Damage Detection Of Bridge Structures Based On Relative Variation Of Wavelet Total Energy

Damage of bridge structure would affect its durability and safety, so operational monitoring,inspection and maintenance of the bridge were attached great importance by all countries in recent years. Main damage detection methods and their limitations were discussed in detail.Dynamic response signals were acquired through dynamic measurement. The wavelet coefficients at different scales were gathered after dealing with the signals at every measured point of structure by using the discrete wavelet transform. Then the wavelet total energy was calculated by summing up all the wavelet energy at each scale. Setting the intact structure as the baseline state, the damage detection method was put forward based on relative variation of wavelet total energy. Bridges were inclined to be damaged under loading and environmental effect. Especially,crack and stiffness reduction in main girder was typical in Girder Bridge,and corrosion and fatigue fracture of the hangers were common in arch bridge. Numerical models of steel girder and steel tube arch bridge were established for damage detection using the proposed method. Real girder and arch bridge model were made in lab,and various damage conditions and parameters were simulated for experimental research. The research results are as follow:(1) By comparing numerical analysis with experimental research of single beam and steel tubular arch bridge, it can be verified that the proposed damage detection method can not only detect single element, node damage and multi-nodes damage of the beam but also can be applied to damage location of single and double damages of hangers in arch bridge at one side.(2) The proposed damage index is sensitive to the damage level to some extent. In numerical analysis, the relationship between damage index and damage level can be fitted, then the damage level is predicted. But in real application, the detection result is not obvious for lower damage as for measuring error and external disturbances, and there still exists further improvement.(3) The proposed damage index shows great robustness to wavelet type, signal length and decomposition scale. As for different impact position, it can also identify the damage place, but the results show some difference. If the impact load is away from the damage location, detection result will be better. The proposed damage index has some requirements for load types, and impact load will be the best choice. What’s more, the multi-damage detection has some limits under moving load excitation. The detection results are affected by the speed of moving load,and higher speed will give better results.