一种基于RBF代理模型的土壤承压特性拟合方法

Fitting Method of Soil Pressure Subsidence Characteristics Based on RBF Surrogate Model

基于有限元理论研究了加载速率和平板尺寸对土壤承压沉陷特性的影响,发现同一沉陷深度的土壤承载力随着加载速率的提高而增大,平板尺寸的增大会降低单位面积土壤承载力。采用Prony指数型函数拟合方法对不同土壤力学参数和加载条件下的土壤压力沉陷曲线进行数值拟合,拟合结果具有较高的精度,R2均大于0.9。基于RBF神经网络建立土壤力学参数、加载速率和平板半径与Prony指数函数参数之间的代理模型,实现通过已知土壤力学参数和加载条件直接预测土壤承压沉陷特性的功能。模型测试效果良好。

Based on the finite element theory,the effects of loading rate and plate size on the characteristics of soil pressure subsidence were studied.It was found that the soil bearing capacity with the same subsidence depth increased with the increase of loading rate,and the increase of flat plate size would reduce the soil carrying capacity per unit area.Prony exponential function fitting method was used to fit the soil pressure subsidence curve under different soil mechanics parameters and loading conditions.The fitting result has higher accuracy,and R2 is greater than 0.9.The RBF neural network was used to establish the surrogate model between the soil mechanics parameters,loading rate,plate radius and the parameters of the Prony exponential function.The function of directly predicting the soil pressure subsidence characteristics through known soil mechanics parameters and loading conditions has been achieved,and the model test works well.