Selecting the forecast interval of wind generations

Abstract

This letter proposes a data‐driven and optimization‐based approach for determining the interval of wind generation forecast without knowing the exact probability distribution of the forecast error. This method requires rough information and is easy to implement. Comparative tests show its potential of practical application. © 2015 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

Estimating DLMP confidence intervals in distribution networks with AC power flow model and uncertain renewable generation

“…Without considering unimodality, it is a well-known result of Chebyshev, who proved that 1/γ 2 is a tight upper bound of (16), regardless of the true distribution of w. Chebyshev's bound is found to be conservative in probabilistic interval forecast of renewable power [63], because the worst-case distribution is discrete, where unfavourable samples which are distant from the forecast get higher chances to occur. When the distribution is restricted to be unimodal, C. F. Gauss found an analytical solution of (16) [62] Pr [ w − w ≥ γσ] ≤ 4/(9γ 2 ), if γ > 2/ 3; 1 − γ/ 3, otherwise;…”

“…The classical Chebyshev's bound can be reduced by more than half in Gaussian inequality (17) by and large. The case with a single renewable plant has also been discussed in [63] using the information of higher-order moments rather than unimodality. It is shown that incorporating fourth-and sixth-order moments can reduce the conservatism of Chebyshev's bound as well.…”