Optimalisasi Sistem Monitoring Dan Prediksi Konsumsi Energi Berbasis IoT Dan Deep Learning Untuk Smart Building Di Kota Palembang

Fadli
Tata Sutabri

Jurnal Ilmiah Nusantara
Vol. 3 No. 1 (2026): Jurnal Ilmiah Nusantara Januari 2026

The increase in urbanization and commercial infrastructure development in Palembang City drives a surge in electrical energy demand. Energy efficiency is crucial, yet conventional reactive building management systems often fail to anticipate waste. This research proposes an integrated Smart Building system combining Internet of Things (IoT) based on Node-RED for real-time monitoring and Deep Learning using Long Short-Term Memory (LSTM) algorithm to predict energy consumption. Simulation dataset was collected for 90 days at 5-minute intervals, covering electrical and environmental parameters. Experimental results show that the proposed LSTM model can predict electricity load 1 hour ahead with high accuracy, achieving Mean Absolute Error (MAE) of 0.78 kW and Root Mean Square Error (RMSE) of 1.05 kW, outperforming ARIMA baseline statistical method. Prediction-based control strategy simulation shows potential energy savings of 8-12% through peak shaving techniques on air conditioning and lighting systems.