Reliability Updating And Bayesian Prediction Of Bridges Based On Proof Loads And Monitoring Data

Structural health monitoring is one of present research hotspots in the field of civil engineering. The research on structural health monitoring generally experiences two stages. The first stage, falling in the mature stage, is to install sensors on the structures and conduct much research on the data transition system, data acquisition technology, system integration technology and other aspects. The second stage is mainly the application of health monitoring information. Novel monitoring systems used in structural engineering contain sensors providing a large amount of monitored data. Proper handling of the continuously provided monitored data is one of the main difficulties in the field of structural health monitoring. In this dissertation, based on Bayesian updating and prediction theory, the structural inspection data (e.g. proof loads and proof load effects) and monitored data, systematic research on reliability updating and prediction of bridge members or bridge system was carry out.The main research contents of this dissertation are described as follows:This paper presents the reliability updating method based on proof loads and resistance degradation model of bridge members. With the truncated method and Bayesian method, the effects of proof loads, proof load effects and resistance degradation model on the reliability of bridge members are analyzed.This paper presents the reliability prediction method of bridge members based on Bayesian dynamic linear model. With inspection data or monitored data, the1-order polynomial function, AR(1) model and ARMA(1,1) model are respectively adopted to build the corresponding Bayesian dynamic linear models (BDLM). And the model monitoring mechanism of BDLM is studied. Considering the diversity of BDLM, the combined BDLM is built based on the multiple BDLMs. Finally based on the single BDLM or combined BDLM, with first order second moment (FOSM) method, the reliability of structural member is predicted.This paper presents the reliability prediction method of bridge members based on Bayesian dynamic nonlinear model (BDNM). Mainly with monitored data,2-order polynomial function and3-order polynomial function are respectively adopted to build the corresponding BDNM. The simulation processes of BDNM are handled with the following two methods. One method is to transform the built BDNM into approximate BDLM with Taylor series expansion technique, the monitoring mechanism of approximate BDLM is also studied. The other method is to directly simulate the processes with Markov Chain Monte Carlo (MCMC) method. Finally based on the built BDNM, with FOSM method, the reliability of structural members is predicted.This paper presents the real-time on-line reliability prediction method of bridge members based on mixed Gaussian particle filter (MGPF). Firstly monitored data-based dynamic model is built, and then the MGPF is introduced. Based on the particle filter method and dynamic model, the distribution parameters of state variables and one-step prediction distribution are predicted. The MGPF solves the problem of particle simulation degeneracy through the resampling method. Finally with FOSM method, the reliability of structural members is predicted.The paper presents the reliability updating and prediction method of bridge system based on proof load effects and monitored data. The proof load effects of structural members are obtained with MIDAS software. Firstly the distribution of stress threshold (genaralized resiatance distribution) is updated with proof load effects. Then the moinitored data(load effect data)-based mixed gaussian particle filter(MGPF) is built, and then based on the updated stress threshold distribution and load effect MGPF, the reliability updating and prediction of structural members is solved, finally with the system reliability method, the reliability of structural system is updated and predicted, and Tianjin Fumin bridge is provided to illustrate the feasibility and application of the given models and methods.