等p路径下砂土本构关系的归一化特性及数值建模方法

NORMALIZATION CHARACTERISTIC OF SAND UNDER THE STRESS PATH OF CONSTANT p AND THE CORRESPONDING NUMERICAL MODELING METHOD

用神经网络对土体进行建模能反映应力路径相关性、反映土的剪胀剪缩以及反映体应力、剪应力对体应变、剪应变的交互影响,因而成为一种比较理想的建模方式.能否在样本有限的情况下获得精度比较高的本构模型正是主要的研究目的.通过研究中密砂在等p路径下的三轴试验曲线,发现其应力-应变关系曲线在常规应力范围内具有归一化特性.选择合适的归一化指标对砂土三轴试验数据进行归一化,以归一化的试验数据为训练样本进行神经网络训练,得到了比较理想的砂土的神经网络本构模型.本构模型仿真值与试验值符合较好,表明所给出的建模方法是合理的.提出的建模方法可以在所有试验数据的基础上自动实现概率寻优,能有效降低噪声信号的干扰、减小试验数据的分散造成的影响.

Neural network is an effective way of modeling constitutive relations of soils because it can effectively reflect the influence of stress path, shear dilation as well as interaction between shear strain or volume stress and shear stress or volume stress. The main objective of this paper is to find a way of modeling constitutive relations of soils with high-precision by using the approach of neural network when the training samples are not enough. Study shows that there is normalization characteristic about the stress-strain curves of middle-dense sands in the range of routine stress under the stress path of constant p. The triaxial test data are normalized by choosing proper normalization parameters. The neural networks are trained by regarding the normalized data as samples and then the constitutive model of sand described by neural networks is obtained. The simulation value of the neural networks accord with the test value well, which shows that the modeling method proposed in this paper is reasonable. It can achieve probabilistic optimization automatically based on all test data by using the modeling method, and can reduce the interference of noise signal, lower the influence caused by dispersive test data.