An optimal inspection and replacement policy under incomplete state information (Q1091924)

The paper deals with a system which has a finite number of deterioration states and is monitored incompletely. It undergoes deterioration according to a discrete-time Markov chain and the costs of operation and replacement increase with deterioration. Assuming a probabilistic relation (this assumption may not hold in real situations, as mentioned by the authors) between the actual state of the system and the outcome of the monitoring, the authors formulate the problem as partially observable Markov decision process and discuss an optimal inspection and replacement policy to minimize the expected total discounted cost over an infinite horizon.