Quarterly electricity consumption prediction based on time series decomposition method and gray model.

Area of Science:

• Energy Systems Analysis
• Time Series Forecasting
• Computational Intelligence

Background:

• Accurate electricity consumption prediction is vital for power system stability and waste reduction.
• Non-linear and seasonal fluctuations pose challenges to traditional forecasting models.
• Existing methods struggle to effectively capture seasonal patterns in energy demand.

Purpose of the Study:

• To develop a novel seasonal prediction method for electricity consumption.
• To address the challenges of non-linear and seasonal fluctuations in energy data.
• To improve the accuracy of electricity demand forecasting.

Main Methods:

• Utilized Seasonal and Trend decomposition using Loess (STL) for time series decomposition.
• Decomposed electricity data into trend, seasonal, and remainder components.
• Applied a gray model for predicting individual components with distinct characteristics.

Main Results:

• The proposed model demonstrated superior prediction accuracy compared to state-of-the-art methods.
• Achieved a Mean Absolute Percentage Error (MAPE) of 1.77% and Root Mean Square Percentage Error (RMSPE) of 2.37%.
• Successfully identified and predicted seasonal fluctuations in quarterly electricity consumption data.

Conclusions:

• The novel STL and gray model approach effectively predicts seasonal electricity consumption.
• This method offers a new tool for optimizing seasonal energy supply schemes.
• The model's accuracy enhances power system management and reduces energy waste.