep neural network
摘要
Abstract
The study of local scour of pile group is crucial to ensure the stability of marine structure foundation.The calculated values of scour depth of pile group obtained by using current codes and empirical formulas are too conservative and too discrete.The depth neural network has strong nonlinear mapping capability.A depth neural network model was established using the neural network and experimental data from previous studies to predict the scour depth and perform sensitivity analysis on the results.It is shown by the analysis that the predicted scour depth of the model established in this paper fits well with the actual test values,and the feasibility and effectiveness of the scour depth neural network model are demonstrated.The theoretical basis for the depth of landfill and later protection of the marine engineering buildings supported by pile groups can be provided,and significant engineering and theoretical significance are attached to it.关键词
深度神经网络/群桩/冲刷深度/ANN预测/深度学习/TensorFlow应用Key words
deep neural network/pile group/scour depth/ANN prediction/deep learning/TensorFlow application分类
交通工程引用本文复制引用
张丽萍,梁丙臣,张黎邦,张嶔,杨博..基于深度神经网络的群桩局部冲刷深度预测[J].水道港口,2023,44(5):747-753,810,8.基金项目
国家自然科学基金项目(51739010) （51739010）
