A Study on the Suitability of Load Demand Forecasting Models for Island Area Using Weather Variables

Long and short-term load demand forecasting are of great importance in the process of electricity policy formulation. In particular, load forecasting in an energy-independence island is essential for an appropriate investment in off-grid electrical power systems using renewable energy sources. The purpose of this study was to evaluate the performance of three models in forecasting the electricity load demand in an island area: the multiple regression model, ARIMA model, and Reg-ARIMA model, which is the combined 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 study concluded that the combined method is a more appropriate model.