Feature selection in wind speed forecasting systems based on meta-heuristic optimization.

Area of Science:

• Renewable Energy Systems
• Computational Intelligence
• Time Series Analysis

Background:

• Accurate wind speed forecasting is essential for integrating wind power into electrical grids.
• Existing methods struggle with wind's inherent unpredictability, impacting grid stability.
• Advanced computational techniques are needed to enhance forecasting accuracy.

Purpose of the Study:

• To develop a highly accurate wind speed forecasting method.
• To optimize a weighted ensemble model using a novel metaheuristic algorithm.
• To improve the balance between exploration and exploitation in optimization algorithms.

Main Methods:

• Proposed a novel weighted ensemble model for wind speed prediction.
• Developed an adaptive dynamic grey wolf-dipper throated optimization (ADGWDTO) algorithm to optimize model hyperparameters.
• Redesigned the Grey Wolf Optimizer (GWO) with dipper-throated bird hunting strategies for enhanced exploration.
• Optimized Multi-Layer Perceptron (MLP), K-Nearest Regressor (KNR), and Long Short-Term Memory (LSTM) models.

Main Results:

• The proposed ADGWDTO algorithm achieved high fitness (0.9209 avg, 0.7432 std dev) for feature selection.
• The optimized ensemble model demonstrated a low root mean square error (RMSE) of 0.0035.
• The method significantly outperformed existing state-of-the-art wind speed forecasting algorithms.
• Statistical analyses (ANOVA, Wilcoxon's rank-sum) confirmed the algorithm's robustness and stability.

Conclusions:

• The novel ADGWDTO algorithm effectively optimizes ensemble models for superior wind speed forecasting.
• The enhanced optimization approach improves prediction accuracy and contributes to power grid stability.
• This method offers a robust and reliable solution for wind energy integration challenges.