Short-term power load forecasting based on gray relational analysis and support vector machine optimized by...

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Data Science

Background:

• Short-term power load forecasting is critical for power system stability and automation.
• Traditional methods struggle with load volatility and the influence of external factors like weather.
• Existing models exhibit limitations in nonlinear mapping and generalization to new data.

Purpose of the Study:

• To develop a more accurate and robust short-term power load forecasting method.
• To address the limitations of traditional forecasting techniques.
• To improve the reliability of power system operations through enhanced load prediction.

Main Methods:

• Factor selection using Pearson correlation coefficient.
• Identification of similar historical days via Gray Relational Analysis (GRA) and K-means clustering.
• Optimization of Support Vector Machine (SVM) parameters (penalty coefficient, kernel function) using the Artificial Bee Colony (ABC) algorithm.

Main Results:

• The proposed GRA and ABC-SVM method demonstrated effectiveness in short-term power load forecasting.
• Simulation results using actual load data from Nanjing validated the method's performance.
• The approach successfully integrated factor analysis, similar day selection, and optimized machine learning.

Conclusions:

• The GRA and ABC-SVM method offers a significant improvement over traditional forecasting approaches.
• This enhanced forecasting technique contributes to safer and more efficient power system operation.
• The study highlights the potential of hybrid intelligent methods in addressing complex forecasting challenges.