Incorporation of forecasts to the optimal operation of dynamic systems using stochastic CEGH models

The optimal operation of a stochastic dynamic system has the classical solution given by the recursion of Bellman. The resolution proposed by Bellman to obtain the Optimal Policy of Operation suffers from the well-known "Bellman’s Curse of Dimensionality" (BCD). The strategies for combating BCD are varied. In the representation of the state of a system, one way to reduce the effect of the BCD is to reduce the dimension of the state by selecting which state variables to represent and which not. This causes a simplification in the representation of the state of the system. Since the Future Cost (FC) function is calculated over the reduced space, it results in a sub-optimal solution. In this paper, we analyze how to consider such loss of information from forecasts in the Dynamic Stochastic Programming (DSP).