TRANSITION PROBABILITIES IN MARKOV CHAIN FOR BRIDGE CONDITION PROJECTION
Markov Chain is commonly used in management for infrastructure facility systems, to model their lifecycle performance and condition. The flexible structure of Markov Chain is its main advantage in describing their evolution. On the other hand, estimation of the transition probabilities in Markov Chain requires special attention. For a long lifecycle, typically tens or hundreds of years, a small amount of error in these probabilities may cause significant deviation from reality. The application to bridge management systems is focused in this paper. First it reviews state of the art in estimating transition probability in relevant areas. A new approach to transition probability estimation for bridge management systems is then proposed and applied to bridge data of Michigan Department of Transportation in the US. Results show, compared with the Pontis methodology, that the proposed approach can predict more realistically the future probability distributions of bridge element condition.
Gong-Kang Fu Dinesh Devaraj
Center for Advanced Bridge Engineering, Wayne State university, Detroit, MI, US Changjiang Scholar C Center for Advanced Bridge Engineering, Wayne State university, Detroit, MI, US
国际会议
长沙
英文
95-100
2007-11-19（万方平台首次上网日期，不代表论文的发表时间）