To main content

Deep Learning Based Non-Intrusive Load Monitoring for Smart Energy Management in Households

Abstract

The increasing digitalization inside the energy sector is transforming the fact that how residential electricity consumption is analysed and optimized. This study aims to propose and evaluate a Deep learning-based model for Non-Intrusive Load Monitoring (NILM), also called energy disaggregation, designed for the real-world applications in smart homes energy management systems. The framework employs Convolutional Neural Network (CNN) based sequence-to-point (Seq2Point) and sequence-to-sequence (Seq2Seq) model to isolate appliance level consumption from the aggregate power, while addressing generalization challenges across diverse households and appliances type. The results suggest that model choice (Seq2Point vs Seq2Seq) effects the results, for example Seq2Point shows better results in detecting high power, short duration events,highlighting the importance of selecting architecture based on appliance-specific characteristics for optimal NILM performance.The practical implications suggest that the proposed study can enable real-time appliance monitoring, adaptive demand-side management, and sustainable energy optimization in future smart grid environments.

Category

Academic article

Language

Other

Author(s)

  • Parvaneh Zavareh
  • Warunee Soythong
  • Basirah Noor
  • Volker Hoffmann
  • Huamin Ren

Affiliation

  • SINTEF Digital / Sustainable Communication Technologies
  • Kristiania University of Applied Sciences

Year

2026

Published in

ENERGY, International Conference on Smart Grids, Green Communications and IT Energy-aware Technologies

Page(s)

1 - 7

View this publication at Norwegian Research Information Repository