Abstract
The optimum control of a stochastic process such as the traffic-flow system and the dam-system has not been so much studied and many questions have been left unsolved. For the operation of the dam systems, it is necessary to determine the optimum quantity (i.e. the largest quantity without lack) of service water obtainable from a river. However this is not easy because the input flows into a dam are stochastic values due to seasonal and random variations.
In this paper, one year is divided into four seasons and the optimum quantity of the service water in each season is determined by means of the statistical decision theory when the total input flow in the last season and the initial water level of the dam in this season are given. The same system also can be considered as a Markov process and the problem to determine the maximum quanity of the service water over a year under the condition that the probability of dearth water is less than 0.1 is solved by use of Dynamic Programming. The above two methods are compared and discussed.
