Wind energy forecasting framework based on new perspective

摘要

As a green renewable energy, in recent years, wind energy has been vigorously developed and utilized by countries all over the world. However, the randomness and nonlinearity of wind itself hinder the process of massive combination of wind energy. Therefore, high-precision wind energy forecasting exerts a great significance on wind power grid. To enhance the forecast quality, this research proposes a new combination model. First, The original data is disintegrated into nonlinear component and linear component by variational mode decomposition (VMD), which discovering hidden information to clean noise and decrease interference with predictions, and then Long Short Term Memory neural networks (LSTM) and ARIMA model are established according to the statistical characteristics, and the ultimate forecasting results are obtained. The research case illustrates that the integrated model is superior to the estimated performance of conventional forecast models, which also implies that it is active for the VMD disintegration model to select the forecast method suitable for the component characteristics to increase the estimated capability of the model. The combination method based on data decomposition and component characteristics proposed in this article is conducive to the valid operation of wind power scheduling, reducing the impact on wind power accommodation space, and raising the stability behavior of power grid running.