Wind Speed Prediction Method Based on Combined Deep Learning Model

The necessity of precise wind speed forecasts is crucial for integrating wind turbines into microgrids. This study introduces a hybrid forecast approach that initially optimizes the key factors of variational modal decomposition (VMD) through the whale optimization algorithm (WOA). Then decomposes t...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2024 9th International Conference on Intelligent Computing and Signal Processing (ICSP) s. 1251 - 1255
Hlavní autoři: Zeng, Chuxi, Xia, Zhipnig, Zhou, Yihuan, Li, Xi
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 19.04.2024
Témata:
combined deep learning model
component
Computational modeling
Correlation coefficient
Data models
Deep learning
gate recurrent unit
Prediction algorithms
Predictive models
Signal processing algorithms
temporal convolutional network
variational modal decomposition
whale optimization
Wind farms
Wind forecasting
Wind speed
Wind speed prediction
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:The necessity of precise wind speed forecasts is crucial for integrating wind turbines into microgrids. This study introduces a hybrid forecast approach that initially optimizes the key factors of variational modal decomposition (VMD) through the whale optimization algorithm (WOA). Then decomposes the initial wind speed data into intrinsic mode functions (IMFs) using the WOA-VMD method. Secondly, the Pearson correlation coefficient is conducted between IMFs and the initial wind speed data, classifying the IMFs into two groups according to their correlation coefficients. IMFs closely correlated are forecasted utilizing a temporal convolutional network (TCN), while those less correlated are forecasted via a gate recurrent unit (GRU). Finally, the forecasting outcomes from each category are then aggregated to formulate the ultimate forecast results. The hybrid WOA-VMD-GRU-TCN model is tested on actual data from six wind farms across China, the results demonstrate that the proposed model markedly enhances forecast precision, minimizes wind speed prediction errors, and exhibits superior robustness compared to alternative methods.
DOI:10.1109/ICSP62122.2024.10743895