Prediction of Peak Load Electricity Consumption in Apartment X Building Using Deep Learning with GRU Method

Yusuf, Yusuf
Rofii, Ahmad
Muliadi, Jemie

Journal of Applied Science and Advanced Engineering
Vol. 4 No. 1 (2026): JASAE: March 2026

This study presents a predictive framework for daily electricity consumption forecasting in Apartment X using a Recurrent Neural Network (RNN) model with the Gated Recurrent Unit (GRU) method. The dataset consists of daily electricity log sheets containing two main variables: Peak Load Time (WBP) and Off-Peak Load Time (LWBP). The preprocessing stage includes data cleaning, normalization using Min–Max Scaling, and sequence formation through a sliding window approach. The GRU architecture comprises two hidden layers, a dropout layer, and optimization using the Adam optimizer. The model’s performance was evaluated using MAE, RMSE, and R². The results show that the GRU model achieved an R² value of 0.623, indicating a good capability in capturing consumption patterns. This study contributes to energy forecasting studies in developing countries, emphasizing smart building energy management applications