Anticipatory response model for multi-agent based energy management system in a standalone microgrid

In this paper, multi-agent architecture was used to provide control for standalone microgrid with distributed generations. Therefore, to achieve a faster control compared to the centralized controller, each agent incorporated with a local prediction or forecasting model to provide anticipatory responses. To accomplish their common goals successfully, the agents cooperated based on facilitator architecture with game-theory. Initially, the agents estimate its own parameters and dynamically adjust them by playing non-cooperative game with other agents. The predictive algorithm is based on autoregressive model where each agent will predict the load demand alongside renewable energy resources in order to dynamically regulate the control parameters. This will provide a faster response where the agents will anticipate future load demand and available renewable resources and adjust their parameters beforehand. Hence, this will minimize the fluctuations of voltage and frequency in the microgrid leading to more efficient power dispatch and lower power losses.