基于截尾概率分布的基坑突涌模糊可靠度分析

岩土工程可靠度分析中,计算参数具有随机性,稳定性评价具有模糊性。传统的Monte Carlo模拟方法计算可靠度时,往往假定参数概率分布在正负无穷之间分布,与真实情况不符,而且其计算效率也往往较低。规范推荐的基坑突涌验算公式,计算结果往往偏于保守。引入截尾概率分布的确定方法,对参数概率分布进行截尾处理;提出拉丁超立方抽样与最大熵原理结合来确定结构响应概率分布的方法,并将其与模糊可靠度原理结合,构建了基于抽样模拟的模糊可靠度计算模型;推荐了考虑土体抗剪强度的突涌验算公式。将所提方法应用于某深基坑工程突涌分析中,其结果表明,基于截尾分布的抽样可以有效避免参数抽样值为负数的情况,所提计算模型效率明显优于传统的Monte Carlo模拟方法,模糊可靠度的计算结果比经典可靠度更符合工程实际情况。

Fuzzy Reliability Analysis on Water Inrush in Foundation Pit Based on Truncated Probability Distribution

In the reliability analysis of geotechnical engineering, the calculation parameters have characteristic of random and the stability evaluation is fuzzy. It is usually assumed that the parameters are distributed between the positive and negative infinity when the Monte Carlo simulation is applied to calculate the reliability, but this is inconformity with the real case. The computational efficiency of the Monte Carlo simulation is often low. The calculation results tend to be conservative by using the formula which is recommended by the specification for check calculation the foundation pit inrush. The parameters probability distributions are truncated by utilizing the method of the truncated probability distribution. The method for determining structure response probability distribution is proposed by combing the Latin hypercube sampling method and the maximum entropy principle, and it is used in fuzzy reliability principle to establish fuzzy reliability calculation model based on sampling simulation. It is used in fuzzy reliability principle to establish fuzzy reliability calculation model based on sampling simulation. The formula which includes shear strength of soil for check calculation the foundation pit inrush is recommended. The presented method is applied to analyze deep foundation pit inrush, the results show that the sampling based on the truncated probability distribution can efficiently avoid the situation that the sampling value is negative. The efficiency of the proposed calculation model is obviously better than the traditional Monte Carlo simulation method. The result of fuzzy reliability is more accord with the engineering practice than the classical reliability result.