Individualized Short-Term Electric Load Forecasting Using Data-Driven Meta-Heuristic Method Based on LSTM Network.

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Computational Science

Background:

• Short-term load forecasting is crucial for efficient power system operation and electricity management.
• Accurate demand prediction is essential for optimizing power plant unit scheduling and resource allocation.
• Current forecasting methods often struggle with parameter selection, impacting prediction accuracy.

Purpose of the Study:

• To develop an advanced approach for short-term electric load forecasting.
• To enhance the accuracy and stability of load predictions using novel optimization techniques.
• To address the limitations of manual parameter tuning in long short-term memory networks.

Main Methods:

• Utilizing long short-term memory (LSTM) networks, a type of recurrent neural network adept at handling sequential data.
• Employing an improved sine cosine algorithm, termed Meta-Optimization with Chaotic Recurrence (MetaREC), for parameter optimization.
• Integrating MetaREC with LSTM to overcome the inaccuracies associated with manual parameter selection in LSTM models.

Main Results:

• The MetaREC algorithm demonstrated superior convergence accuracy and speed compared to other optimization methods on test functions.
• The MetaREC-LSTM model outperformed baseline models, including standard LSTM, LSTM with conventional sine cosine algorithm, and LSTM with whale optimization.
• Simulations on a real electric load dataset confirmed the high accuracy and stability of the MetaREC-LSTM approach for short-term load forecasting.

Conclusions:

• The proposed MetaREC-LSTM approach significantly enhances the accuracy and stability of short-term electric load forecasting.
• MetaREC provides an effective solution for optimizing LSTM parameters, addressing a key limitation in previous methods.
• This research offers a promising tool for improving electricity management and power system operations through reliable demand prediction.