基于模糊点数据的地基液化判别

以世界范围内19次大地震106个场地的地基液化现场实测数据为基础,通过对地基液化评价模型建立过程中模糊不确定因素的分析,引入模糊模式识别理论建立各实测数据点隶属度,模拟数据点在构建液化判别模型的不同作用;并通过LOGISTIC回归将数据点的模糊性映射到液化判别模型,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%,为利用现场实测资料进行地基液化判别提供了简便、较为准确的新途径.

Estimation of Soil Liquefaction under Fuzzy Points Data

According to the field testing liquefaction data of cone penetration test from 19 strong earthquakes around the world, the fuzzy uncertainties in construction of soil liquefaction discrimiant equation are analyzed. The uncertainties in field testing data play different roles in establishment of discriminate equation. In order to obtain more accurate discriminant equation, fuzzy uncertainties must be included, i.e., field testing data shall be considered as a kind of fuzzy point data. The fuzzy pattern recognition set is employed to fit the uncertainties in field testing data, and then the uncertainties are reflected in liquefaction discriminant equation by LOGISTIC. The discrimnant equation based on the fuzzy point data is more reliable, and the result of evaluation are more reasonable.