Wind speed short-term prediction using recurrent neural network GRU model and stationary wavelet transform GRU hybrid model

被引：31

作者：

Fantini, D. G. ^{[1
]}

Silva, R. N. ^{[2
]}

Siqueira, M. B. B. ^{[1
]}

Pinto, M. S. S. ^{[2
]}

Guimaraes, M. ^{[3
]}

Brasil Junior, A. C. P. ^{[1
]}

机构：

[1] Univ Brasilia, Dept Mech Engn, Energy & Environm Lab, BR-70910900 Brasilia, DF, Brazil

[2] Univ Estadual Maranhao, Dept Comp Engn, BR-65000000 Sao Luis, Maranhao, Brazil

[3] Dept Dam Safety & Technol DSBE, Dept Dam Safety & Technol DSB E, BR-74923650 Aparecida De Goiania, Go, Brazil

来源：

ENERGY CONVERSION AND MANAGEMENT | 2024年 / 308卷

关键词：

Recurrent neural networks; Wavelet transform; Forecast wind speed; Hybrid deep learning; Deep learning applied to renewables; Dispatchability; DECOMPOSITION; MULTISTEP; STRATEGY;

D O I：

10.1016/j.enconman.2024.118333

中图分类号：

O414.1 [热力学];

学科分类号：

摘要：

This study aims to evaluate the application of the wavelet transform (WT) as a pre-processing and hybridization technique for Recurrent Neural Networks (RNN). The modeling approach presented here aims to enhance hourly wind forecasting by improving its accuracy. For this strategy of study, a model based on the Gated Recurrent Unit (GRU) was employed. We propose a methodology for integrating wavelet transforms with RNNs, along with an analysis of the potential errors arising from incorrect partition and processing of training and validation data. Ultimately, our observations suggest that employing WT as a pre-processing step for GRU input data does not yield improvements that would justify its use.

引用

页数：13