Dynamic Equivalence Method of Wind Farm Considering the Wind Power Forecast Uncertainty

Abstract

The clustering of wind turbine generators (WTGs) in many dynamic multi-machine equivalence methods of wind farm (WF) is according to the deterministic data. However the forecasted wind power and speed are full of uncertainty. Considering this point, a WF dynamic multi-machine equivalence method is presented for the power system dispatch early warning calculation. The joint probability density functions (PDFs) of wind power and speed errors in forecast are selected as clustering index. The improved KL distance is applied to evaluate the similarities among the wind power forecast error distributions of all WTGs. All WTGs are divided into several groups by k-means algorithm. The WTGs and lines in each group are equivalent to a single-machine equivalent model. Considering both inaccurate and accurate scenarios of wind power forecast in the simulation case, the WF output responses curves of the detailed model, the conventional equivalent model and the proposed equivalent model are compared. The result indicates that the proposed equivalence method has better accuracy when the forecasted value is inaccurate.