Energy Consumption Forecasting Using Semantic-Based Genetic Programming with Local Search Optimizer

Area of Science:

• Computer Science
• Artificial Intelligence
• Operations Research

Background:

• Energy consumption forecasting (ECF) is vital for economic policy and utility operations.
• Inaccurate ECF leads to increased operating costs for electric utilities.
• Existing methods require improvement for enhanced prediction accuracy.

Purpose of the Study:

• To propose a novel semantic-based genetic programming framework for accurate energy consumption forecasting.
• To develop a system capable of generating near-optimal predictions on unseen data.
• To enhance the efficiency and accuracy of ECF models.

Main Methods:

• Integration of a semantic genetic programming approach with a local search method.
• Utilizing semantic genetic operators for enhanced exploration capabilities.
• Employing a local searcher for exploitation and refinement of solutions.

Main Results:

• The proposed framework achieves high probability of finding near-perfect solutions.
• Experimental results demonstrate superior performance compared to state-of-the-art techniques.
• The system produces lower prediction errors on the same dataset.
• Forecasting accuracy on unseen data is significantly improved.

Conclusions:

• The semantic-based genetic programming framework is highly suitable for energy consumption forecasting.
• Combining semantic genetic programming with local search accelerates the search process.
• The enhanced models provide accurate forecasts, even for previously unseen data.
• This approach offers a significant advancement in energy consumption prediction.