| Long span prestressed concrete girder bridges have boomed in recent years, butthe problem of bridges’ deflection in service has great influenced the comfortablenessof driving and the structural safety. As the vertical displacement under action ofdeadloads and effect of self-weight, deflection reflects the bridges’ wholecharacteristics, which is the very important parameter in the health-state of existingbridges. The total deflection sigal can be viewed as the superposition of long-termdeflection, annual temperature difference effect, daily temperature difference effectand traffic loads effect, which is obtained by the real-time running state monitoringsystem. So it can correctly assess the work performance and evaluate the health statusof bridges during its service life by rejecting environmental factors effect only byobtaining the deflection value caused by each individual factor.A large number of scholars calculate the corresponding deflections of bridge inthe means of applying the main factors on the bridge through finite elementsimulation. But in the practical engineering, it is difficult to accurately simulate thedegradation situation and environment factors of bridge because of the loss rate of thepretension is unclear, the bridge’s service environment and external factors are randomuncertain and some other factors. Therefore, it is uneasy to accurately calculate singlevalue in the deflections of bridge.Through multi-scale analyse theory, it can be found that the defection of bridgehas the characters of time multi-time scale: the defection caused by live load isdynamic, the defection caused by temperature is dynamic, but it has the characters ofslight variation frequency; the frequency of long time defection caused by concreteshrinkage and creep and prestress loss is close to0. The differentiation amongdifferent defection statistics characters can provide a possibility for the research forbridge deflection signal—in the angle of signal processing, a mathematical theoryneeded to be found to separate the defection signal which belong to discrete-timeseries, the effects of actions can be got in different periods, that is also the singlefactor contributing to the bridge defection.The traffic loads effect is a high-frequency signal in the total deflection sigal,which can be separated by a general classic filtering algorithm. The temperatureeffects and long-term deflection of bridge are low frequency signals, which is very hard to be separated by the filtering algorithm and wavelet decompositionreconstruction algorithm which is the genral signal processing means in science andengineering. Independent component analysis(ICA) is a novel approach for blindsource separation, which can effectively separate the relatively independent signalsform the source.It not only can effectively extract weak signals under strong noiseenvironment but also has good separating when the source vibration signals blendingon the frequency range. Empirical mode decomposition(EMD) is adaptive and notneed set the prior information, which is good used in noise reduction and multi-scaledecompositions, although the leverage point effects is obvious in separating result.The paper has a complementary advantage in using two algorithms, puting forward anew technical ideas that using EMD method to expand one-dimensional tomulti-channels in order to build ICA separating model.The article described the theory of single factor causing bridge’s deflectionsignal. Through the finite element numerical method, the responding simulativedefection signal can be got through applying single factor load on each bridgemodel, after the simulative signal have been multiplied, then the correlationcoefficient between separation results and source signals can be simulated, itsmaximum value reaches0.983, its minimal value reaches0.814; after then themeasured deflection signal can be separated, its maximum value reaches0.926, itsminimal value reache0.8281, the result proves the feasibility and accuracy ofseparation model. |
PDF Full Text Request