The foundations of some ocean engineering structures are built to withstand not only the vertical gravity load V, but also the horizontal load H induced by sea waves and current. The horizontal load includes the conce...The foundations of some ocean engineering structures are built to withstand not only the vertical gravity load V, but also the horizontal load H induced by sea waves and current. The horizontal load includes the concentrated force load, the moment load M, and the torque load T termed also as combined loading. It is of academic and engineering significance to study the deformation law of submarine seabed due to combined loading. On the basis of the three-dimensional elastic mechanics solution of circular foundation, numerical methods are used to analyze the deformation law of submarine soil under circular foundation with six degrees of freedom. The finite element analysis results give the elastic deformation law of soil in three dimensional spaces, modify the theoretical elasticity solution, and presents nonlinear soil deformation mechanism under the circular foundation with six degrees of freedom.展开更多
为了预测深基坑支护桩水平变形的长期发展规律,在卷积神经网络(convolutional neural network,简称CNN)数据空间特征提取基础上,结合长短时记忆神经网络(long and short term memory,简称LSTM)分析数据的时序性和注意力机制(attention m...为了预测深基坑支护桩水平变形的长期发展规律,在卷积神经网络(convolutional neural network,简称CNN)数据空间特征提取基础上,结合长短时记忆神经网络(long and short term memory,简称LSTM)分析数据的时序性和注意力机制(attention mechanism,简称AM)的划分特征权重,构建了能够预测支护桩变形的AM-CNN-LSTM模型。以北京地区某深基坑工程为背景,基于灰色关联方法明确了影响支护桩最大变形的因素,通过构建的模型分析支护桩的单点变形规律,并与反向传播神经网络(back propagation neural network,简称BPNN)、CNN和传统CNN-LSTM模型的预测所得结果进行比较分析。研究结果表明:支护桩最大变形值与深基坑开挖深度、临空天数、支撑内力、土壤性质、桩的尺寸和嵌固深度等因素关联度较高;AM机制显著提升了初始数据信息挖掘深度和变形预测精度,通过梯度下降法不断更新直至满足误差要求;与BPNN、CNN及CNN-LSTM模型相比,AM-CNN-LSTM模型的应用对于支护桩的长期变形预测稳定性较好;通过与实测数据对比,AM-CNN-LSTM模型的预测精度误差在5%~10%以内。展开更多
基金The financial support for this study is through the grants 50909050 from the National Natural Science Foundation of ChinaZR2009FQ004 from the Natural Science Foundation of Shandong Province
文摘The foundations of some ocean engineering structures are built to withstand not only the vertical gravity load V, but also the horizontal load H induced by sea waves and current. The horizontal load includes the concentrated force load, the moment load M, and the torque load T termed also as combined loading. It is of academic and engineering significance to study the deformation law of submarine seabed due to combined loading. On the basis of the three-dimensional elastic mechanics solution of circular foundation, numerical methods are used to analyze the deformation law of submarine soil under circular foundation with six degrees of freedom. The finite element analysis results give the elastic deformation law of soil in three dimensional spaces, modify the theoretical elasticity solution, and presents nonlinear soil deformation mechanism under the circular foundation with six degrees of freedom.
文摘为了预测深基坑支护桩水平变形的长期发展规律,在卷积神经网络(convolutional neural network,简称CNN)数据空间特征提取基础上,结合长短时记忆神经网络(long and short term memory,简称LSTM)分析数据的时序性和注意力机制(attention mechanism,简称AM)的划分特征权重,构建了能够预测支护桩变形的AM-CNN-LSTM模型。以北京地区某深基坑工程为背景,基于灰色关联方法明确了影响支护桩最大变形的因素,通过构建的模型分析支护桩的单点变形规律,并与反向传播神经网络(back propagation neural network,简称BPNN)、CNN和传统CNN-LSTM模型的预测所得结果进行比较分析。研究结果表明:支护桩最大变形值与深基坑开挖深度、临空天数、支撑内力、土壤性质、桩的尺寸和嵌固深度等因素关联度较高;AM机制显著提升了初始数据信息挖掘深度和变形预测精度,通过梯度下降法不断更新直至满足误差要求;与BPNN、CNN及CNN-LSTM模型相比,AM-CNN-LSTM模型的应用对于支护桩的长期变形预测稳定性较好;通过与实测数据对比,AM-CNN-LSTM模型的预测精度误差在5%~10%以内。
