Updating Probabilistic Model of Traffic Loads on Bridges Using In-Service WIM Data

Abstract

These days, a Weigh-In-Motion (WIM) system enables us to estimate traffic loads on a bridge based on site-specific traffic environment. However, since the traffic environment of a bridge may change significantly during its service life, it is necessary to monitor the in-service traffic environment and to update the probabilistic model of traffic load. This study aims to develop a methodology to update distribution parameters of random variables in the probabilistic traffic load model by Bayesian inference. Three main methods are used together to establish the updating methodology: conjugate prior distributions, Bayesian linear regression, and Gibbs sampling. The proposed method is demonstrated by numerical examples using WIM data from two sites in South Korea.