Optimal renewable microgrid energy dispatch in the presence of uncertainty wind speed forecasting

The stochastic nature of wind speed and other renewable energy resources (RES) requires robust approaches to the energy management of renewable energy microgrids (RES MG). In this study, enhancements to the control hierarchy of RES MGs are proposed by taking advantage of probabilistic forecasting of wind speed and solar radiation with machine and deep learning models as a way to introduce uncertainty to the tertiary operations of the system. Using uncertainty through prediction intervals provides improvements over the conventional deterministic approaches for long-term planning, resources allocation and risk management of electrical grids with RES MG participation. The formulated probabilistic economic dispatch problem also takes into consideration grid stability, reduced carbon emissions, operation-and-maintenance costs, and load uncertainty as part of the optimal cost solution. Uncertainty-based optimization techniques such as stochastic and chance-constrained programming are used to solve the optimization problem under different criteria. The results provide further motivation for the adoption of probabilistic forecasting of renewable energy sources in the long-term operations of RES MG for a more reliable and cost-effective clean energy grid.