Research On Bridge Structure Damage Detection Based On The Staitstical Characters Of Time-domain Response

With the rapid development of modern economy, the traffic load that bridge inservice has to bear increases gradually. Meanwhile, external harsh environment leads tocontinuously aging problem of the materials, accompany with the easy occurrence ofcumulative damage for bridge structure, so as to result in the insufficient carryingcapacity. Considering the safety and economic factors, it is necessary to carry outdamage detection and health monitoring of the existing bridge structure, learn aboutthe bridge health status, ensure the safety of the structure of the operation, and reducethe collapse accident.Currently, vibration-based damage detection is an area of wide attention andresearch activity, as it does not affect the normal work of the structure and it can realizelong-term or on-line monitoring and it’s easy and convenient operate. But mostly peopleuse modal analysis method to get the modal parameters of the system, and to identifythe damage then. Modal analysis theory has a good theoretical basis, and relatedresearch has got a not bad effect of damage detection, but it still need to solve a seriesof problems such as power leakage. In this paper, by the application of statistical analysisprocedures such as the time series analysis, we use the time-domain response(acceleration-time histories) to identify damage directly. Compared with modal analysis,this kind of method is more convenient, at the same time to avoid the loss of usefulinformation. The specific research work is shown as follows:1. The regression coefficients of Time series model describe how the structureresponse of a certain moment is influenced by a former time, and reflect the inherentcharacteristics of the structure. Further, we can establish the relationship between theregression coefficient and the structural physical characteristics. Therefore, the structureof the damage can be reflected by the change of the regression coefficients. Based on the regression coefficients of time series model, a weighted distance index is presentedin this paper. The numerical example shows the index can identify the damage, but whenat a low damage degree, the recognition effect is not ideal. Meanwhile, the index can’tlocate the damage.2. The autocorrelation coefficients show the correlation between the randomsignal’s former and latter sampling times, whose description of the system is similar tothe time series regression coefficients. In this paper, damage detection was carried outbased on the first-order autocorrelation coefficient of the vibration signals. Comparedwith the weighted distance, the new index is more sensitive to small damage, but alsocan not realize damage location.3. In this paper, we compute the sum of the squares of the signal value of a certainpoint on the structure in an appropriate period, written as SS. Before and after thedamage, the change of the SS of a damage position is bigger than other position. Basedon this principle, the damage location is carried out on the former numerical example,the results shows that the index for identifying the damage location has a good effect,and can reflect the damage degree, but the index cannot be used for online monitoringof structure damage.Considering the capacity of the above three indexes of damage detection andlocation, the damage identification can be operated in two steps: first, the weighteddistance or autocorrelation coefficient is used for damage detection, then, the index ofSS is used for damage location.