Abstract
The disaggregative estimation method of Markov transition probabilities has been proposed by the development of the Markov deterioration hazard model in 2005. This makes improvement of accuracy of deterioration forecasting in infrastructure asset management and the number of applications of statistical deterioration forecasting models in practice is increasing. In the Markov deterioration hazard model, it is assumed that transition of condition states occurs between only adjacent condition states. This assumption, however, may cause a problem that the Markov deterioration hazard model cannot estimate deterioration processes which includes transition of condition states between not adjacent condition states. In this study, to overcome this problem, the multi-destination Markov deterioration hazard model is proposed with use of multi-destination hazard models. The proposed methodology is demonstrated in a numerical example and empirical study using inspection data of existing bridges, and its effectiveness is discussed.
