Deterioration prediction of bridge by Markov chain model and Bayesian theory

Abstract

This manuscript presents a bridge deterioration prediction method by using Markov chain model and Bayesian theory. Markov chain model works by defining discrete condition states and accumulating the probability of transition from one condition state to another over discrete time intervals. The probability of transition is generally expressed by the matrix. Though the previous studies have predicted the bridge deterioration by developing deterioration curves by using the Markov chain model, the predicted value will not be necessarily suitable for the measured value in the future. Therefore, this study demonstrates a method to predict deterioration progress as a prediction interval by taking account of the uncertainty by the Monte Carlo simulation. In addition, the method to update the prediction interval after the inspection is developed by Bayesian theory. This research was developed by using inspection results of existing bridges in Japan, and the proposed mechanism is convenient for bridge engineers to take rational decisions on the maintenance management plan of steel bridge infrastructures.