Multi-stage Bridge Damage Identification Based On RBF Neural Networks

With the rapid development of modern transportation, the requirements of security and operation of bridge structure are becoming higher and higher. It becomes an important question need to be solved urgently that the bridge structure and component damage could not be discovered accurately in time. The bridges will gradually wear down with the reduction of bearing capacity under the effect of the natural environment and applied environment. This will cause substantial loss of people’s life and their properties. Therefore, it’s very important to analyze the technology of bridge damage.Structural damage will cause dynamic characteristics to change correspondently. So if the mapping relationship between structural damage-and changes of dynamic characteristics can be established,the darnage can be identified using dynamic test of the structures. The neural network technique has great superiority in identifying the damage of structures for its strong non-linear mapping ability and anti-interference capability. On the basis of collection and analy si s of the data ab out structural damage identification and artificial neural networks, RBF neural networks method was applied to detect bridge damage with the help of ANSYS and MATL AB. The multi-stage damage identification approach based on neural net works has b een raised in this paper.The multi stage damage identification approach is divided into three steps. Firstly, damage anomalous filter whichis set up by RBF neural networks has been used to alarm the damage in bridge structure. Secondly, the location of the element damage is determined bythe neural network with inputting the combined damage index Finally, the damage degree of the element is determined by the neural network with inputting the change rate of squared mo dal frequency. The result indicates that the multi-stage damage identification approach based on neural networks can alarm the damage. The location and the damage degree of bridge structure can be determined within a certain range of measurement noise.