An aggregation and disaggregation procedure for the maintenance of a dynamic system under partial observations
5th European Workshop on Probabilistic Graphical Models, PGM 2010, Helsinki, Finlandiya, 13 - 15 Eylül 2010, ss.217-224, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Basıldığı Şehir: Helsinki
- Basıldığı Ülke: Finlandiya
- Sayfa Sayıları: ss.217-224
- Boğaziçi Üniversitesi Adresli: Evet
Özet
We study the maintenance of a dynamic system consisting of several components each of which age with a constant transition probability of failure. The state of the components are hidden. However, partial observations exist in each time period. The decision of whether to replace a component or to do nothing is to be made in each decision epoch. A hierarchical solution procedure is proposed to solve the problem. An aggregate model is developed by aggregating states and actions so that it can be solved with exact partially observable Markov decision process (POMDP) techniques. Then disaggregation is performed by simulating the process using a dynamic Bayesian network (DBN) and applying troubleshooting approaches in the decision epoch where replacement is planned in the aggregate policy.