Research On Bridge Reliability Based On Vine Copula And Bayesian Dynamic Models

Bridge health monitoring systems have produced a huge amount of data in the long-term service periods.Through reasonably handling with these data,the service statuses of the existing bridges can be effectively understood.Based on these data,how to predict structural dynamic reliability with considering the existing uncertainties of bridge system,the correlation between failure modes and so forth,is one of the key and urgent scientific problems in the field of structural health monitoring.In this dissertation,based on the dynamic monitored data,the reliability of bridge systems is systematically studied with vine copula models and Bayesian dynamic models.The main research contents of this dissertation are described as follows:(1)The time-invariant reliability analysis method of bridge systems is presented based on vine copula model considering the nonlinear correlation of failure modes.With pair-copula models and vine copula models,the nonlinear correlation of multi random variables is studied,further,with structural health monitoring data and First Order Second Moment(FOSM)reliability method,the influence rules of nonlinear correlation between failure modes on the invariant reliability of structural system are obtained.(2)The time-variant reliability prediction method of bridge structures is presented based on the decoupled monitored data and Bayesian dynamic models.With moving average method,the coupled monitored data of bridge structures is decoupled.With Kendall rank correlation coefficient,the correlation between decoupled data is analyzed.Based on time series models,the Bayesian dynamic models including T distribution and normal distribution are built to predict the trend of the decoupled data.Further,with FOSM method,the dynamic reliability of bridge structures is predicted.(3)The dynamic reliability analysis method of bridge systems with considering the time-variant nonlinear correlation of failure modes,is presented.Based on the health monitored data of bridge structures,the Bayesian dynamic vine copula models about multi monitored variables which combine the vine copula models and Bayesian dynamic models,are built.Further,with FOSM method,the dynamic reliability of bridge systems considering the time-variant nonlinear correlation of failure modes is solved.And the proposed method is compared with the existing dynamic reliability analysis method of bridge systems without considering the time-variant nonlinear correlation of failure modes for illustrating the feasibility.