A Study on Day-Ahead Load Forecasting for Special Days in Jeju Island

Jeju Island is faced with considerable challenges in day-ahead load forecasting due to its high share of renewable energy, unique industrial structure, and strong seasonal tourism influence. Particularly on special days such as national holidays and vacation periods, significant fluctuations in load patterns are observed, which complicate accurate forecasting. While existing forecasting models have been developed primarily for mainland power grids, their direct application to Jeju Island has been shown to yield lower accuracy, due to the island’s distinctive demand characteristics. To overcome these limitations, a statistical forecasting model based on similar-day analysis has been developed. In this model, analogous weekdays are selected using key meteorological and calendar variables, and Jeju’s high weather sensitivity along with the impact of behind-the-meter solar generation is incorporated. A case study using real grid data from Jeju Island was conducted. As a result, the proposed model achieved a mean absolute percentage error (MAPE) of 3.24% for hourly demand on the day ahead. This approach is expected to enhance the reliability of power market operations and supply planning, particularly for regions like Jeju Island where conventional forecasting methods are less effective.