Wind power forecast based on variational mode decomposition and long short term memory attention network

Xiao Zhou, Chengxi Liu, Yongjian Luo, Baoying Wu, Nan Dong, Tianying Xiao, Haojun Zhu

Energy Reports · 23 August 2022

In one sentence. VMD decomposition with an attention-LSTM encoder-decoder improves wind power forecasts from NWP inputs over several benchmark models.

Abstract

Wind power forecast is becoming more and more important as the ever-increasing penetration of renewable energies brings uncertainties to the power systems. Although the numerical weather prediction (NWP) has already been widely applied on wind power forecast, it involves high computational burden with complex meteorological models, which has great uncertainty in real environment, so machine learning methods are the significant supplement for accurate wind power forecast. This paper proposes a deep learning model to improve the prediction accuracy based on the NWP data. Variational Mode Decomposition (VMD) is applied to extract time-series information. Furthermore, an encoder–decoder structure consisting of a dual attention-long short term memory (LSTM) neural network is constructed to enhance the forecasting accuracy. The comparison between the proposed model and several benchmark models, show the superiority of proposed model in effectively enhancing the prediction performance.

Labels

Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.