Point Estimate Method Addressing Correlated Wind Power for Probabilistic Optimal Power Flow

Increasing levels of wind power integration pose a challenge in system operation, owing to the uncertainty and non-dispatchability of wind generation. The probabilistic nature of wind speed inputs dictates that in an optimization of the system, all output variables will themselves be probabilistic. In order to determine the distributions resulting from system optimization, a probabilistic optimal power flow (POPF) method may be applied. While Monte Carlo (MC) techniques are a traditional approach, recent research into point estimate methods (PEMs) has displayed their capabilities to obtain output distributions while reducing computational burden. Unfortunately both spatial and temporal correlation amongst the input wind speed random variables complicates the application of PEM for solving the POPF. Further complications may arise due to the large number of random input variables present when performing a multi-period POPF. In this paper, a solution is proposed which addresses the correlation amongst input random variables, as well as an input variable truncation approach for addressing the large number of random input variables, such that a PEM can be effectively used to obtain POPF output distributions.