Research On Finite Element Method Updating For Beam Bridge Based On Static And Dynamic Response Surface

As an important part of the transportation network,bridges are the key infrastructure for economic development and people's livelihood improvement.There are many bridges in service in China,most of which are small and medium-span bridges,to ensure the smooth flow of transportation networks,scientific assessments of the damage and health status must be made for existing bridges.The updating of finite element model is used as basic work for structural damage identification and health monitoring,and obtains a reference model that can reflect the actual stress of the structure.The traditional model updating method optimizes by repeatedly running the finite element model,there is the problem of low computational efficiency and convergence difficulties.The model updating method based on response surface uses a function to fit the finite element model and optimizes the mathematical model obtained by fitting,which greatly improves the optimization efficiency.In this thesis,after reviewing the development history of model updating and introducing the basic theory of the model updating method based on response-surface,a model updating method based on the static and dynamic response surface was adopted to perform the finite element models of numerically I-beam,full-foot track beam,and 3-span continuous beams respectively,contrasted the updating effect of static method,dynamic method and joint method,at the same time,it discussed the function and applicable conditions of constructing objective function with dynamic weights.Get the following basic conclusions:Firstly,the response surface obtained by the center composite design and the complete quadratic polynomial has good accuracy,meets the requirements of simple-supported and continuous-beam structural analysis,and does not require the use of higher-order polynomials with large amounts of computation or adding too many sample points.Second,the static method has a good static updating effect,static updating effect obtained by dynamic method is poor,and the combined method has the best comprehensive updating effect.Comparing to the results of the power test,the static test value has a higher accuracy,and the dynamic updating effect of the static method depends on the deviation of only the power-sensitive parameter from its initial value.If there is a large error in the dynamic response,not only the power updating effect of the combined method cannot be effectively improved,but also the static updating effect will be reduced.Finally,when there is a large difference in the relative error of each response,the dynamic weighting method is more effective in improving the updating effect.The model updating essence is a multi-objective optimization problem.If there is a large error in one or several responses,it will have an adverse effect on the updating effect,the dynamic weighting method can reduce the target function residual by assigning a smaller weighting to a larger error and improve the effect of updatings.In general,the degree of error in different types of responses is inconsistent,if the difference is large,equal weights cannot be used to obtain good updating effects,if the difference is small,dynamic weights lose their significance.