Abstract: Traditional damage identification methods are difficult to accurately identify the damage location and extent of irregular-shaped bridge because of the complex structure of the bridge. To overcome the shortcoming of these damage identification methods and the uncertainty of factor influence, the factor effect on damage identification indicator is synthetically analyzed, and a novel damage identification approach for irregular-shaped bridge is proposed based on modal flexibility difference curvature and Support Vector Machine optimized by Genetic Algorithm (SVM-GA). First, the damage identification effect on the indicators (modal shape change, modal curvature difference, modal flexibility difference and curvature) is studied under the impacts of damage severity, number of sensors and modal orders. Second, the two stage damage identification method for irregular-shaped bridge is presented. On the basis of determining the damage location using modal flexibility difference curvature, the damage extent is forecasted and identified based on SVM-GA. Damage identification results demonstrate that the proposed method is feasible and accurate.

Key words: road engineering, irregular-shaped bridge, damage identification, modal flexibility difference curvature, genetic algorithm(GA), support vector machine(SVM)