基于孔压静力触探的岩土参数正态性检验方法对比

Comparison of Normality Test Methods on Geotechnical Parameters Based on Piezocone Penetration Test

岩土参数的正态分布是岩土工程可靠分析的基本假定之一。目前,岩土工程参数的正态性检验以KS(Kolmogorov-Smirnov)检验为主,然而诸多研究表明,对于小样本数据,KS检验的正态性检验结果并不可靠。本文以废黄河泛滥沉积相粉土中的孔压静力触探(CPTU)锥尖阻力qt数据为研究对象,首先采用ANOVA(方差分析)检验,从CPTU测试资料中进行总体的样本筛选,然后应用SW(Shapiro-Wilk)检验、KS检验、LF(Lilliefors)检验和AD(Anderson-Darling)检验等4种正态性检验方法,在不同样本容量条件下,对测试数据进行正态性检验评价。当测试数据的样本容量充足时,分析结果表明:SW检验结果最为严格,KS检验结果最不保守,而LF检验和AD检验的严格性介于两者之间;对于常规岩土工程设计,采用KS检验结果即可满足稳定性分析要求,对于复杂岩土工程设计,采用SW检验对设计参数进行正态性检验,可降低设计的不确定性。

Normality assumption is one of the fundamental prerequisites of geotechnical risk assessment. Currently the Kolmogorov-Smirnov(KS) test is mainly applied to assess the normality of geotechnical parameters in geotechnical literatures;However, many studies demonstrated that the power of the KS test might be the lowest among the common formal normality tests for a small sample size. With the piezocone penetration test data collected from the floodplain silts of abandoned Yellow River, the performance of four formal normality tests was assessed, including Shapiro-Wilk(SW) test, KS test, Lilliefors(LF) test, and Anderson-Darling(AD) test. The ANOVA test was utilized to identify the samples of the same population. The normality tests were performed on different sample volume. When the sample size is sufficient, the results show that: Among the four formal normality tests, the KS test provides the least conservative assessment on rejecting normality assumption, the SW test is the strictest, while the strictness of LF test and AD test is between the two;For conventional geotechnical engineering design, the KS test result can meet the stability analysis requirements, while for complex geotechnical engineering design, the SW test is used to test the normality of the design parameters to reduce the design uncertainty.