Short-Term Wind Generation Combined Forecast Considering Meteorological Similarity

High-precision short-term wind generation prediction results are conducive to making a scientific generation plan and improving the wind power absorption capacity of the power grids. Based on the analysis of the relationship between the numerical weather prediction and wind power, this paper proposes a short-term wind generation combined forecast model considering meteorological similarity to improve the prediction accuracy of short-term wind power. In this method, the meteorological similarity day model, the extreme gradient boosting algorithm and the back propagation neural network algorithm are selected for achieving the short-term wind power prediction. Then, the particle swarm optimization algorithm is applied to determine the weight of each single forecasting model. Finally, the prediction results are obtained through the combination of the single model prediction results. With the realistic wind power data collected from a wind farm in Xinjiang province, the short-term wind forecasting task is achieved by the proposed method. The simulation results illustrate that the combined model proposed in this paper can effectively improve the forecasting performance of the benchmark models.