土石混合体地基系数支持向量机模型与应用

Foundation Coefficient Model of Soil-rock Mixture based on Support Vector Machine and Its Application

针对缺乏土石混合体地基系数评价模型这一现状,构建了基于支持向量机的土石混合体地基系数评价模型。该模型以土体弹性模量、泊松比以及块石含石率作为输入参数,地基系数作为输出参数。为建立训练数据,基于土石混合体弹性模量与泊松比计算模型,建立了FLAC3D平板载荷模型开展数值实验。基于数值实验数据,训练支持向量机模型并进行初步验证。将该支持向量机模型用于预测现场试验获得的土石混合体地基系数,进一步验证了该模型的合理性与准确性。通过敏感性分析发现,土体弹性模量对土石混合体地基系数的影响最大,其次为含石率。基于支持向量机模型,讨论了5种已有地基系数计算模型应用于土石混合体地基系数计算的可行性,确认了可用于评估土石混合体地基系数的显式模型。

A foundation coefficient evaluation model of soil-rock mixture was established based on support vector machine(SVM). In this model,the elastic modulus of soil,Poisson’s ratio of soil,and volumetric block proportion(VBP)were taken as input parameters and the foundation coefficient was taken as an output parameter. In order to establish the training data,the FLAC3 D plate load model was built based on the elastic modulus and Poisson’s ratio calculation model of soil-rock mixture. According to the numerical test data,the support vector machine model was trained and preliminarily verified. The support vector machine model was used to predict the foundation coefficient of soil-rock mixtures obtained from field experiments,and the rationality and accuracy of the model were further verified. Through sensitivity analysis,it could be found that the elastic modulus of soil has the greatest influence on the foundation coefficient of soil-rock mixture,followed by the volumetric block proportion. Based on the support vector machine model,the feasibility of applying the five existing foundation coefficient calculation models to soil-rock mixture foundation coefficient calculation was discussed,and the explicit model for evaluating soil-rock mixture foundation coefficient was confirmed.