Modeling of Stress Spectrum Using Long-Term Monitoring Data and Finite Mixture Distributions

Abstract

This study focuses on how to exploit long-term monitoring data of structural strain for analytical modeling of multimodal rainflow-counted stress spectra by use of the method of finite mixture distributions in conjunction with a hybrid mixture parameter estimation algorithm. The long-term strain data acquired from an instrumented bridge carrying both highway and railway traffic is used to verify the procedure. A wavelet-based filtering technique is first applied to eliminate the temperature effect inherent in the measured strain data. The stress spectrum is obtained by extracting the stress range and mean stress from the stress time histories with the aid of a rainflow counting algorithm. By synthesizing the features captured from daily stress spectra, a representative sample of stress spectrum accounting for multiple loading effects is derived. Then, the modeling of the multimodal stress range is performed by use of finite mixed normal, lognormal, and Weibull distributions, with the best mixed distributi...