Abstract:Based on neural network technique and hanger local vibration measurement, identification of damage location and extent for suspension bridges is discussed. First, damage identification is simulated numerically based on a high-precision FE model validated by experiment. Some potential damage cases for suspension bridges are simulated. BP network for damage identification is trained and tested by taking hanger tension indices as input. The index is constructed by fundamental natural frequency of hanger local mode. The damage locations and extents are indicated by the output vector of the network. Then, model test for identification of some damage cases is carried out on the test model of suspension bridge. The model is specially designed aim at damage identification and health diagnosis of suspension bridge. The both results of numerical simulation and model test show the method is effective and practical. The outstanding feature of the method is that a good result can be obtained by using only fundamental natural frequencies of a few hangers. Because measurement of fundamental natural frequencies of a few hangers is much easier than some other damage-oriented measurement, the method has great practical value.