Metaheuristic strategy for the hierarchical predictive control of large scale energy networks

Abstract

The short term optimization and control of energy networks is of great interest for Energy Industries because of the technical, economical and environmental benefits which could be gained from an appropriate management. However, models of such complicated systems are strongly non linear due to the energy propagation modeling and suffer from important uncertainties. Furthermore, some of the control variables are binary ones (on/off status of production units) and some of them are real ones (amount of energy to be produced). In this article, the goal is to encompass the whole technological string "production - distribution - consumption" by defining a suitable hierarchical predictive control strategy. In a first step, a global predictive control law is defined to compute the global amount of energy to be produced by each production site. In the second stage, this energy is dispatched between the production units of each site by a local predictive law. Due to the complexity of the system, an exact solution of the on line optimization problems to be solved in the predictive control strategy is untractable, and metaheuristic optimization methods are used. The global law is computed by a Particle Swarm Optimization (PSO) method whereas the local law is computed by ant colony and genetic algorithm. Numerical results exhibit more than satisfactory results and prove the viability of the approach.