Estimation of probability density function of long-term strain measurement for reliability assessment

摘要

Cable-stayed Ting Kau Bridge in Hong Kong is instrumented with a long-term structural health monitoring system called Wind And Structural Health Monitoring System(WASHMS)which has collected voluminous data on structural responses as well as the imposed loadings.Obviously,the probability density functions(PDFs)of the measured structural responses under traffic loadings,wind loadings,and environmental effects are indicative of performance and condition of the bridge,and they play a vital role for probabilistic structural condition assessment.Among a variety of measurands,the strain response is probably the most important as it directly indicates the safety reserve of the component.This paper presents the reliability-based condition assessment of bridge components directly using PDF of strain measurement data.A non-parametric density estimation approach based on Parzen window is first applied to estimate the density function of strain data at every measurement point.Then structural reliability method is used to determine the reliability index and failure probability based on the estimated PDF of strain data and the PDF of design stress capacity.Reliability indices are useful parameters to evaluate the damage condition of typical members and to assign the level of maintenance action.Some preliminary results of application to the reliability assessment of instrumented structural components in cable-stayed Ting Kau Bridge are presented by using long-term strain monitoring data.