Data-Driven Optimal Generation Scheduling Applying Uncertainty in Microgrid

Microgrids provide efficient means of incorporating renewable energy resources (RES) into the power network. The deployment of an energy management system into a microgrid is essential for achieving efficient utilization of resources and ensuring stable grid operation at a favorable cost. However, the inherent intermittent nature of consumer loads and RES introduces uncertainty, posing significant challenges for system design. This paper proposes a generation scheduling approach that can optimally manage energy resources in a microgrid in the presence of load and generation uncertainties. First, a data-driven machine learning algorithm is employed to forecast PV and wind generation as well as electrical power demand from weather data and actual dataset. Next, optimal unit commitment based on energy prices to minimize system costs is conducted using Mixed Integer Linear Programming (MILP). This approach provides optimal generation scheduling among PV and wind turbine generation systems as well as the required power from the utility grid. Simulation results for different case studies is carried out in order to demonstrate the performance of the proposed method for hourly RES and load profile forecast. Furthermore, results indicate that optimal generation scheduling can be effective in minimizing the operating cost under the worst-case of RES and load uncertainty.