본문 바로가기
  • Home

A Study on Electricity Demand Forecasting Using the ARIMAX Model

  • Journal of Internet of Things and Convergence
  • Abbr : JKIOTS
  • 2025, 11(5), 8
  • Publisher : The Korea Internet of Things Society
  • Research Area : Engineering > Computer Science > Internet Information Processing
  • Received : September 9, 2025
  • Accepted : October 8, 2025
  • Published : October 31, 2025

Hun Shim 1

1한국폴리텍 IV 대학 청주캠퍼스

Accredited

ABSTRACT

Based on Korea Power Exchange’s electricity consumption data by contract type, this study quantitatively forecasted Korea’s electricity demand. To comprehensively incorporate key explanatory variables such as real GDP growth rate, industrial production index, real electricity price index, cooling/heating degree days, and the expansion of data centers and electric vehicles, an ARIMAX (Autoregressive Integrated Moving Average with Exogenous Variables) model was developed. The model’s optimal order was determined using the AIC and BIC criteria and the Ljung–Box test, achieving high predictive accuracy with a MAPE of less than 2% and RMSE within 8TWh. Under the baseline scenario, total electricity demand in 2034 is projected to reach approximately 625TWh, with an uncertainty range of ±25TWh depending on the pace of data center and electric vehicle adoption. Notably, the industrial and commercial/service sectors are expected to account for about 75% of the total increase, highlighting demand management and efficiency improvement as key policy priorities. These findings can inform mid to long term decision-making in areas such as the Basic Electricity Supply and Demand Plan, industrial and EV charging infrastructure development, and power system flexibility enhancement.

Citation status

* References for papers published after 2024 are currently being built.