| Since the1980s, with the rapid development of Chinese economy, a great dealachievements have been made in the field of bridge construction. Meanwhile, likeother contries all over the world, the bridge projects in China are facing the bridgeaccidents. For this reason, many bridge constructions have installed different sizesof health monitoring systems at home and abroad. Up to now, the technology of thebridge health monitoring system installation has already been well developed. Afternearly30years of monitoring, a large amount of data has been accumulated, how tohandle these data is becoming a hotspot of research gradually. However, the theoryof modeling and analysis of monitoring data has not yet been developed to maturity,therefore, how to cope with the huge monitoring data accurately and timelybecomes a key process of condition assessment and performance prediction ofbridges.The data acquised by the bridge health moritoring systems usually has a strongrelationship with the time. It is well known that the method of time series analysistakes the data itselves as a starting point, and explores the intrinsic law inside theseries. If we use the time series analysis method to establish models of the healthmonitoring data of structures, then the global situation of the bridge constructionwould be easily seized, and the prediction of the future state would be possible.Therefore, in this thesis, taking data-driven modeling as an objective, the feasibilit yof the application of time series analysis method to modeling and analysis of bridgemonitoring data is discussed. The basic ARMA (Auto-Regressive Moving-Average)model in time series anlaysis is selected to model and analyze the health monitoringdata of bridges, and the reliability indices of bridges are calculated based on thebuilt ARMA model. The main contents are as follows:First, the key steps of the building of ARMA model are described in detail, andthe Levenberg-Marquarat optimization algorithm is introduced in the parameterestimation of the ARMA model to improve and optimize the parameters estimatedby the common algorithm. Based on the vertical acceleration of the girder of YitongRiver Bridge, use of the improved estimation algorithm is made to build the ARMAmodel, to analyze and predict, and to verify the feasibility of the time series analysismethod. Compared with the standards of Ontario Highway Bridge Design Code(OHBDC-1995), which provides the limit value of acceleration, the drivingcomfortness level of Yitong River Bridge is assessed.Second, in combination with the data obtained by the inclinometer of YitongRiver Bridge, the deflection of the girder is calculated in MATLAB. The maximum deflection of the girder is picked up, and the ARMA model is established to processand analyze the deflection data. After that, the statistical analysis of the deflectiondata is performed, and the kernel density estimation method is used to estimate theprobability density function of the deflection data. The reliability indices of thegirder of Yitong River Bridge are calculated using MCMC (Markov Chain MonteCarlo) simulation method corresponding to the serviceability limit states.Meanwhile, the impact of the limit value of the vertical deflection of bridgestructures on the reliability assessment for serviceability limit states is investigated.Last, in combination with the data obtained by the pressure rings on thesuspenders of Yitong River Bridge, the key components are considered, the cableforces are identifed, and the ARMA model is established to process and analyze thecable forces. Use of kernel density estimation method is made to estimate theprobability density function of the suspender force data, and the reliability indicesof three suspenders of Yitong River Bridge are calculated using MCMC (MarkovChain Monte Carlo) simulation method. Meanwhile, the effects of the variations ofstrength and diameter of the strands on the reliability assessment of the suspendersare discussed. |
PDF Full Text Request