CA mortar void identification for slab track utilizing time-domain Markov chain Monte Carlo-based Bayesian approach

Abstract

This paper investigates the feasibility and practicability study on the use of Markov chain Monte Carlo (MCMC)-based Bayesian approach for identifying the cement-emulsified asphalt (CA) void of the slab track system utilizing the measured vibration data. A newly developed model class identification algorithm was extended and integrated with the MCMC-based Bayesian approach for the first time to identify the CA mortar void that may partly extend to a neighborhood region. Not only the most probable values of the scaling factors to the mortar stiffness can be calculated, but also the damage probability of model parameters using the posterior probability density function (PDF) can be estimated, and the void can be clearly identified by the MCMC-based Bayesian approach. The proposed methodology was experimentally verified and positive outcomes were obtained. The detection results illustrate that the proposed method not only can successfully assess the void location of the CA mortar but also provide the information of the damage severity, and the posterior PDFs of model parameters can be also calculated by using kernel density estimation to quantitatively describe the uncertainty of the model.