Study On Time Series Analysis Based Cable-stayed Bridge Damage Detection Method

Bridge occupies an important position in the national economy, assuming theresponsibilities of resource transportation and commodity circulation. It will causeheavy economic losses and bad social influence once an accident occurs. The bridgewill be under constant effects of normal load, environmental load, corrosion as well asother factors. Long-term effect leads to structure damage accumulation and resistancereduction, causing catastrophic accident in extreme cases. Therefore damageidentification and early warning for the completed bridges are very significant, whichcan identify early structural damage and avoid the occurrence of major accidents.Although there is a lot of existing damage diagnosis methods for bridge structure, yetmost are in the simulation stage lacking in effective verification with experimentaldata. However, it is not realistic to sabotage and study on the actual bridges, thereforeverification of damage identification methods through empirical models is verysignificant. Based on the existing accomplishment in the bridge damage detection,this paper conducts following research with emphasis on bridge structural damageidentification methods:1. Based on analysis of the data of existing cable-stayed bridge design, a kind ofdouble pylon cable-stayed bridge damage identification test model is designed. Thismodel, which is aimed to validate damage identification method can simulate variousdamage conditions, and is convenient to be disassembled and assembled, adapted tovarious test environment.2. A damage identification method based on the time series analysis theory isapplied, and is programmed through the MATLAB software. This method is ofuniversality, which only needs acceleration response under ambient excitation tocomplete the damage identification.3. The verification is carried out on the designed cable-stayed bridge test modelto validate the damage identification method based on time series analysis. Theexperiment is conducted under ambient environment, and four damage cases are setup. The damage identification is applied using response signal acquired by Imote2sensors. Based on the results, the effect of this method is discussed.