Short-term Wind Power Interval Prediction Method Based on Empirical Mode Decomposition (EMD) and Non-parametric Estimation Method

This paper proposes a combined short-term wind power interval prediction method based on empirical mode decomposition (EMD), BP neural network, and non-parametric kernel density estimation method, effectively improving the prediction accuracy. This method first uses EMD to decompose the historical wind power and reconstruct the sequence using the run-length test. BP neural network models are then established for each decomposed and reconstructed subsequence and meteorological factors. Finally, the results of the subsequence prediction models are superimposed to obtain the wind power prediction value, achieving a 5.57% improvement in combined prediction accuracy. Combining the characteristics of power prediction and its prediction error distribution, the non-parametric kernel density estimation method is used to solve the probability density function of the prediction error for each power interval. The wind power prediction confidence interval is then obtained based on the deterministic power prediction value.