A new Frontier in electric load forecasting: The LSV/MOPA model optimized by modified orca predation algorithm.

Area of Science:

• Energy Systems
• Artificial Intelligence
• Data Science

Background:

• Accurate electric load forecasting is crucial for efficient energy management and policy.
• The complexity and dynamic nature of electric load data present significant forecasting challenges.

Purpose of the Study:

• To propose a novel hybrid technique, LSV/MOPA, for enhanced electric load forecasting.
• To evaluate the performance of LSV/MOPA against state-of-the-art methods.

Main Methods:

• Developed a hybrid model combining Long Short-Term Memory (LSTM) and Support Vector Regression (SVR).
• Optimized the hybrid model using a Modified Orca Predation Algorithm (MOPA).
• Applied and validated the LSV/MOPA model on 20 years of historical electric load data from four regions in South Korea.

Main Results:

• The LSV/MOPA model demonstrated superior performance across all regions, achieving the lowest Mean Absolute Percentage Deviation (MAPD) error.
• Achieved specific MAPD errors: 3.65 (North), 12.8 (South), 8.6 (Central), 30.8 (East).
• Exhibited faster convergence and better generalization capabilities compared to benchmark models like SVR/FFA, LSTM/BO, LSTM-SVR, and CNN-LSTM.

Conclusions:

• The LSV/MOPA model offers a highly effective and superior approach for electric load forecasting.
• The study highlights the potential of LSV/MOPA for applications requiring accurate predictive modeling in various sectors.