基于逆高斯随机过程的混凝土梁桥抗力退化模型  

Resistance degradation model of concrete beam bridgebased on inverse Gaussian stochastic process

作　　者：徐望喜 钱永久[1] 金聪鹤 龚婉婷 Xu Wangxi;Qian Yongjiu;Jin Conghe;Gong Wanting(School of Civil Engineering,Southwest Jiaotong University,Chengdu 610031,China)

机构地区：[1]西南交通大学土木工程学院,成都610031

出　　处：《东南大学学报（自然科学版）》2024年第2期303-311,共9页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金资助项目(51778532)。

摘　　要：为准确体现混凝土梁桥抗力退化的时变性与随机性,运用逆高斯(IG)随机过程建立了结构抗力退化模型.基于现场检测数据,采用贝叶斯更新理论对IG随机过程进行实时更新.针对IG随机过程中的参数因先验分布与后验分布非共轭而难以采样估计的问题,提出了混合Gibbs采样法.结合数值案例论证了该方法的可行性,并将其应用于某一混凝土梁桥的抗力预测分析.研究结果表明:IG随机过程可利用现场检测信息对桥梁抗力退化过程实时更新;混合Gibbs采样法可解决IG随机过程的高维参数估计问题,克服了形函数中指数q值的经验性给定缺陷;随着服役年限的增大,桥梁结构的累积退化量逐步增大,当桥梁服役年限为60 a时,其累积退化量为服役年限为30 a时的3.48倍.与Gamma随机过程相比,IG随机过程避免了部分初始参数的假定,可得到更为精确的桥梁抗力退化模型.To accurately reflect the time-dependence and randomness of the resistance degradation of concrete girder bridges,the inverse Gaussian(IG)stochastic process was used to establish the structural resistance degradation model.Based on the field test data,the IG stochastic process was updated in real-time by Bayesian updating theory.In addition,a mixed Gibbs sampling method was proposed to address the difficulty in estimating high-dimensional parameters in the model due to the non-conjugate prior and posterior distributions.The feasibility of the method was demonstrated by numerical cases,and the resistance prediction of a concrete girder bridge was conducted.The research results show that the IG stochastic process can employ the field detection information to update the bridge resistance degradation process in real time.Moreover,the mixed Gibbs sampling method can solve the problem of high-dimensional parameter estimation,and overcome the defect of empirical given value of exponential q in the shape function.With the increase of service life,the accumulated deterioration of bridge structure gradually increases.When the service life of the bridge is 60 a,the accumulated deterioration is 3.48 times that of 30 a.Compared with the Gamma stochastic process,the IG stochastic process avoids the assumption of several initial parameters,and thus an accurate bridge resistance degradation model is obtained.

关 键 词：桥梁工程 抗力 逆高斯(IG)随机过程 混合Gibbs采样 

分 类 号：U441.2[建筑科学—桥梁与隧道工程]