Analisis Peramalan Beban Listrik Jangka Pendek Libur Akhir Tahun Menggunakan Metode Exponential Triple Smoothing dan Koefisien Energi

Abstract

A short-term forecast of the electricity load for the year-end holiday season is important. The load demand pattern during this period fluctuates due to changes in people's annual activity patterns. Therefore, high-precision short-term load forecasting is required to detect the impact of changes in regional activity patterns on electricity consumption in order to improve the availability, stability and economic levels of the Java-Madura-Bali system during the year-end holidays. The Department of Energy and Natural Resources by Grid Code SDC (Scheduling and Dispatch Code) 6.2 sets a minimum standard of less than 5% error in implemented forecast results in metric MAPE units. The purpose of this study is to compare the Energy Coefficient method, the Exponential Triple Smoothing method, and the Linear Regression method with his MAPE metric and find a better forecasting method that satisfies the criterion. The results of the forecasting method with the best MAPE value will be used for the short-term electricity load forecast for the holiday season at the end of 2023. Based on the forecast results analyzed using standard MAPE metrics, the exponential triple smoothing method showed the highest accuracy for short-term load forecasts during the year-end and New Year holidays. The MAPE value obtained by this method is 2.52% based on the daily peak load increase scenario in January, 1.67% in December, and the MAPE value determined based on the daily energy increase scenario in January is 2.94% and 1.67% in December. 2.98% in December.