Hilbert-Huang Transform Method For Damage Identiifcation Of Cable-stayed Bridge With Low Towers

Damage detection technique for bridges is one of the most research active areasin the modern bridge engineering. The scientific level of damage detectiontechniques plays an important role for the bridge construction quality and safety, aswell as the safety in the service period of bridges. However, the attention was onlypaid for the health monitoring and damage identification of bridges in the serviceperiod. Little attention was paid for the damage detection of bridges in theconstruction state. In the construction state, the bridge structure systems would bechanged several times. Its loading conditions are very complicated and changeable.This may result in the serious accidents that would cause the major economic andlife losses. Thus, the damage detection is necessary for bridges in the constructionstate.Hilbert-Huang transform is a new approach of signal analysis and treatmentdeveloped in the recent years. Because this approach can decompound the signalunder its own scale, it is intrinsically different from the traditional Fourier transform.Hilbert-Huang transform has a large advantage in the non-stationary signal analysis.Damage identification problems for a cable-stayed bridge with low towers underconstruction state are investigated in this work. Xijiang Large Bridge on the highwayfrom Jiangmen to Zhaoqing in Guangdong province is taken as an example. TheFEM model is established by the software MIDAS. Then, the modal singles underundamaged and damaged states are obtained from the FEM modal analysis. Themodal signals are input into MATLAB to perform the Hilbert-Huang and wavelettransform in order to estimate the structure damage location and degree. Two resultsby Hilbert-Huang transform and wavelet transform are discussed and compared. Acriterion for the selection of the optimized modal shape is obtained. In addition, adamage identification index is then proposed. The limitations of Hilbert-Huangtransform for the damage identification of cable-stayed bridges with low towers arefound and the solution methods are proposed. Four typical construction conditionsare considered, i.e., biggest double cantilever condition, side span closure condition,second main span closure condition and main span closure condition.The results show that the Hilbert-Huang transform method is valid and has astronger anti-noise feature. Meanwhile, an effective method is found to overcome theend effect by selecting the suitable modal shape. On the other hand, the accuracy of wavelet method depends on the choice of wavelet basis which is inconvenient to setup the standard of damage identification. Moreover, the wavelet method needsdenoising and this may result in the failure of damage detection. Because the historyof Hilbert-Huang transform is short, the theoretical aspect needs further develop.