Real‐Time Wind Power Forecast Error Estimation Based on Eigenvalue Extraction by Dictionary Learning

Abstract

Because of the fluctuation and uncertainty characteristics of wind power, it is difficult to achieve a perfect wind power forecast. The forecast error may lead to an imbalance between the load demand and power supply. The object of recent research on forecast error is to achieve the probability distribution of forecast error based on the statistics of historical data. This statistical error achieved from a probability distribution cannot reveal the real-time condition of wind power. A real-time forecast Error estimate method based on dictionary learning (EEDL) was proposed. In EEDL, several coefficients that have strong relevance to the forecast error are computed. The dictionary learning method is used to extract the eigenvalues of forecast error from these coefficients. Based on the eigenvalues, a real-time error estimation model was built to obtain the forecast error. EEDL was compared to the estimation method based on a Probability distribution function (PDF). The performance of EEDL was also compared to the error estimation method based on a PDF while using different forecast techniques.