Fuzzy time series for short-term residential load forecasting in smart grids

Area of Science:

• Energy Systems Engineering
• Artificial Intelligence
• Time Series Analysis

Background:

• Accurate energy consumption forecasting is crucial for smart grid operations.
• Short-term load forecasting (STLF) impacts demand-side management and renewable energy integration.
• Traditional forecasting models face challenges in precise prediction due to variable influencing factors.

Purpose of the Study:

• To develop an improved fuzzy time series (FTS) methodology for residential electricity consumption forecasting.
• To enhance prediction accuracy at hourly, daily, and weekly intervals.
• To address limitations of traditional FTS models, such as overfitting and fuzzification processes.

Main Methods:

• Utilized a fuzzy time series (FTS)-based approach for load forecasting.
• Implemented techniques to mitigate overfitting during data partitioning.
• Refined the fuzzification process within the FTS model.

Main Results:

• Achieved up to 40% improvement in hourly load forecasting accuracy.
• Demonstrated up to 58% and 84% improvements in daily and weekly forecasting, respectively.
• Validated the methodology using real residential electricity consumption data.

Conclusions:

• The proposed FTS model significantly enhances residential electricity demand forecasting accuracy.
• The improved forecasting supports reduced peak-demand uncertainty in smart grids.
• This methodology contributes to more resilient, flexible, and sustainable smart grid systems.