A two-stage deep learning-based hybrid model for daily wind speed forecasting.

Area of Science:

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

Background:

• Increasing global wind energy adoption necessitates accurate wind speed (WS) prediction for turbine optimization.
• Current WS models often leverage artificial intelligence (AI) but require further reliability enhancements.
• Developing advanced AI-driven WS models is crucial for efficient wind energy generation.

Purpose of the Study:

• To develop and evaluate novel hybrid AI models for improved wind speed (WS) forecasting.
• To investigate the performance of combined Time Varying Filter-based Empirical Mode Decomposition (TVFEMD) with Gradient Boosting (GB) and Long Short-Term Memory (LSTM) models.
• To establish a superior hybrid WS prediction model for wind energy applications.

Main Methods:

• Hybridization of TVFEMD with standalone Gradient Boosting (GB) and Long Short-Term Memory (LSTM) models.
• Development of TVFEMD-GB and TVFEMD-LSTM hybrid models for comparative analysis.
• Creation of a simultaneous hybrid TVFEMD-GB-LSTM model, integrating all three components.

Main Results:

• Novel hybrid models demonstrated superior performance compared to standalone GB and LSTM models.
• The integrated TVFEMD-GB-LSTM model achieved the highest accuracy in WS forecasting.
• The study successfully validated the enhanced predictive capabilities of the proposed hybrid approach.

Conclusions:

• The developed hybrid AI models offer a significant advancement in wind speed (WS) prediction accuracy.
• The TVFEMD-GB-LSTM model presents a promising alternative for reliable WS forecasting in wind energy.
• This research opens new avenues for optimizing wind turbine performance through advanced AI techniques.