【摘要】 滑坡各影响因素之间存在较强的相关性。偏最小二乘回归(PLS)方法在一个算法下同时实现了回归建模(多元线性回归分析)、数据结构简化(主成分分析)以及两组变量间的相关分析(典型相关分析),给因变量之间存在较高相关性的回归带来极大的便利。为了探讨PLS方法在滑坡稳定性影响因素分析中的适用性,选取位于三峡水库区巫山县的邓家屋场滑坡为试验区。在考虑三峡水库蓄水后引起的地下水力坡度变化、地震因素以及建筑物因素对邓家屋场滑坡稳定性的影响条件下,建立了滑坡稳定性系数与以上指标的PLS回归方程,以期达到对邓家屋场滑坡稳定性影响因素敏感性的分析目标。计算建立的回归模型为:K*=0.181468×C*+0.274876×φ*-0.611369×J*-0.238604×α*-0.105219×ΔW*。表达式各因素影响力的排序为:地下水力坡度>滑动面内摩擦角>地震加速度>滑动面内聚力>滑体均匀加载。结果表明,采用偏最小二乘回归方法对滑坡稳定性影响因素进行分析具有物理意义明确、计算简单、建模效果好、解释性强的特点,是一种可行的解决方案。更多还原

【Abstract】 Sensitivity analysis is helpful for the stabilization designing of landslide.However,three issues hinder such quest for:(1) different measurement units for the influence factors controlling the behavior of an active landslide;(2) possible correlation among the influence factors;(3) variation in each of the factors.In this study,partial least squares(PLS) regression is employed to cope with the preceding issues.PLS is a recent technique that generalizes and combines features from principal component analysis and multiple regression.It is particularly useful when we need to predict a set of dependent variables from a(very) large set of independent variables(i.e.,predictors).First,normalization of each factor is used to create a dimensionless indicator.Second,principal components analysis(PCA),a multivariate statistical technique,is applied to eliminate the possible correlation among the influence factors.Based on PCA and inverse transform,applied to a landslide example,the sequence of importance of the influence factors was obtained.The results show the PLS is a suitable tool to do the sensitivity analysis of stability for landslide stabilization designing.更多还原