Cable Force Determination And Finite Element Model Updating Of Tied-arch Bridge Based On Radial Basis Function Neural Network

Follow the further development of finite element technique, finite element (FE) model updating has been widely applied to different engineering structures over past two decades. Parameters based FE model updating method is the process of decreasing the gap between the measured value and the calculated value of the FE model through the selecting and updating of design parameters, and establish the finite element model to reflect and predict the state of the existing structure.In the civil engineering field, an error between actual structure and the FE model is always caused by certain assumptions in the construction of FE model and effects from external load and environment. In this case, the FE model is usually unable to reflect the actual state of the structures. Therefore, the FE model should be updated in order to ensure it is capable of expressing the actual structural condition.In recent years, the health monitoring and the safety status assessing of large cross bridges have become hot research field at home and abroad. FE model updating can provide a reliable and effective model for inversion of structure working and identification of structural damage including site and extent, etc. Therefore, FE model updating has important engineering significance.The main work includes:1. Review the course of development of the finite element updating, and summarize the finite element model updating methods, to analyze the advantages and disadvantages of each method;2. The introduction of using radial basis function neural network method as finite element updating tool of this thesis, considering FE model updating characteristics, analyze the input and output parameters of radial basis function neural network and select the structural parameters to construct the network. Determine the technology roadmap using the radial basis network to update the finite element model and calculate construction force finally.3. Assuming an existing state, update a prestressed concrete simply supported girder finite element model, and verify the feasibility of the proposed method.4. Determine the tied arch bridge construction force through RBF neural network in order to achieve extended finite element model updating application.5. Establish the finite element model for beam arch combination bridge using grillage method based on construction drawings and employ the principal component analysis to preprocess the input and output samples of network. The measured response of actual bridge is used to update the finite element model by RBF network examined. The gap between the calculation results of updated finite element model and measured value is much more decreased than the original design finite element model.