EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

AI-Based Smart Energy Monitoring and Predictive Control System

Nishikant Talokar, Tanish Tandon, Swaroop Choukade, Samarth Sutar, Bhausaheb Dane

International Research Journal of Innovation in Science and Technology · 16.11.2026 · 1 (4) · 75-83

DOI: 10.67308/irjist.097 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Research Journal of Innovation in Science and Technology, online erschienen am 16.11.2026. Verfasst von Nishikant Talokar und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Efficient energy management has become increasingly important due to rising electricity demand, climate change concerns, and the growing integration of renewable energy sources. Conventional energy monitoring systems often rely on fixed thresholds and manual intervention, limiting their ability to predict demand and respond proactively to abnormal consumption patterns. This paper presents an AI-based smart energy monitoring and predictive control system that integrates Internet of Things (IoT) technologies with machine learning algorithms for energy monitoring, demand forecasting, anomaly detection, and automated load control. The proposed architecture combines monitoring, forecasting, and relay-based control within a low-cost edge-cloud framework. Linear regression and Long Short-Term Memory (LSTM) models are evaluated using a synthetic household energy consumption dataset comprising 11,520 samples generated at 15-minute intervals. On the held-out test set, the LSTM achieved a root mean square error (RMSE) of 0.202 kW, compared with 0.255 kW for linear regression, with a mean absolute percentage error (MAPE) of 20.5%. The predictive control strategy achieved a peak demand reduction of 20.7%, compared with 6.1% for reactive control. Although all 23 injected anomaly events were detected, the sample-level detection precision was 43.5%. The mean LSTM inference time was 0.59 ms on the development machine. The study also discusses cybersecurity, scalability, and implementation challenges and identifies hardware validation, federated learning, and digital twin integration as potential directions for future research. Since the evaluation is simulation-based, validation using a physical hardware prototype and real-world energy consumption data remains necessary.

Zitieren

APA

Talokar, N., Tandon, T., Choukade, S., Sutar, S., & Dane, B. (2026). AI-Based Smart Energy Monitoring and Predictive Control System. International Research Journal of Innovation in Science and Technology, 1(4), 75-83. https://doi.org/10.67308/irjist.097

Vancouver

Talokar N, Tandon T, Choukade S, Sutar S, Dane B. AI-Based Smart Energy Monitoring and Predictive Control System. International Research Journal of Innovation in Science and Technology. 2026;1(4):75-83. doi:10.67308/irjist.097

BibTeX
@article{Talokar2026,
  author = {Talokar, Nishikant and Tandon, Tanish and Choukade, Swaroop and Sutar, Samarth and Dane, Bhausaheb},
  title = {AI-Based Smart Energy Monitoring and Predictive Control System},
  journal = {International Research Journal of Innovation in Science and Technology},
  year = {2026},
  volume = {1},
  number = {4},
  pages = {75--83},
  doi = {10.67308/irjist.097}
}
Kostenlos starten →