Ride comfort is among the most critical indicators of service quality of high-speed trains. In this study, a Bayesian approach is developed for successive evaluation of ride comfort in terms of Sperling index using online monitoring data acquired by an FBG-based monitoring system instrumented on a high-speed train during its route operation. The posterior probability density function (PDF) of Sperling index and the relevant statistical parameters are updated evolutionarily when new monitoring data is available. A proximity measure technique is applied to identify the datasets with outliers, and the posterior PDF of Sperling index is refined with a better confidence interval after removing the datasets with outliers in the Bayesian inference.
doi: 10.12783/SHM2015/262