【摘要】 针对有限样本下信息不完备致使边坡可靠度结果围绕真值离散问题,研究了小样本条件下可靠度分析改进算法及合理勘察样本数量取值。研究表明:基于Bayes原理建立的可靠度全概率分析法可有效约束可靠指标散布范围,减少小样本统计不确定性影响;通过多批次随机抽样的可靠指标散布特征分析,可确定远小于传统基于t分布结果、能与岩土参数变异水平相匹配的勘察合理取样数目。更多还原

【Abstract】 Analysis results of slope reliability randomly deviate from true value caused by incomplete information of limited samples. Improved algorithm of reliability analysis under limited samples condition and reasonable investigation samples number are researched. Results show that based on Bayes principle,the total probability analysis method of reliability can effectively constraint the distribution range of reliability index and diminish the influence of statistical uncertainty under limited samples condition. Reasonable investigation samples number that is far less than results based on traditional t distribution and matches the variation level of geotechnical parameters which can be determined by analysis of distribution characteristic of multi-batch random reliability index.更多还原