Enhanced framework embedded with data transformation and multi-objective feature selection algorithm for forecasting...

Area of Science:

• Renewable Energy Systems
• Artificial Intelligence in Energy
• Computational Intelligence

Background:

• Global interest in wind energy necessitates precise wind power forecasting for effective grid management.
• Existing forecasting methods often struggle with complex, high-dimensional wind energy datasets.
• Optimizing feature selection and data processing is key to enhancing predictive model performance.

Purpose of the Study:

• To propose an integrated framework for accurate wind power forecasting.
• To leverage advanced algorithms for optimal feature selection and data transformation.
• To develop a robust hybrid deep recurrent network-long short-term memory model for wind power prediction.

Main Methods:

• Utilized the multi-objective none-dominated sorting genetic algorithm III (NSGA-III) for optimal feature selection from wind energy data.
• Implemented a data transformation mechanism on selected features prior to model input.
• Developed and applied a hybrid deep recurrent network (DRN) and Long Short-Term Memory (LSTM) architecture for wind power modeling.

Main Results:

• The proposed framework demonstrated superior effectiveness and robustness compared to classical approaches.
• Achieved significantly lower Mean Squared Error (MSE) of 2.6593e-10 and Root Mean Squared Error (RMSE) of 1.630e-05.
• Outperformed existing methods, highlighting the benefits of the integrated feature selection and hybrid deep learning approach.

Conclusions:

• The integration of a data transformation mechanism with the NSGA-III and hybrid DRN-LSTM model offers substantial improvements in wind power forecasting accuracy.
• The proposed framework provides a robust and effective solution for managing wind power generation.
• This study offers valuable insights and a foundation for future research in advanced wind energy forecasting techniques.