Forecasting methods for wind power scenarios of multiple wind farms based on spatio-temporal dependency structure

In this paper, we first proposed a new wind power generation scheme for multiple wind farms that considers both spatial and temporal dependence of wind power time series based on copula theory in a computationally efficient way. Different vine copulas were studied and compared. Then a novel two-stage spatio-temporal sampling method is used to generate wind power scenario samples. Finally, we propose a conditional quantile sampling method that converts the samples into specific wind power generation scenarios. A case study was conducted on Wind Integration National Dataset (WIND) of National Renewable Energy Laboratory (NREL) to benchmark the performance of the new method with existing methods. The results show that the proposed method can generate scenarios with more accuracy and efficiency, which is conducive to subsequent operations.

Automated summary

This paper proposed a new wind power generation scheme for multiple wind farms that considers both spatial and temporal dependence of wind power time series in an efficient way based on copula theory. Different vine copulas were studied and compared, and a novel two-stage spatio-temporal sampling method was used to generate wind power scenario samples. The proposed conditional quantile sampling method can convert samples into specific wind power generation scenarios with more accuracy and efficiency.