Modelagem estocástica em usinas virtuais de energia utilizando transformada de incerteza

Abstract

The increase in distributed generation (DG) participation is a reality, influencing the way it generates, distributes and consumes electricity. Virtual Power Plant (VPP) will play an important role in integrating decentralized power generation systems with markets. However, the scheduling of distributed energy resources present in the VPP are important issues and require commitment forecasts from participating units. Predicting production or demand profiles is not a trivial task, as they rely heavily on weather characteristics, and the predictability of consumer demand is inherently variable. The development of methodologies and tools to meet these future uncertainties in different scenarios is relevant to the interests in the electricity markets. Probability forecasts are growing as a tool for managing variability. In this context, this thesis proposes a methodology for forecasting generation dispatches and individual user demands by modeling stochastic uncertainty using Unscented Transform (UT) in a VPP. These forecasts using UT were more satisfactory for more assertive decision making and better estimates for likely scenarios, minimizing decision risks for VPP aggregators, thus reducing their uncertainty about operations. The results validate the efficiency of the proposed technique using data and simulations. The UT, when inserted in this scenario, presents a good performance from a technical point of view in a VPP as evidenced in this thesis. Knowledge of the predictability and uncertainties that UT provides in a VPP will leverage the assertiveness of electrical scenarios, optimize economic results, leverage smart statistics, and improve electrical system planning.