Short-term wind speed prediction based on improved Hilbert-Huang transform method coupled with NAR dynamic neural...

Area of Science:

• Renewable Energy Systems
• Data Science and Predictive Modeling
• Power Systems Engineering

Background:

• Wind energy is a rapidly growing renewable source, vital for clean electricity generation.
• Wind speed intermittency, randomness, and uncontrollability pose significant challenges to power system stability and operational costs.
• Accurate wind speed forecasting is essential for safe equipment operation, grid integration, and stable power system functioning.

Purpose of the Study:

• To enhance the accuracy and stability of wind speed prediction models for wind energy generation.
• To reduce the economic costs associated with wind power generation and promote sustainable energy development.
• To address the limitations of traditional Empirical Mode Decomposition (EMD) methods in processing complex wind speed data.

Main Methods:

• Utilized an improved Hilbert-Huang Transform (HHT) with Complementary Ensemble Empirical Mode Decomposition (CEEMD) to process wind speed data, mitigating issues like component mode mixing and noise.
• Employed mathematical analytical methods to compute component weights for constructing a dynamic neural network model.
• Optimized the HHT-CEEMD approach with a Neural Adaptive Regression (NAR) model for time series forecasting.

Main Results:

• The developed HHT-CEEMD-NAR model demonstrated significant improvements in wind speed forecasting accuracy.
• The model effectively handled the stochastic and nonlinear characteristics inherent in wind speed data.
• Application in the Xinjiang region showed reduced root mean square errors and an enhanced coefficient of determination compared to traditional methods.

Conclusions:

• The improved HHT-CEEMD-NAR model offers a robust and accurate predictive tool for wind speed forecasting.
• This approach enhances the efficiency and reliability of wind power integration into the grid.
• The study provides a valuable methodology for optimizing wind power generation scheduling and ensuring stable power system operation.