A Year-ahead Prediction Method of Monthly Wind Power Generation

The monthly distribution quantity of annual electricity of the wind farm is the basic data for the preparation of annual power generation plan and maintenance plan. With the increasing proportion of wind power, the reduction of its prediction accuracy directly affects the plan's enforceability. The lack of historical data is the primary problem facing the medium and long-term electricity forecasting of wind farms. In view of this, the existing historical data are expanded by means of wind historical data and data rolling; the diversity and multi-scale characteristics of wind power influencing factors make it difficult for a single method to obtain satisfactory prediction accuracy. In view of this, a combined prediction method based on improved empirical mode decomposition for monthly distribution of annual electricity of wind farms is proposed. Aiming at the fluctuation characteristics and laws of each component after the decomposition of historical data, appropriate prediction methods are used to improve the prediction accuracy, and then the overall accurate prediction is achieved through data restoration. The example results show that the above measures can effectively improve the prediction accuracy.