A hybrid multi-objective optimizer-based model for daily electricity demand prediction considering COVID-19

Area of Science:

• Energy Systems
• Computational Intelligence
• Epidemiology

Background:

• COVID-19 lockdowns significantly impacted global electricity consumption patterns.
• Ensuring electricity supply security is vital for public welfare during epidemics.
• Existing electricity forecasting models often neglect pandemic-specific factors and stability metrics.

Purpose of the Study:

• To develop a robust electricity consumption prediction model adaptable to pandemic conditions.
• To improve both the accuracy and stability of electricity demand forecasts.
• To address the limitations of traditional forecasting methods during health crises.

Main Methods:

• A hybrid prediction system integrating data processing, modeling, and optimization.
• Utilizing an improved complete ensemble empirical mode decomposition with adaptive noise for data preprocessing.
• Employing a multi-objective optimizer and support vector machine for accurate and stable predictions.

Main Results:

• The proposed hybrid model demonstrated superior performance over benchmark models in prediction accuracy and stability.
• Incorporating daily infection data as an input parameter significantly enhanced model performance.
• The model proved effective in predicting daily electricity demand in the US during the pandemic.

Conclusions:

• The developed hybrid model offers a reliable solution for electricity demand forecasting in pandemic scenarios.
• Considering epidemiological data alongside consumption data is key to improving forecast accuracy and stability.
• The model shows significant potential for real-world application in energy management and security.