Holt-Winters-Based Time Series Modeling for Smart Grid Energy Forecasting Using Household Power Data

To achieve green and sustainable development, reducing energy losses through intelligent residential electricity management has become a critical challenge. Accurate load forecasting is the cornerstone of such systems, yet existing methods often struggle with multi-temporal-scale prediction and household-level adaptability. In the context of energy management, residential electricity consumption needs to enable flexible and intelligent decision-making. For sound intelligent decision-making, accurate forecasting of future electricity demand is essential, regardless of whether it pertains to an entire residential area or individual households. Consequently, electric energy load forecasting has attracted growing attention in recent years and presents certain challenges. This paper presents a household electricity consumption data prediction method grounded in the Holt-Winters model. The method is implemented using a benchmark dataset of electricity consumption data from an apartment. The model was trained and tested on datasets with 15-minute and 1-hour time-step resolutions respectively. Experimental results demonstrate that the model exhibits good performance on both datasets.