Multi-scale Models for Decomposing Uncertainties in Load and Wind Power

Abstract

AbstractIn this chapter, we develop data-driven statistical representation of uncertain wind and load power in Azores island.We next derive auto-regression models for 10-min, 1-h and 24-h forecastmodels of wind and load power. Finally, we derive Markov models for the short-term and long-term characterization of the wind and load power, and obtain the short-and long-term decision trees for stochastic decision-making in operations and planning. These trees are more detailed than typical binomial trees, as they have the probabilities of several most likely states and their transitional probabilities. These models can be used for short-term model-predictive dispatch and unit commitment on a daily basis. The longer-term annual decision trees could be used for more dynamic and probabilistic decision-making regarding the best choice of technology to invest in under major uncertainties.591KeywordsWind PowerFinite Impulse ResponseLoad PowerInfinite Impulse ResponseUnit CommitmentThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.