Dynamic dependence modelling of wind power uncertainty considering heteroscedastic effect

Abstract The inherent stochastic nature of wind power poses significant challenges to the safe and economic operation of power systems, and an accurate model of wind power forecast error can give operators great help in dealing with the negative impact of wind power uncertainty on the power system. To obtain accurate forecast error intervals, a wind power forecast error (WPFE) model based on dynamic Copula theory is proposed in this paper. By combining Autoregressive integrated moving average (ARIMA) method and generalized autoregressive conditional heteroscedasticity (GARCH) model, the marginal distribution of WPFE model with the ability to describe its time-varying feature is achieved. In this proposed model, 4 dynamic Copula functions are estimated and evaluated by the combination method of Inference Functions for Margins (IFM) and Akaike Information Criterion (AIC), and the best-fitted one is selected to be applied in a set of synchronous data of wind power and its forecast. The result shows that compared with the conventional method, the proposed method can provide more accurate forecast error intervals. This advantage can benefit the scheduling of the power system with high wind power penetration rate, which is also verified by a stochastic unit commitment case in modified IEEE 118-Bus system.