Impact of Cooling and Heating Degree Hours on Short-Term Electricity Demand Forecasting

초록

In this study, we investigates the effectiveness of incorporating Cooling Degree Hours (CDH) and Heating Degree Hours (HDH) as derived features for short-term electricity demand forecasting. CDH and HDH quantify cooling and heating loads by accumulating hourly temperature deviations from a reference temperature, offering a representation of thermal demand. Using electricity consumption and weather data, three machine learning models—Random Forest, Support Vector Regression, and Artificial Neural Network— were evaluated. Experimental results show that adding CDH and HDH significantly improves prediction accuracy. Furthermore, the HDH calculated for working hours consistently outperformed the 24-hour calculations, indicating that winter heating demand primarily occurs in office environments, where electric heating is dominant. These findings highlight the value of CDH and HDH as predictive features and emphasize the importance of incorporating time-specific indicators to enhance the forecasting accuracy.

키워드