Structural Identification Of A Long-span Concrete-filled Steel Tubular Arch Bridge Based On Static And Dynamic Tests

Mode updating and parameter identification belongs to the category of structural identification, which is conducted by pairing the model with field test results, and is of significant importance for condition assessment and structural health monitoring. Due to its complexity inherent in long-span arch bridges, plenty of epestimic uncertainty may be induced during the FE modeling procedure. This paper discussed the two structural identification methods to identify uncertain parameters using the field test results. The first method is based on one optimal model which is best coincide with tests data. Significant updating parameters are selected by sensitivity analysis and model calibration is carried out based on the optimization of objective function which denotes the relative difference between FE model and test data. The second method is based on the theory of uncertainty and probability theory. Model calibration is based on a set of candidate models to simulate the real structure, during which parameters are identified according to the posterior probability calculated by the bayes formula. Major contents of this research are drawn as follows:(1) Field tests including ambient vibration tests and static loading tests carried out by loaded-trucks were summarized. Results of elevation observation showed that real structure is in strict accordance wit h design blueprints. Uncertainties due to geometry parameters are mitigated by precise modeling guieded by complete field investigation. As the early stage of structural identification, field test results laied the foundation of model calibration and param eter identification.(2) Ambient vibration tests were carried out on both bridge decks and arch ribs and two kinds of operational modal analysis methods were employed. Before using the complex mode indicator function(CMIF) method, two kinds of pre-processing techniques were used. In order to research the impact of inputs considering acceleration signals of reference point in each sub-structure tests, modal parameters identified were compared. When constructing the frequency response function of the whole structure, considering different kind of auto spectrum may lead to misidentification of modal parameters. Besides, spatial mode coupling is obvious according to modal analysis on arch ribs and b ridges decks.(3) During model calibration, uncertain parameters are identified through optimization of objective function. In order to avoid iterative search trapped into local minima, model calibration based on artificial intelligence are employed. Afterwards, it was applied to numerical simulation model and FE model of Laihua Bridge. In order to overcome the shortcoming of genetic algorithm in local search ability, simulated annealing algorithm and a hybrid algorithm were also investigated.(4) After the recognizition of error compensation, multi-model identification based on bayes theory was used in this research and uncertain parameters were identified by bayes estimation. According to this method, both the maximum likelyhood of uncertain parameters and its distribution could be acquired.