Extensive oedometer tests and physical tests have been conducted on remolded and reconstituted marine soils with different initial water contents and liquid limits. The oedometer test data can be well fitted with a st...Extensive oedometer tests and physical tests have been conducted on remolded and reconstituted marine soils with different initial water contents and liquid limits. The oedometer test data can be well fitted with a straight line in the bilogarithmic ln(1+ e ) ～ lg p plot. The initial effective stress corresponding to the initial void ratio is determined by extrapolation of the bilogarithmic compression line. This new way of extrapolating oedometer test data to the initial void ratio overcomes the difficulty of measuring the residual effective stress of soils in the remolded state. The initial stable compression line of ocean floor deposits under different deposition environments corresponds to the compression line at the sensitivity equal to one. This initial stable compression line obtained in this study is consistent with the available compression line at the sensitivity equal to one which is proposed based on the experimental data of remolded undrained strength and the theoretical concept of Cam clay critical state line.展开更多
软基水闸在水流侵蚀等因素作用下易发生底板脱空现象,极大地威胁水闸的安全运行.针对软基水闸底板脱空检测问题,提出一种基于门控循环单元(Gate Recurrent Unit, GRU)神经网络代理模型和多目标灰狼算法(Multi-Objective Grey Wolf Optim...软基水闸在水流侵蚀等因素作用下易发生底板脱空现象,极大地威胁水闸的安全运行.针对软基水闸底板脱空检测问题,提出一种基于门控循环单元(Gate Recurrent Unit, GRU)神经网络代理模型和多目标灰狼算法(Multi-Objective Grey Wolf Optimizer, MOGWO)的软基水闸底板脱空动力学反演方法.基于GRU神经网络构建表征软基水闸结构模态参数与脱空参数间非线性关系的数学代理模型,基于水闸结构固有频率、归一化振型建立软基水闸脱空参数反演的多目标优化函数,并采用MOGWO优化算法求解多目标优化问题的Pareto最优解.将所提方法应用于室内软基水闸物理模型两种脱空工况的反演计算.GRU神经网络代理模型精度优于多层前馈(Back Propagation, BP)神经网络代理模型及三阶多项式响应面模型,且反演脱空面积和模型实际脱空面积的相对误差分别为6.76%、5.58%,反演效果明显优于单目标反演方法.展开更多
文摘Extensive oedometer tests and physical tests have been conducted on remolded and reconstituted marine soils with different initial water contents and liquid limits. The oedometer test data can be well fitted with a straight line in the bilogarithmic ln(1+ e ) ～ lg p plot. The initial effective stress corresponding to the initial void ratio is determined by extrapolation of the bilogarithmic compression line. This new way of extrapolating oedometer test data to the initial void ratio overcomes the difficulty of measuring the residual effective stress of soils in the remolded state. The initial stable compression line of ocean floor deposits under different deposition environments corresponds to the compression line at the sensitivity equal to one. This initial stable compression line obtained in this study is consistent with the available compression line at the sensitivity equal to one which is proposed based on the experimental data of remolded undrained strength and the theoretical concept of Cam clay critical state line.
文摘软基水闸在水流侵蚀等因素作用下易发生底板脱空现象,极大地威胁水闸的安全运行.针对软基水闸底板脱空检测问题,提出一种基于门控循环单元(Gate Recurrent Unit, GRU)神经网络代理模型和多目标灰狼算法(Multi-Objective Grey Wolf Optimizer, MOGWO)的软基水闸底板脱空动力学反演方法.基于GRU神经网络构建表征软基水闸结构模态参数与脱空参数间非线性关系的数学代理模型,基于水闸结构固有频率、归一化振型建立软基水闸脱空参数反演的多目标优化函数,并采用MOGWO优化算法求解多目标优化问题的Pareto最优解.将所提方法应用于室内软基水闸物理模型两种脱空工况的反演计算.GRU神经网络代理模型精度优于多层前馈(Back Propagation, BP)神经网络代理模型及三阶多项式响应面模型,且反演脱空面积和模型实际脱空面积的相对误差分别为6.76%、5.58%,反演效果明显优于单目标反演方法.
