The Research On The Validity Of Extreme Response Estimation Method For Highly Reliable Structural System

For the high reliability structure system,because of the internal contradiction between the uncertainty of material performance,geometric parameters,the effect of external load effect and the highly reliable structural system requiring probabilistic metric structure reliability,the study of the general method for the failure probability estimation of the high reliability structure system needs to be come over.In order to solve the reliability analysis of the highly reliable structural systems,there exists two key problems: the joint failure probability and the efficiency of the system Failure probability.In order to address the first problem,the joint failure probability problem of high reliability structure system is transformed into the multivariable distribution estimation problem of structural extremum response.In the present,the multivariate joint probability distribution is decomposed into Copula function and marginal distribution through Copula theory.The appropriate Copula function can be studied by comparison and analysis.The multivariable distribution estimation of the extreme response of high reliability structure system is decomposed into the marginal distribution of each component.According to the proposed generalized extreme value distribution and the shifted Generalized Logarithmic normal distribution,the response extreme distribution on the component level is fitted respectively,and the validity of the distribution of the components is compared and analyzed to solve the problem of the failure probability estimation of the high reliability structural system.In order to address the second problem,based on the extreme value theory,the extreme value distribution of the structural response under the external load can be estimated efficiently by the accelerated simulation method based on the generalized extreme value distribution and the shifted Generalized Logarithmic normal distribution.At the same time,the estimation of the Gumbel Hougaard Copula function though the parameter estimation method based on the Kendall-tau rank correlation coefficient is discussed in this paper.The efficiency of the algorithm accelerates the estimation efficiency of failure probability from two different angles.In view of the above problems,the basic principles and processes of the selected methods are introduced in detail.Finally,the method is used to estimate the extreme value of the non-Gauss vector response process for the frame structure containing linear and nonlinear structures and the ideal model of the substation under static and dynamic external loads.The failure probability of the structure system is compared and the numerical accuracy of the selected method is carried out according to the calculation results.A comparative study of computational efficiency is given,and suggestions for engineering application are given.The research in this paper has a certain reference value for the reliability evaluation and design of the highly reliable structural system.