基于神经网络的深厚软土地层参数反演分析

在深基坑工程数值分析中,土体本构关系的选取及其参数的确定对分析的结果会产生重要的影响。结合一个海中深厚软土基坑实例,根据现场实测数据,利用参数敏感性分析确定地层的关键参数,提出一个基于参数敏感性分析和BP神经网络的动态施工反演分析方法。分析结果显示,围护桩水平位移监测值与反演值吻合较好,说明基于参数敏感性分析和BP神经网络的动态施工反演分析方法能够很好地反映土体的非线性、不确定特征,依据该反分析方法获得的土体参数可用于估算下一工况的基坑位移。本次反演得到的土体参数可作为当地深厚软土地层类似工程土体参数的参考取值。

Back Analysis of Deep Soft Soil Parameters Based on Neural Network

In the numerical analysis of deep foundation pit engineering,selection of soil constitutive model and parameters has an important impact on the results of numerical analysis.A deep sea soft soil foundation was taken as an instance in analysis.According to the measured data,the parameter sensitivity analysis method was used to determine the key parameters of stratum,and a dynamic construction back analysis method based on parameter sensitivity analysis and BP neural network was proposed.It is shown that the monitoring values and calculation values of retaining pile horizontal displacement are in good agreement,which illustrates that the method can reflect the soil nonlinearity,uncertainty characteristics.And the soil parameters obtained by the method can be used to estimate foundation displacement in the next working condition.The related inversion values of the soil parameters can provide reference for the similar projects in local deep soft soil ground when the soil parameters are needed.