We suggest a heuristic solution procedure for Partially Observable Markov Decision Processes with finite action space and finite state space with infinite horizon. The algorithm is a fast, very simple general heuristic; it is applicable for multiple states (not necessarily ordered) multiple actions and various distribution functions. The quality of the algorithm is checked in this paper against existing analytical and empirical results for two specific models of machine replacement. One model refers to the case of two-action and two-system states with uniform observations (Grosfeld-Nir ), and the other model refers to a case of many ordered states with binomial observations (Sinuany-Stern et al. ). The paper also presents the model realization for various probability distribution functions applied to maintenance and quality control.
- Dynamic programming: Bayesian programming
- Partially observed Markov decision process: algorithms
- Reliability/maintenance: machine replacement
- Suboptimal design