Bridge Structural Damage Location Identification under Seismic Action

In order to detect the bridge structural damage caused by earthquake, a location identification method for structural seismic-induced damage based on the pattern recognition in statistical learning theory is put forward. Hilbert-Huang transform is adopted to process the inputting ground motion and structural dynamic responses, and then Hilbert marginal spectrum transfer function is defined as the damage location index in time-frequency domain. Furthermore, the model of damage location identification is set up by support vector machine. Taken one certain cable-stayed bridge as case study to verify the proposed seismic-induced damage location method, and the results show that when the single region of bridge is damaged under seismic action, the proposed method obtains the preferable results of location identification, while the misjudgments still appear under conditions of multi-regions damage cases, but the whole identified rate is still acceptable.