Study Of A Wavelet Packet Energy Based Damage Detection Index For Bridges

Bridge is an important hub for the transportation infrastructure. The performanceof bridge structure is of great significance to the transportation system. The bridge isconstantly subjected to various loads during its lifetime, and the damage willaccumulate inside the structure with a continuation of service time. If these problemsare ignored in a long-term, leaving the bridge without promptly maintenance, it willcause catastrophic accidents. Therefore effective early damage identification isbecoming an important direction in the field of bridge health monitoring. The existingresearch techniques usually identify damage by constructing a damage index, whichcan reflect the state of structure system. Based on wavelet packet analysis and waveletpacket energy, a novel damage index and an identification process are proposed. Thispaper studies the following research focusing on a bridge structural damageidentification method:1. Based on the existing damage detection methods and the application ofwavelet analysis theory, a wavelet packet energy based damage detection index (DDI)is proposed. The index is a square root, which is extracted from the sum square ofwavelet packet energy change rate. A damage identification method based on the DDIis programmed through the Matlab software. The wavelet packet transform is used toanalysis acceleration signals. Then damage is detected by calculating the DDI.2. Numerical simulation of a simple supported beam is tested and random whitenoise is applied to the model as a random environmental excitation. The beam issimulated in various damage scenarios and noise levels. The validity of the damageindex is verified by the results. It is shown that the results of damage identification areaffected by both noise and damage extent.3. The index is further experimentally investigated by a double pyloncable-stayed bridge model test in different damage scenarios. Firstly the numericalmodel is tested to get response signals. Based on these signals the effect of damagedetection is preliminary observed. Then experiments are conducted under ambientrandom excitation. These measured acceleration signals are analyzed using waveletpacket transform. Based on the results, the parameter sensitivity of the presentedmethod is discussed.