2015 Volume 71 Issue 2 Pages 70-89
In recent years, statistical deterioration prediction methods for infrastructures have been advanced rapidly. These methods mostly use inspection data evaluating the progression of deterioration with discrete soundness level. However, the criteria for judging soundness level regarding deterioration (hereinafter called judging criteria) may be changed during the service period of structures. In this case, conventional statistical deterioration prediction methods cannot be applied. In this study, the authors formulate a model for the relation between the soundness levels before and after the revision to the judging criteria, and propose a hidden Markov deterioration model as the model combined with a Markov deterioration model defined by the new standard soundness level after the revision to the judging criteria. In addition, the authors propose a method for conducting the Bayesian estimation of the above model with the MCMC method, and discuss the applicability and effectiveness based on the cases of applying it to tunnel luminaires.