摘要
This paper investigates non-stationary discrete time Markov decision models with Borel state spaces and action sets, in which fuzzy criterion functions are employed to be a measure of performance. A class of Markov fuzzy criterion decision models is formulated. Under certain conditions, it is proved that the optimal equations hold and there exists a deterministic Markov optimal policy and the structure properties and the stability theorem of optimal policy are obtained. For models where transition laws are determined by system state equations, two conditions to be verified easily are presented and two applied examples are given.
This paper investigates non-stationary discrete time Markov decision models with Borel state spaces and action sets, in which fuzzy criterion functions are employed to be a measure of performance. A class of Markov fuzzy criterion decision models is formulated. Under certain conditions, it is proved that the optimal equations hold and there exists a deterministic Markov optimal policy and the structure properties and the stability theorem of optimal policy are obtained. For models where transition laws are determined by system state equations, two conditions to be verified easily are presented and two applied examples are given.
