摘要
针对土性指标的回归模型通常没有考虑土性指标的非线性趋势分量以及土层其他因素引起的土性指标突变,建立了一个非线性回归模型,并考虑土性随机分量之间的自相关性的问题,使土性指标的统计模型具有普遍性.同时,为了在回归模型中体现土层其他因素对土性指标的影响,构造了一个虚拟变量并将之应用于土性指标的统计模型中,从而建立了第二个非线性回归模型.最后以两组不排水抗剪强度实测数据为例对两类模型的准确性进行了验证,结果表明:在所能见到的文献中,提出的第二个模型的回归方差是最小的.
The non-linear trend and soil other factors causing sudden changes of soil property index are not considered in ordinary regression method. A non-linear regression model is put forward which considers the soil auto-correlation. After a dummy variable is introduced into the regression model in order to take account of the effects of other factors on soil index, another statistical model is also proposed. Two groups of measured data of undrained shear strength were used to validate the proposed models. The results show that regression variance obtained by the model which introduces the dummy variable is the least among the literature that can be found.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2007年第2期4-7,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(50478090)
湖北省自然科学基金资助项目(2005ABA303)
关键词
空间变异性
回归模型
非线性最小方差
不排水抗剪强度
spatial variability
regression model
nonlinear minimum variance
undrained shear strength
