From Energy Disaggregation to Appliances Load Forecasting: A Supervised Deep Learning Approach
Hamid QASSEMI, Inoussa Habou LAOUALI, Saad Dosse BENNANI, Hakim El FADILI
Abstract. Forecasting individual electrical appliance consumption is a major challenge for energy optimization and smart building management. This paper presents a prediction model based on a sequence to point (Seq2Point) CNN-LSTM architecture, designed to estimate appliances consumption from its historical data. Unlike traditional autoregressive models, the Seq2Point approach predicts consumption at a specific time step without relying on previous outputs. This approach reduces the propagation of temporal errors. In this paper we evaluate the performance of the model using UK-DALE dataset under tree different scenarios. The experimental results demonstrate that the proposed approach has a strong ability to forecast the consumption of appliances. The proposed model when calculating the average over four appliances achieves a Mean Absolute Error (MAE) of 2.29 corresponding to reduction of 26% compared to the reference model (MAE=3.10).
Keywords
Load Forecasting, Seq-To-Point, CNN, LSTM, Appliance Consumption
Published online 4/25/2026, 8 pages
Copyright © 2026 by the author(s)
Published under license by Materials Research Forum LLC., Millersville PA, USA
Citation: Hamid QASSEMI, Inoussa Habou LAOUALI, Saad Dosse BENNANI, Hakim El FADILI, From Energy Disaggregation to Appliances Load Forecasting: A Supervised Deep Learning Approach, Materials Research Proceedings, Vol. 64, pp 987-994, 2026
DOI: https://doi.org/10.21741/9781644904091-122
The article was published as article 122 of the book Energy Futures
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
References
[1] Energy, S.P., Heat, G.: Transition to Sustainable Buildings Strategies and Op-portunities to 2050. (2013).
[2] G.-F. Angelis, C. Timplalexis, S. Krinidis, D. Ioannidis, et D. Tzovaras, « NILM applications: Literature review of learning approaches, recent developments and challenges », Energy and Buildings, vol. 261, p. 111951, avr. 2022. https://doi.org/10.1016/j.enbuild.2022.111951
[3] A. Ruano, A. Hernandez, J. Ureña, M. Ruano, et J. Garcia, « NILM Techniques for Intelligent Home Energy Management and Ambient Assisted Living: A Review », Energies, vol. 12, no 11, p. 2203, juin 2019. https://doi.org/10.3390/en12112203
[4] M. Kaselimi, E. Protopapadakis, A. Voulodimos, N. Doulamis, et A. Doulamis, « Towards trustworthy Energy Disaggregation: A review of challenges, methods and perspectives for Non-Intrusive Load Monitoring », 5 juillet 2022, arXiv: arXiv:2207.02009. Consulté le: 16 novembre 2024. [En ligne]. Disponible sur: http://arxiv.org/abs/2207.02009
[5] P. Boopathy et al., « Deep learning for intelligent demand response and smart grids: A comprehensive survey », Computer Science Review, vol. 51, p. 100617, févr. 2024. https://doi.org/10.1016/j.cosrev.2024.100617
[6] P. Huber, A. Calatroni, A. Rumsch, et A. Paice, « Review on Deep Neural Networks Applied to Low-Frequency NILM », Energies, vol. 14, no 9, p. 2390, avr. 2021. https://doi.org/10.3390/en14092390
[7] J. Ouzine, M. Marzouq, S. D. Bennani, K. Lahreche, et H. El Fadili, « Overview of Non-Intrusive Load Monitoring: Probabilistic and Artificial Intelligence approaches », E3S Web Conf., vol. 351, p. 01021, 2022. https://doi.org/10.1051/e3sconf/202235101021
[8] I. H. Laouali, H. Qassemi, M. Marzouq, A. Ruano, S. B. Dosse, et H. El Fadili, « A Non Linear Autoregressive Neural Network Model for Forecasting Appliance Power Consumption », in WITS 2020, S. Bennani, Y. Lakhrissi, G. Khaissidi, A. Mansouri, et Y. Khamlichi, Éd., Singapore: Springer Singapore, 2022, p. 759‑768
[9] M. Abumohsen, A. Y. Owda, et M. Owda, « Electrical Load Forecasting Using LSTM, GRU, and RNN Algorithms », Energies, vol. 16, no 5, p. 2283, févr. 2023. https://doi.org/10.3390/en16052283
[10] C. Wang, X. Li, Y. Shi, W. Jiang, Q. Song, et X. Li, « Load forecasting method based on CNN and extended LSTM », Energy Reports, vol. 12, p. 2452‑2461, déc. 2024. https://doi.org/10.1016/j.egyr.2024.07.030
[11] C. Zhang, M. Zhong, Z. Wang, N. Goddard, et C. Sutton, « Sequence-to-point learning with neural networks for nonintrusive load monitoring », 18 septembre 2017, arXiv: arXiv:1612.09106. doi: 10.48550/arXiv.1612.09106
[12] J. Kelly et W. Knottenbelt, « The UK-DALE dataset, domestic appliance-level electricity demand and whole-house demand from five UK homes », Sci Data, vol. 2, no 1, p. 150007, mars 2015. https://doi.org/10.1038/sdata.2015.7
