考虑替代模型不确定性的结构动力特性全局敏感性分析  

Global Sensitivity Analysis of Structural Dynamic Characteristics Considering Metamodel Uncertainty

作　　者：万华平 钟剑[1] 任伟新[1] 

机构地区：[1]合肥工业大学土木与水利工程学院,合肥230009

出　　处：《应用数学和力学》2018年第1期1-10,共10页Applied Mathematics and Mechanics

基　　金：国家自然科学基金(51508144;51608161);中国博士后科学基金(2015M581981;2016M602007);香江学者计划项目(XJ2016039);安徽省自然科学基金(1608085QE118)~~

摘　　要：结构参数的不确定性必然导致其动力特性具有不确定性,全局敏感性分析是定量各不确定参数动力特性影响大小的有效手段.但是全局敏感性分析具有计算花费高的问题,为此采用快速Gauss(高斯)过程模型来降低计算成本.此外,该文采用的是全局Gauss过程模型而不是它的均值函数来进行全局敏感性分析,考虑了替代模型不确定性,给出了敏感性指标的分布.该方法的可靠性通过具有解析敏感性指标值的测试函数得到验证.最后,将该方法用于安庆铁路长江大桥动力特性的敏感性分析.Uncertainty of structural parameters will unavoidably lead to uncertainty of structural natural frequencies. The global sensitvity analysis（ GSA） is an effective approach to quantify the contributions of individual parameters to the induced uncertainty of dynamic characteristics. However,the GSA has the issue of high computational cost that needs to be addressed. The fast-running Gaussian process model（ GPM） was used as a surrogate for the costly computer models,to reduce the computational burden of the GSA. Moreover,the influence of the metamodel uncertainty associated with the GPM was taken into account. The effectiveness of the presented GPM-based method for the GSA was verified with a test function. Finally the GPM-based approach was applied to the GSA of structural dynamic characteristics of the Anqing Yangtze River Railway Bridge.

关 键 词：替代模型不确定性 动力特性 Gauss过程模型 全局敏感性分析 桥梁结构 

分 类 号：U443.6[建筑科学—桥梁与隧道工程]