<aside> 💡 UPEC 2010
</aside>
The huge amounts of data required by Smart grids operation are impossible to be processed by human operators in a timely manner. New intelligent systems should provide a clear decision regarding the system state. This paper proposes a new methodology based on supervised learning using AdaBoost and CARTs as decision support system for power system state classification. The methodology proves to be time efficient and precise, with low false negative rates. This approach could help in Smart Grids design and deployment, as it could be easily integrated into the existing EMS/SCADA systems.
Even though today’s electrical networks do not fully comply with the requirements of Smart Grids, it is expected that, in a foreseeable future, most of them will become “smarter”. Smart Grids are defined as the electrical transmission and distribution networks which, by incorporating information technology and communication capabilities, become able to predict, self-heal and self-adjust to changes.
All in all, we believe that machine learning could provide the support needed to develop the functionalities and performances required by a power grid to become a Smart Grid.