기상변수를 이용한 도서지역 전력수요 예측 모형의 적합성에 관한 연구

Abstract

Long and short-term load demand forecasting are of great importance in the process of electricity policy formulation. Inparticular, load forecasting in an energy-independence island is essential for an appropriate investment in off-grid electrical powersystems using renewable energy sources. The purpose of this study was to evaluate the performance of three models in forecasting theelectricity load demand in an island area: the multiple regression model, ARIMA model, and Reg-ARIMA model, which is thecombined model of the two preceding ones. Using the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error(MAPE) as a forecasting accuracy criterion and comparing the predicted and real values of the three islands in 2015, the studyconcluded that the combined method is a more appropriate model.