Comparison of statistical counting methods in SHM-based reliability assessment of bridges

Structural condition assessment technology has gained widespread applications for providing desired solution to assess safety and serviceability of civil engineering structures. The structural reliability assessment which incorporates structural health monitoring (SHM) data is capable of providing authentic information about in-service performance of the instrumented structure and accommodating uncertainties in the measurement data. Because the peak values of the measurands which illustrate the critical condition/status of the structure are random in nature, it is important to adopt appropriate statistical counting methods to extract favorable peak values for reliability assessment. Several algorithms, such as the sampling method, the peak counting method, and the pointwise counting method, have been proposed for peak counting. In the present study, different statistical counting methods for the selection of peak data targeted for SHM-based reliability assessment of instrumented bridges are examined and compared, through the application of the above methods for the purpose of constructing peak-stress histograms and formulating probability density functions by use of long-term strain monitoring data from an instrumented bridge. Peak covering rate is defined to serve as a common basis for relating the amount of peak data and the control parameters for peak counting and helping determine the values of the control parameters in different statistical counting methods which ensure identical peak covering rate. The reliability indices obtained from the different statistical counting methods under the same amount of peak data are also compared.