Gerenciamento eficientemente energético de microrrede de energia elétrica utilizando um sistema híbrido inteligente: uma abordagem multiagentes com percepção Fuzzy e Q-Learning.

Abstract

The evolution of microgrids can be noted through time, evolving their concept and aspects, as well as their control techniques. The applications of Artificial Intelligence in these control techniques have proven to be effective, allowing the microgrid to be managed satisfactorily. The combination of two or more intelligences is a new alternative, with several possibilities that allow complementing their benefits and deficiencies in order to manage a microgrid in a more comprehensive way. This combination is known as a hybrid intelligent system, and this work uses a multiagent system with reinforcement learning with Q-Learning and fuzzy perception to energetically manage a microgrid. The microgrid in question is composed of wind and photovoltaic systems, an energy storage system composed of batteries, and priority and non-priority loads. The combination of artificial intelligences aims to optimize the management of these resources, choosing which action to take regarding the charging and discharging of batteries, and connection and disconnection of loads. The performances of the implemented systems are compared, and it’s verified if the use of this hybrid intelligent system can offer an energy-efficient management of a microgrid.