Stochastic model predictive control for robust operation of distribution systems

Resumen

There are many systems in which uncertainties are present in their model, either in the same system description or as disturbances. Among many random variables we can mention the electrical demand of a generation network, the amount of rainfall in an irrigation system, the number of people occupying a room in a system of heating. Among others, they are examples of stochastic systems, in which the idea of scenarios can be considered for their solution. Specifically, the stochastic model predictive control seeks to generate a solution for several scenarios that can be established under a probabilistic condition. In this work is carried out an analysis and comparison regarding performance among the three well-known stochastic model predictive control (MPC) approaches, namely, multi-scenario (MS), tree-based (TB), and chance-constrained (CC) MPC. The possibility of application in several distribution sectors is also analyzed. Moreover, some improvements are proposed in terms of robustness. To this end, the stochastic MPC controllers are designed and implemented in a real renewable-hydrogen-based microgrid as well as to the drinking water network of Barcelona via simulation. Finally, an application of CC-MPC to inventory management in a hospital pharmacy, is also presented. Stochastic MPC controllers are applied in a hierarchical and distributed fashion. In this sense, a scenario-based hierarchical and distributed MPC is applied for water resources management by considering dynamical uncertainty. In addition, a multicriteria optimal operation of a microgrid considering risk analysis and MPC is shown. For all applications, their design has considered the important role that uncertainty plays in these systems. Finally, in order to analyze different types of the so-called insider attacks in a distributed MPC scheme is presented. In particular, the situation where one of the local controllers sends false information to others is considered to manipulate costs for its own advantage. Then, some mechanisms based on stochastic MPC techniques are proposed to protect, or, at least, relieve the consequences of the attack in a typical DMPC negotiation procedure is addressed in this work.