摘要
以世界19次大地震106个场地的地基液化实测数据为基础,引入多级模糊模式识别方法并建立各实测数据点隶属度,通过logistic回归将数据点的模糊性映射到液化势判别模型,导出液化概率评价方程,建立地基液化概率和液化安全系数的函数关系,预测液化和非液化的可靠性分别达到92.9%和80.2%。
According to the field testing liquefaction data of cone penetration test from 19 strong earthquakes around the world, the multi-polar fuzzy recognition theory is introduced to establish each field measuring point＇s grade, the uncertainties, which the fuzzy pattern recognition set is employed to fit, are reflected in the liquefaction discriminant equation by logistic regression. Then the liquefaction probability evaluation equation is deduced and the functional relationship between the foundation liquefaction and liquefaction safety factors is established. By using them, reliability of forecasting the liquefaction and non-liquefaction is 92.9% and 80.2 % individually.
出处
《土工基础》
2008年第5期51-54,共4页
Soil Engineering and Foundation
关键词
多级模糊模式识别
地基液化势
静力触探
LOGISTIC回归
multi-polar fuzzy pattern recognition, foundation liquefaction, cone penetration test, logistic regression
