Research on the prediction of short-term wind power based on wavelet neural networks

Abstract

The theory of applying wavelet neural networks to predict the time series of short-term wind power has important practical significance for the construction of emergency response systems in smart grids. Based on obtaining the low-frequency approximate part and high-frequency random part of wind power data through wavelet decomposition and reconstruction, and analyzing the advantages and disadvantages of various models, a short-term wind power prediction model is established. The simulation of the model realizes the complex mapping from the relevant historical wind power to the prediction target. The prediction results obtained from the measured wind power data of a wind farm show that the model has a fast prediction speed and high accuracy. It can effectively improve the prediction accuracy of short-term wind power.