Bayesian Based Linear Modeling for Pavement Smoothness Prediction
暂无分享,去创建一个
Pavement smoothness as indicated by the International Roughness Index (IRI) is a critical component in evaluating pavement performance. Numerous statistical models have been developed. However, because adequate databases to support the development and updating of these models are often lacking, they had some success with certain limitations. A promising solution lies in the use of Bayesian based approach, which considers the uncertainties of the data and allows the combination of engineering judgments and supplemental field monitoring data. Adopting the same influencing factors as those used in the Mechanistic Empirical Pavement Design Guide (MEPDG) smoothness models, Bayesian based linear models are developed in this paper. The coefficients in the MEPDG equations are used as the Bayesian priors, and the posterior is obtained by updating the priors with the actual observed data from the Long Term Pavement Performance (LTTP) database developed in the United States. The distributions of the regression coefficients are generated and pavement smoothness is predicted with available new data sets. The application shows that the Bayesian based method provides promising results and is useful for modeling pavement performance.