基于SVM的固化土无侧限抗压强度模型

为实现较少试验次数下固化土无侧限抗压强度(qu)的准确预测, 提出了基于支持向量机(SVM)的固化土qu的预测模型. 以固化剂各组分掺入比、龄期、初始含水量、固化剂掺量等因素为输入量, 固化土的qu作为输出量, 以径向基为核函数, 采用网格搜索法和交叉验证法进行参数优化, 建立了基于SVM的固化土qu的预测模型. 算例分析表明: 该模型适用于任意条件下固化土qu的精确预测, 且在较小试验成本下实现与响应面法相当的预测精度.

Unconfined compressive strength model of solidified soil based on SVM

Based on the mechanism of support vector machine (SVM), an optimization model is developed to accurately predict the unconfined compressive strength (qu) of solidified soil under the condition of the limited number of tests. In the proposed model, a series of factors such as the components of the solidified agent, curing age, initial water content, the content of solidified agent, etc. are considered to be the input parameters. The associated values of qu for each case are used for the output parameter. The radial basis function is adopted as the core function, in which the key parameters are optimized using the grid search method and cross-validation method. The simulated results show that the proposed model behaves well in predicting the qu of solidified soil under any conditions. It is worth mentioning that the proposed model has the equivalent prediction accuracy with the response surface method but with fewer unconfined compression strength tests.