Some aspects of stochastic dynamic programming with application to large scale multiobjective systems

This disertation presents a comprehensive an-alysis of an important class of dynamic programmingoptimization schemes of the stochastic, large-scaleand multiobjective type.A stochastic dynamic programming (SDP) modelis proposed for a stochastic periodically stationarysystem. For this SDP, the optimal decision that op-timizes the total expected return over an infinitetime horizon is considered. It is shown that the sy-stem is specified by both a deterministic state var-iable and a stochastic state variable with accompan-ied transitional probabilities.For the above SDP forulation with multio-bjectives, a recurive function is defined and suitablemanipulations are invoked using either the constraintmethod or the weighting method. In the constraintmethod, one of the objectives is incorporated in theobjective function, while the other objectives aresatisfied by imposing suitable constraints on the