基于结构可靠性方法的贝叶斯模型更新:氯盐侵蚀耐久性案例分析  

BAYESIAN MODEL UPDATING BASED ON STRUCTURAL RELIABILITY METHOD:CASE STUDY FOR CHLORIDEINDUCED DURABILITY ANALYSIS

作　　者：肖雄 王泽宇 李全旺[1] 曹迈 XIAO Xiong;WANG Ze-yu;LI Quan-wang;CAO Mai(Department of Civil Engineering,Tsinghua University,Beijing 100084,China)

机构地区：[1]清华大学土木工程系,北京100084

出　　处：《工程力学》2022年第S01期336-341,共6页Engineering Mechanics

基　　金：清华大学国强研究院资助项目(2021GQC0003);国家重点研发计划项目(2019YFE0112800)。

摘　　要：贝叶斯模型更新是一种基于先验知识和最新观测数据,具有减少工程模型不确定性的计算能力的技术。通过将贝叶斯更新问题重新解释为结构可靠性问题,最新的BUS算法(即采用结构可靠性方法的贝叶斯更新)为快速和稳健的后验密度估计开辟了新的途径。为了实现氯盐侵蚀耐久性模型参数快速准确更新的目标,该文采用基于子集模拟的BUS方法。通过研究一个工程指导的复杂混凝土耐久性模型,证明了计算效率和精度。结果表明,模型参数的统计特性可以根据现场监测的信息成功地更新,并与简单拒绝采样估计的结果接近。Bayesian model updating refers to a technique based on prior knowledge and up to date observations,which has the computational capability to reduce the uncertainties in engineering models.By reinterpreting Bayesian updating problem into a structural reliability problem,the state-of-the-art BUS algorithm(i.e.,Bayesian updating with structural reliability methods)opens a new avenue for a fast and robust posterior density estimation.Towards the goal of fast and accurate parameter updating for a chloride-induced durability model,this paper adopts the subset simulation-based BUS approach.The computational efficiency and accuracy of the approach are demonstrated through investigating an engineering-guided sophisticated concrete durability model.Results showcase that the statistical properties of model parameters can be successfully updated by the information monitored in-situ and are close to the results estimated through simple rejection sampling.

关 键 词：模型更新 贝叶斯更新 拒绝采样 结构可靠性 子集模拟 混凝土耐久性 

分 类 号：TU311.2[建筑科学—结构工程]