Forecasting Congestions in Interconnected Power Systems with High Shares
of Renewable Energy: A Probabilistic Approach using Copulas
Abstract
This paper proposes a novel approach to forecast congestions in
high-voltage grids with high shares of distributed photovoltaic (PV)
infeed. The approach is based on a physical PV model using intra-day
numerical weather prediction (NWP) input data. Subsequently,
probabilistic forecasts are generated based on Kernel density estimators
(KDE) and Copula, describing the multivariate spatial dependencies for
the marginal distributions of forecasting and approximation errors.
Finally, a probabilistic power flow (PPF) using a linearized AC version
is proposed, combining the benefits of high accuracy with high
computational performance. To assess and quantify the overall advantages
of this approach, a case study is carried out for an existing power
system.