Deep Learning Method Based on Gated Recurrent Unit and Variational Mode Decomposition for Short-Term Wind Power...

Area of Science:

• Renewable Energy Systems
• Artificial Intelligence in Energy
• Time Series Forecasting

Background:

• Wind power interval prediction (WPIP) is crucial for managing power system uncertainty.
• The intermittent nature of wind power presents challenges for high-quality prediction interval (PI) production.

Purpose of the Study:

• To propose a novel hybrid model for WPIP.
• To improve the quality of prediction intervals for wind power.
• To reduce the computational time for WPIP.

Main Methods:

• A hybrid model combining variational mode decomposition (VMD) and gated recurrent unit (GRU) neural networks was developed.
• VMD decomposes wind power data into simpler modes, with GRU models trained for each mode.
• An adaptive optimization method using constructed intervals (CIs) was employed for training label generation.

Main Results:

• The proposed hybrid model demonstrated superior performance compared to traditional interval prediction models.
• Higher quality prediction intervals were achieved.
• The method required less training time.

Conclusions:

• The novel hybrid VMD-GRU model offers an effective solution for WPIP.
• This approach enhances prediction accuracy and efficiency in wind power forecasting.