Real-time NILM: A lightweight and low power approach

Ze Dong, Xiaohu Zhang, Shiyi Cai, Yuhang Yang, Wei Jiang, Yaqi Song, Wenqing Zhao and Dongyang Zhang

Energy, 2025, vol. 335, issue C

Abstract: Non-intrusive load monitoring (NILM) is a great significance for energy conservation, cost reduction, and environmental protection. It helps with better energy management and planning by accurately identifying the energy consumption of each load. However, existing research approaches have high complexity and poor real-time performance, making it difficult to be applied in edge devices. Therefore, this paper proposes a real-time lightweight NILM approach that can be applied to edge devices. This approach is composed of the following main steps: (i) We propose a Time&Voltage Dual-Reference-Based Time-Domain Window Subtraction (TVDR-TDWS) Method to separate the transient waveform when a single load is activated from the aggregated transient current data. (ii) A lightweight model with high accuracy and few parameters is designed for load monitoring. The experimental results show that the proposed model can accurately identify the load from the aggregated current data by training with the transient current waveform of a single load activation. The accuracy rates on the TDHA dataset and the PLAID dataset reach 94.69% and 81.92% respectively. (iii) Finally we deployed the lightweight model on a microcontroller with the Cortex-M7 architecture(STM32H7 series), and it has a significant model inference time down to 9.067 ms. This makes a new approach to NILM in practical field applications.

Keywords: Non-intrusive load monitoring; Feature extraction method; Full sequence capture; Feature serialization; Lightweight model; Cortex-M7 (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544225037429
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:335:y:2025:i:c:s0360544225037429

DOI: 10.1016/j.energy.2025.138100

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().