Damage Identification Research Of Bridge Based On FEMU And SVM

Having experienced decades of development, the theory and methods about damage identification has already been gradually matured, but the research about damage identification is mostly based on a simple correction of the finite element model, whereas the model error has a great influence on the damage identification process. Therefore, in this paper, considering the model experiment of a Combined Cable-stayed Suspension Bridge as an example to update the initial finite element model by using the response surface method firstly, and then use the SVM (Support Vector Machine) to carry out a damage identification research based on the updated finite element model.The main research contents are as follows:1. In this paper, at first, the latest research about damage identification both in China and around the world has been introduced, and then the existing problems have been summarized, and eventually damage identification will be researched based on FEMU (Finite Element Model Updating) and SVM.2. The basic principles and implementation process of damage identification which is based on FEMU and SVM has been expounded systematically, and the key link of theoretical methods has been illustrated, and also at the same time problems along with their solutions have been discussed which were related to FEMU.3. The initial finite element model of the combined cable-stayed suspension bridge was established by ANSYS. Based on the measured vertical displacement and vertical bending frequency and adopting the response surface method, two types of damage conditions and two loading stages were combined to carry on the finite element model updating. The results shows that the updating effect is superior, and can be used in the research of damage identification;4. Carried out the damage identification research based on the updated finite element model and vertical displacement data by the SVM method, the results shows that the effect of damage identification is effective.