土工参数概率特性的贝叶斯优化估计

详细信息    查看全文 | 推荐本文 |

英文篇名：Bayesian Optimization Estimation for Probability Characteristic of Geotechnical Parameters 作者：徐志军 ; 张健 ; 肖昭然 ; 刘军 英文作者：XU Zhijun;ZHANG Jian;XIAO Zhaoran;LIU Jun;School of Civil Engineering and Architecture,Henan University of Technology; 关键词：土工参数 ; 概率特性 ; 贝叶斯优化 ; 均值 ; 标准方差 英文关键词：geotechnical parameters;;probability characteristic;;Bayesian optimization;;mean;;standard variance 中文刊名：人民黄河 英文刊名：Yellow River 机构：河南工业大学土木建筑学院; 出版日期：2019-01-10 出版单位：人民黄河 年：2019 期：01 基金：国家自然科学基金资助项目(51608177) 语种：中文; 页：103-106 页数：4 CN：41-1128/TV ISSN：1000-1379 分类号：TU43

摘要

足量精确的土工参数测试数据是计算其均值和标准方差的关键,在工程实际中,受到各种不确定性因素的影响,获得足量精确的数据较为困难。基于数理统计方法,建立土工参数数据优化处理模型,将数据优化处理为"好数据"和"一般数据",剔除离散性较大的数据;利用随机加权法将数据量较小的"好数据"和"一般数据"进行加权处理。最后利用贝叶斯方法对土工参数概率分布的均值和标准方差进行优化估计。通过算例分析表明:贝叶斯优化后的标准方差大大降低;贝叶斯优化后的概率分布均值与"好数据"的均值相差很小,但与一般数据的均值和所有数据的均值相差较大。证明了贝叶斯优化后的均值和标准方差更接近工程实际,由此估计出的工程安全性更加科学合理。
        The key procedure to compute the mean and standard variance of geotechnical testing data is sufficient and exact data. However,due to various uncertainties in engineering practice,it is more difficult to obtain sufficient and exact data. This paper used the mathematical statistical method to set up the optimization model of geotechnical parameters,which could classify the geotechnical parameters into"good data"and"general data"and get rid of the singular data. Then the smaller sample between the"good data"and"general data"was advanced by using random weighing method. Finally,Bayesian method was introduced to optimize the mean and standard variance of geotechnical data.The results through case study show that the standard variance of optimized data is greatly decreased. The difference between the mean of"optimized probability distribution"and"good data"is pretty small. The mean of"general data"and"all data"is greater than that of"optimized probability distribution"and"good data",which verifies the presented method can scientifically and reasonably estimate the safety of engineering.

引文

[1] PHOON K K, CHING Jianye. Risk and Reliability in Geotechnical Engineering[M]. New York:Taylor&Francis,2015:339-433.
    [2]黄天朗,宫凤强,伍婷玉.小样本岩土参数概率分布的正交多项式推断方法[J].工程力学,2017,34(8):85-95.
    [3]王新,黄磊群.随机加权法辅助小样本岩土参数概率分布的边坡稳定性分析[J].公路工程,2017,42(3):131-136.
    [4]骆飞,罗强,蒋良潍,等.小样本岩土参数的Bootstrap估计及边坡稳定分析[J].岩石力学与工程学报,2017,36(2):370-379.
    [5]孟庆山,雷学文.土工参数的统计方法及其工程应用[J].武汉冶金科技大学学报(自然科学版),1999,22(4):414-417.
    [6]宫凤强,李夕兵,邓建.小样本岩土参数概率分布的正态信息扩散法推断[J].岩石力学与工程学报,2006,25(12):2559-2564.
    [7]徐志军,肖昭然,蔡正银,等.基于随机加权熵的岩土参数概率分布的拟合研究[J].防灾减灾工程学报,2014,34(s1):6-10.
    [8] ANG A H S,TANG W H. Probability Concepts in Engineering:Emphasis on Applications to Civil and Environmental Engineering. 2nd Edition[M]. New York:John Wiley&Sons,2006:106-110.
    [9]郑忠国.随机加权法[J].应用数学学报,1987,10(2):247-253.