基于贝叶斯推断的裂纹梁损伤参数识别  

Damage parameter identification of cracked beam based on Bayesianian inference

作　　者：梁岗[1] 沈奎双 马玉付 LIANG Gang;SHEN Kuishuang;MA Yufu(Logistics Engineering College,Shanghai Maritime University,Shanghai 201306,China)

机构地区：[1]上海海事大学物流工程学院,上海201306

出　　处：《上海海事大学学报》2018年第3期85-91,共7页Journal of Shanghai Maritime University

基　　金：国家高技术研究发展计划("八六三"计划)(2005AA11Z123);上海市重点学科建设项目(S30602)

摘　　要：针对在结构损伤诊断过程中存在的不确定性和对称结构损伤参数难以识别的问题,以梁损伤前后两阶频率变化平方比为基础损伤指标,采用对基础损伤指标进行积分处理的方法,构建新的损伤指标。基于贝叶斯结构损伤诊断理论,建立损伤参数的后验概率分布。采用马尔科夫链蒙特卡洛(Markov chain Monte Carlo,MCMC)方法解决贝叶斯方程中存在的高维积分问题。仿真和算例分析说明,该方法可实现对损伤参数的有效估计。In order to solve the problems of the uncertainty in the process of structural damage diagnosis and the difficulty in identifying damage parameters of symmetrical structure,the two order frequency-change square ratio of beam before and after damage is taken as the basic damage index,and a new damage index is constructed by integrating the basic damage index.The posterior probability distribution of damage parameters is established based on the Bayesianian structural damage diagnosis theory.The Markov chain Monte Carlo(MCMC)method is adopted to solve the problem of high dimensional integral in Bayesianian equations.Simulation and example analysis show that the method can effectively estimate damage parameters.

关 键 词：贝叶斯推断 简支梁 裂纹 损伤识别 马尔科夫链蒙特卡洛(Markov CHAIN MONTE Carlo MCMC)方法 

分 类 号：U653.921[交通运输工程—港口、海岸及近海工程] TH215[交通运输工程—船舶与海洋工程]