Reference Tracking Controller Applied to the Operation of Hydrothermal Systems

Abstract this paper considers the long term operation planning of hydrothermal systems modelled by a Markov jump system, using reference tracking controller. The reference is obtained via optimization of a certain cost functional in a deterministic setup, using deterministic dynamic programming. The control is designed employing a linear Markov jump model and quadratic costs. For selecting the quadratic cost weighting matrices, we propose a scheme based on Monte Carlo simulation of the system with the deterministic dynamic programming solution. We present some preliminary case studies suggesting that the reference tracking controller represent an interesting alternative to the deterministic or stochastic dynamic programming solutions.