Wind Speed Prediction for Multiple Wind Farms Based on Improved Vine-Copula Method

The traditional Vine-Copula method employs Vine structure to represent the dependent structure between wind speed of wind turbines in wind farms, which can fully capture the spatial correlation among them, but lacks the consideration of temporal correlation. Therefore, a wind speed forecast method utilizing an improved Vine-Copula model for multiple wind farms is proposed. In this method, the Copula joint distribution function samples points of the wind speed for the Vine turbine using a Markov process to reflect its change on the time scale. Then, the Vine-Copula model is constructed between wind turbines to solve the spatial correlation problem and achieve wind speed prediction. Using the actual measured data from wind turbines at a wind power plant in Gansu, China, the effectiveness of the proposed method is demonstrated.