Very short-term probabilistic forecasting of wind power based on OKDE

Accurate wind power (WP) forecasting plays an important role in the safe and stable operation of power systems, in the context of a high proportion of WP. In this paper, a two-stage WP very short-term probabilistic forecasting model is proposed. Firstly, the different frequency components are obtained based on complementary ensemble empirical mode decomposition (CEEMD) of historical data. Secondly, a long short-term memory (LSTM) based WP very short-term deterministic forecasting model was developed. An optimized kernel density estimate (OKDE) is established to fit the errors by means of band width (BW) constraints, and the quantile is calculated at different confidence levels. Finally, the deterministic forecasting results are combined with the quantile to calculate the probabilistic forecasting interval. Case Studies are also carried out to verify the effectiveness of the model proposed in this paper.