We propose a "threshold probability" decision-making in fuzzy environment as a multi-stage stochastic control process. The objective value is the threshold probability that total membership function is greater than or equal to a given lower grade. Under the controlled Markov chain we optimize the threshold probability not in large general policy class but in smell Markov policy class. This choice turns out to lead a valid recursive formula. We show that there exists an optimal policy in Markov class.
展开▼
