Model Gabungan (Ansambel) Sarima Dan Jaringan Saraf Tiruan Untuk Peramalan Beban Listrik

Abstract

This study aims to investigate the efficacy of employing artificial neural networks in conjunction with a seasonal autoregressive integrated moving average (SARIMA) ensemble for forecasting electrical load. The SARIMA ensemble comprises members generated by varying autoregressive orders or moving averages. Subsequently, these SARIMA ensemble members are integrated using artificial neural networks. The datasets encompass monthly electrical load data pertaining to households, businesses, industries, and the public, spanning from January 2016 to December 2020. The findings demonstrate that across various categories, SARIMA ensemble-based artificial neural networks demonstrated superior predictive performance compared to alternative models. Future research endeavors should focus on exploring diverse methodologies for both creating and amalgamating ensemble members.