基于贝叶斯网的结构延性需求分析  

Bayesian network based structural ductility demand analysis

作　　者：易伟建[1] 李浩[1] 

机构地区：[1]湖南大学,长沙410082

出　　处：《土木工程学报》2009年第11期15-22,共8页China Civil Engineering Journal

基　　金：国家自然科学基金(50678064)

摘　　要：概率延性需求分析是基于性能的结构抗震评估与设计中的重要一环。根据搜集到的1918条地震记录,对具有不同屈服水平系数及周期的单自由度体系作了弹塑性时程分析,通过拟合得到简化的延性需求计算公式;针对单一地面峰值加速度作为地震强度指标的不足,根据回归分析构造一个新的参数用于描述地震频谱特性对延性需求的影响;以地面峰值加速度和此新参数作为地震强度指标向量,建立给定地震强度条件下结构延性需求的概率关系,并形成相应的贝叶斯网。引进马尔科夫链蒙特卡罗模拟方法,用于更新给定地震强度观察值条件下延性需求的后验分布,从而实现延性需求计算的本地化。通过算例分析了各地震强度参量、给定观测值条件下对延性需求计算结果的影响。Ductility demand probabilistic analysis is important for performance based seismic evaluation and design. Nonlinear time history analysis of SDOF with various yield strength coefficients and periods are performed using 1918 earthquake ground motion records to obtain simplified formula for ductility demand through curve fitting. Considering the insufficiency of PGA for being as single intensity measure, a new parameter is constructed through regression analysis to quantify the effect of earthquake spectrum characteristics on the ductility demand. With PGA and the new parameter being as IM vector, the relationship of ductility demand probabilistic with given earthquake intensity is formulated, and a Bayesian network is established. Markov Chain Monte Carlo method is introduced to update the ductility demand posterior probabilistic distribution with given observation of earthquake intensity, thus localizing the calculation of the ductility demand. Examples analyze how the parameters of earthquake intensity and the condition of given observations affect the calculation result.

关 键 词：结构抗震 延性需求 贝叶斯网 马尔科夫链蒙特卡罗 后验分布 

分 类 号：TU313[建筑科学—结构工程] TU312.1