Short-term load forecasting system based on sliding fuzzy granulation and equilibrium optimizer.

Area of Science:

• Electrical Engineering
• Data Science
• Artificial Intelligence

Background:

• Short-term electricity load forecasting is crucial for power system operations and planning.
• Existing models struggle with noise and nonstationarity in load data, leading to uncertainty.
• Accurate forecasting is essential for efficient energy management and grid stability.

Purpose of the Study:

• To develop a robust short-term load forecasting system addressing noise and nonstationarity.
• To enhance forecasting accuracy and stability using a hybrid approach.
• To provide both point and interval forecasts for comprehensive load prediction.

Main Methods:

• A modified information processing technique using sliding fuzzy granulation to denoise load data and capture uncertainty.
• Deep neural networks (DNNs) to model complex nonlinear patterns in electricity load.
• An advanced meta-heuristics algorithm to optimize DNN weighting coefficients for improved stability.

Main Results:

• The proposed system demonstrates superior effectiveness and stability compared to existing methods.
• Experimental results validate the system's ability to handle noisy and nonstationary load data.
• Comprehensive evaluation using multiple metrics confirms the forecasting performance gains.

Conclusions:

• The hybrid forecasting system effectively mitigates noise and nonstationarity in electricity load data.
• The integration of sliding fuzzy granulation, DNNs, and meta-heuristics significantly improves forecasting accuracy and reliability.
• This approach offers a more stable and comprehensive solution for short-term electricity load forecasting in power management.