ABSTRACT
In Japan, many bridges have stood for a long time since they were constructed and aging is progressing by corrosion and other factors. In order to maintain the intact of bridges, efficient and reliable inspection techniques for screening the inspection points are required. In this study, we discussed a damage identification method focused on the degree of bridge member deterioration and the position of damaged members by using a Self-Organizing feature Map (SOM), which is a type of neural network. It is considered damage by changing the thickness of the stiffening girder to represent a deteriorated member. It is used the power spectrum as the learning and recognized data for the SOM. In order to calculate the power spectrum, an FEM analysis model was created based on a small arch bridge experimental model. In conclusion, it was found that there is the possibility of performing damage identification using the proposed method.
