Some SHM Benchmarks Of Bridges And The Model Updating Based On Substructure And Response Surface Methods

It is well known that a well-defined and precise analytical model of bridge is very important, and it is the research foundation for damage identification, safety evaluation and life prediction in bridge structural health monitoring (SHM). In this paper, the bridge SHM benchmark problem and model updating are studied and discussed. The main content of this dissertation is as follows:(1) Several bridge model benchmark experimental systems with increasing complexity are designed and constructed for bridge SHM benchmark study. They are a steel truss bridge model, a single span and simply supported grid bridge model, a single tower cable-stayed bridge model and the SHM laboratorial model of a long-span cable-stayed bridge. Each bridge benchmark system includes the following three steps:a) the design and realization of bridge physical model; b) the establishment of model experimental system, then, various static and dynamic experiments are carried out on the model experimental system; c) the initial Finite Element Model is constructed, and the benchmark numerical model of physical model can be achieved by updating the initial Finite Element Model (FEM) based on the experimental data of the bridge model. These bridge benchmark model experimental systems can provide an ideal and convenient experimental platform for bridge SHM study.(2) The parameters selection methods of sensitivity analysis and variance analysis with respect to design parameters are performed for model updating based on FEM simulation of a cable-stayed bridge laboratorial model to investigate the reasonability. Results show that the sensitivity analysis has clear conception and can be easily applied to practical engineering; variance analysis based on testing samples generated by design of experiment (DOE) in the global design space could quantitatively show the significance levels of design parameters to modal eigenvalues.(3) To investigate the problems of multi-parameter and limited testing data in bridge model updating, various numerical simulation of model updating are implemented based on different types and amounts of information and their combination. Results indicate that the quantity and quality of characteristic information for model updating have significantly impacts on the modification results, and the model updating by utilizing integrated data of different characteristic information can considerably improve the updated results. Meanwhile, substructure model updating approaches are proposed for existing and under-construction cable-stayed bridges. The numerical simulation and experimental demonstration indicated that the proposed methods are valid and reliable with high efficiency.(4) The response surface method based on radial basis functions (RBFs) is investigated to model the input-output system of large-scale structures. As a methodology study, the complicated implicit relationships between the design parameters and response characteristics of cable-stayed bridges are employed in the construction of response surface (RS) models. Several methods of DOE are investigated to study the sample selection for RS modeling, and the shape parameter of RBFs for RS approximation of the interested features of cable-stayed bridges is studied. Some suggestions for selecting samples and optimal shape parameters are proposed. The RS methods based on approximate functions of different RBFs are investigated. Meanwhile, the commonly used RS method of polynomial function is also carried out for comparison. For each RS method, the approximation accuracy is evaluated and the anti-noise ability is also discussed. Results demonstrate that RS methods based on RBFs have high approximation accuracy and exhibit a better performance than that based on polynomial function.(5) The RBFs-based RS method is applied to model updating of large and complicated bridge. Firstly, the numerical simulation of model updating based on RBFs-based RS approach is carried out on a space truss bridge model; then, simulation analysis and experimental verification are performed on a single tower cable-stayed bridge model and a scaled model of a long-span cable-stayed bridge. Simulation analysis shows the updated results have high accuracy, and the model updating based experimental data indicates that the results are reasonable. It is demonstrated that the proposed approach is valid for model updating of large and complicated bridge structures.