A Bayesian Estimation Method to Improve Deterioration Prediction for Infrastructure System with Markov Chain Model

Title:

A Bayesian Estimation Method to Improve Deterioration Prediction for Infrastructure System with Markov Chain Model

Authors: Kiyoshi Kobayashi (Graduate School of Management, Kyoto University, Kyoto 606-8501, Japan);
Kiyoyuki Kaito (Department of Civil Engineering, Osaka University, Suita, Osaka 565-0871, Japan);
Nam Lethanh (Institute of Construction and Infrastructure Management, Swiss Federal Institute of Technology (ETH), Zurich 8093, Switzerland).
Issue: Vol 1, No 1 (2012)
Pages: 1-13
Section: Research Paper
DOI: 10.7492/IJAEC.2012.001
Citation: Kiyoshi Kobayashi, Kiyoyuki Kaito and Nam Lethanh (2012). "A Bayesian Estimation Method to Improve Deterioration Prediction for Infrastructure System with Markov Chain Model." International Journal of Architecture, Engineering and Construction, 1(1), 1-13.
Publisher: International Association for Sustainable Development and Management (IASDM)
Abstract: In many practices of bridge asset management, life cycle costs are estimated by statistical deterioration prediction models based upon monitoring data collected through inspection activities. In many applications, it is, however, often the case that the validity of statistical deterioration prediction models is flawed by an inadequate stock of inspection dates. In this paper, a systematic methodology is presented to provide estimates of the deterioration process for bridge managers based upon empirical judgments at early stages by experts, and whereby revisions may be made as new data are obtained through later inspections. More concretely, Bayesian estimation methodology is developed to improve the estimation of Markov transition probability of the multi-stage exponential Markov model by Markov chain Monte Carlo method using Gibbs sampling. The paper concludes with an empirical example, using the real world monitoring data, to demonstrate the applicability of the model and its Bayesian estimation method in the case of incomplete monitoring data.
Keywords: Bayesian estimation, Markov chain Monte Carlo, Gibbs sampling, statistical deterioration prediction, Markov chain model, infrastructure management.
Full Text: PDF HTML