岩土参数概率分布类型及其选择标准

 岩土参数的概率分布类型对风险评价和可靠度分析具有相当大的影响，对其研究是一项基础性的重要工作。从分析总结各种概率分布类型的产生背景、所描述对象的物理意义着手，将其分门别类，发现大致有5大系列，即Bernoulli分布系列、Poisson分布系列、极值分布系列、中心极限分布系列和用于统计分析的分布系列。然后研究岩土参数被建模为随机变量时影响其不确定性的因素，得出用中心极限分布系列对其描述较为合理，在此基础上，详尽地对比隶属于中心极限分布系列的正态、对数正态分布的各种特性，提出在二者之间如何选择的标准，即当变异系数小于30%且偏度系数小于0.025时选择正态分布，反之，选择对数正态分布。另外，论证泊松比更适合建模为b 分布的事实。最后，用一统计实例诠释本文所提观点的正确性。研究成果对岩土工程的风险评价和可靠度分析具有重要的参考价值。

TYPES AND SELECTION CRITERIA OF PROBABILITY DISTRIBUTION OF ROCK AND SOIL PARAMETERS

The types of probability distribution of rock and soil parameters have lots of effects on the risk assessment and reliability analysis in geotechnical engineering. So it is a primary work to study rock and soil parameters. Based on the background and physical meaning of various probability distributions,all of them are approximately classified into five series by Bernoulli distribution series,Poisson distribution series,extreme value distribution series,central limit distribution series,and distribution series using for statistical analysis. Subsequently,the factors that affect the uncertainties of rock and soil parameters as random variables are studied. As a result,the central limit distribution series are considered more reasonable for soil parameters. Based on there,the characteristics of normal and lognormal distributions of central limit distribution series are compared in detail. Then,a selection criteria that normal distribution is chosen when coefficient of variation less than 30% and skew coefficient less than 0.025 and lognormal contrarily is recommended. In addition,the β distribution which is more suitable for Poisson's ratio is also studied. Finally,a statistical case is studied to illustrate the validity of the viewpoint. It is believed that the research results have important reference values for the risk assessment and reliability analysis in geotechnical engineering.