第十届（2019）海峡两岸高校师生土木工程监测与控制研讨会

197 / 2019-07-31 17:13:49

Stochastic Modeling of Stay Cable Force Using Monitoring Data From a Cable-Stayed Bridge

Cable force; temperature; SHM; cointegration; stochastic modeling.

Plenary track > 土木工程结构健康监测

This paper analyzes one year’s acceleration and temperature data of a stay cable from the Structural Health Monitoring (SHM) system of a cable-stayed bridge. The cable force is calculated from acceleration data by frequency method, and modeled as potentially non-Gaussian and non-stationary stochastic time series. It is determined that the cable force and temperature data are both non-stationary time series with daily and seasonal periodic fluctuations. The cointegration relationship between the cable force and temperature is then applied to obtain stationary cable force residuals by linear fitting. The marginal distribution of the residuals are subsequently determined. The time series of mean value of the cable force residual is obtained in separated 30 minutes. The autocorrelation and partial autocorrelation function of the derived mean value series are analyzed. Then the stochastic AR model of the mean value series was established. The fitting value of the AR model is subtracted from the cable force residuals of same time period to remove the daily fluctuation of cable force. It turns out the residuals is close to white noise time series. The marginal distributions of the model residuals can be fitted best by t Location-Scale distribution. The corresponding model residual threshold to 1% is then derived. Taking the difference between the actual bridge monitoring value and the predicted value of AR model as the bridge damage-sensitive features, the health condition of bridge can be monitored continuously.