Short-Term Electric Load Forecasting Study Using Linier Regression and Time Series Models At PT. PLN (Persero) Tarakan

Abstract

In the electricity operating system, it is very necessary to balance the electrical power transferred by the power plant with the electricity load consumption at the customer. Fluctuations in the electrical load cause the operation of the electrical power system to become unreliable. Electrical load forecasting studies are very necessary to ensure optimal electrical power system conditions. PT. PLN (Persero) Tarakan as an electrical energy service provider really needs electrical load forecasting studies. This study applies the time series analysis method and as a comparison uses the linear regression method. This method is used to forecast short-term needs in the electrical power operating system. The results of the study showed that the mean absolute error percentage (MAPE) with the linear regression method was 7.87 percent. With the ARIMA time series method, the MAPE was 14.68 percent. However, by looking at the seasonal plot in the time series or SARIMA method, the MAPE was 6.13 percent. The results of this study indicate that the SARIMA model is the best forecasting method.

Kata Kunci:  Regresi Linier, Beban Listrik, Time Series, PT. PLN (Persero) Tarakan.