摘要
超固结比(OCR)和不排水抗剪强度(Su)是土的基本力学参数,对土体沉降变形分析和稳定性计算具有重要影响。采用数据融合技术,结合孔压静力触探(CPTU)测试数据,提出了江苏典型黏性土超固结比和不排水抗剪强度的预测模型。利用特征级数据融合技术(回归树、模型树)与决策级数据融合技术(自举聚合、堆叠泛化)对预测模型的可行性进行分析。将土的超固结比和不排水抗剪强度的预测值、室内试验所得到的参考值以及CPTU传统方法所得到的估计值进行对比分析。结果表明,模型树预测结果比回归树要好,决策级融合算法可以提高回归树的预测结果,但对模型树的预测结果影响较小;叠加回归树和模型树的预测结果会使其预测的不排水抗剪强度比回归树预测的结果要好,但比模型树预测的结果要差;对于几种数据融合模型,OCR预测值大致相当,回归树模型在预测OCR方面稍优于其他数据融合模型,数据融合技术能更好地预测土的超固结比和不排水抗剪强度。
The overconsolidation ratio(OCR) and the undrained shear strength(Su) are the basic mechanical parameters of soils,which can influence the deformation analysis and strength calculation of soils. A prediction model for OCR and S_u of typical clay in Jiangsu Province is proposed by using the data fusion technique and the data of piezocone penetration test(CPTU). The feasibility of the prediction model is analyzed by using the feature-level data fusion techniques(regression tree, model tree) and decision-level data fusion techniques(bagging, stacking). The predicted OCR and Su, the reference values obtained by the laboratory tests and the estimated values obtained by the existing calculation methods are compared and analyzed. The results show that the predicted results of the model tree are better than those of the regression tree. The decision and fusion algorithms can improve the predicted results of the regression tree, but they have little influences on the predicted results of the model tree.The superimposed regression tree and model tree can make the predicted Su better than that of the regression tree, but worse than that of the model tree. For several data fusion models, the predicted OCR is approximately close. The regression tree model is slightly better than other data fusion models in predicting the OCR. Compared with other prediction methods, the data fusion model can better predict the OCR and S_u.
作者
刘晓燕
蔡国军
邹海峰
李学鹏
刘松玉
LIU Xiao-yan;CAI Guo-jun;ZOU Hai-feng;LI Xue-peng;LIU Song-yu(Institute of Geotechnical Engineering,Southeast University,Nanjing 211189,China)
出处
《岩土工程学报》
EI
CAS
CSCD
北大核心
2019年第7期1270-1278,共9页
Chinese Journal of Geotechnical Engineering
基金
国家重点研发计划项目课题(2016YFC0800200)
国家自然科学基金项目(41672294,41877231,51578146)
