Autors: Gunay M., Demir Y., Zhilevski, M. M.
Title: Multi-Task NILM with Anomaly Detection Using a Hybrid CNN–BilSTM–Transformer Model
Keywords: 1D-CNN, anomaly detection, appliance classification, BiLSTM, deep learning, NILM, non-intrusive load monitoring, Transformer Attention, UK-DALE

Abstract: Non-Intrusive Load Monitoring (NILM) enables estimation of the energy use of individual appliances in smart buildings from a single aggregate meter. In practice, however, this task is not straightforward. Signals from different appliances can overlap, and the measured data may also include distortions such as spikes, drops, and noise. To address these issues, this study presents a multi-task triple-hybrid deep learning framework that handles appliance classification and anomaly detection together. The model brings together 1D-CNN, BiLSTM, and Transformer Attention so that local patterns, temporal dependencies, and wider contextual information can be learned within the same structure. It also uses a dual-output design to classify appliance categories and detect anomaly types simultaneously. Experiments were carried out on Building 1 of the UK-DALE dataset with four appliances: kettle, microwave, washer dryer, and fridge freezer. For the anomaly task, synthetic disturbances were added to segmented signal windows and grouped as normal, spike, drop, and noise. To check how well the proposed framework handled different scenarios, it was tested on both the UK-DALE and REDD datasets. Looking at the main UK-DALE results, the model correctly identified appliances 99.48% of the time and spotted anomalies with 98.80% accuracy. A secondary test on the REDD dataset yielded an 86.44% classification score. This proves the architecture can adjust to completely new power grid environments without losing its edge. On top of that, when pitted against standard benchmark models like Seq2Point, this triple-hybrid design clearly does a better job of mapping out complex signal changes. As a result, it yields much stronger anomaly detection metrics.

References

  1. Hart G.W. Nonintrusive appliance load monitoring Proc. IEEE 1992 80 1870 1891 10.1109/5.192069
  2. Parson O. Ghosh S. Weal M. Rogers A. Non-intrusive load monitoring using prior models of general appliance types Proc. AAAI Conf. Artif. Intell. 2012 26 356 362 10.1609/aaai.v26i1.8162
  3. Medeiros A.P. Canha L.N. Bertineti D.P. de Azevedo R.M. Event classification in non-intrusive load monitoring using convolutional neural network Proceedings of the 2019 IEEE PES Innovative Smart Grid Technologies Conference-Latin America (ISGT Latin America) IEEE Piscataway, NJ, USA 2019 1 6
  4. De Diego-Oton L. Fuentes-Jimenez D. Hernandez A. Nieto R. Recurrent LSTM architecture for appliance identification in non-intrusive load monitoring Proceedings of the 2021 IEEE International Instrumentation and Measurement Technology Conference (I2MTC) IEEE Piscataway, NJ, USA 2021 1 6 10.1109/I2MTC50364.2021.9460046
  5. Sykiotis S. Kaselimi M. Doulamis A. Doulamis N. ELECTRIcity: An efficient transformer for non-intrusive load monitoring Sensors 2022 22 2926 10.3390/s22082926 35458907
  6. Nie Z. Yang Y. Xu Q. An ensemble-policy non-intrusive load monitoring technique based entirely on deep feature-guided attention mechanism Energy Build. 2022 273 112356 10.1016/j.enbuild.2022.112356
  7. Zhang C. Zhong M. Wang Z. Goddard N. Sutton C. Sequence-to-point learning with neural networks for non-intrusive load monitoring Proc. AAAI Conf. Artif. Intell. 2018 32 2604 2611 10.1609/aaai.v32i1.11873
  8. Rashid H. Singh P. Stankovic V. Stankovic L. Can non-intrusive load monitoring be used for identifying an appliance’s anomalous behaviour? Appl. Energy 2019 238 796 805 10.1016/j.apenergy.2019.01.061
  9. Kelly J. Knottenbelt W. Neural NILM: Deep neural networks applied to energy disaggregation BuildSys 2015—Proceedings of the 2nd ACM International Conference on Embedded Systems for Energy-Efficient Built Environments Association for Computing Machinery, Inc. New York, NY, USA 2015 55 64 10.1145/2821650.2821672
  10. Huzzat A. Khwaja A.S. Alnoman A.A. Adhikari B. Anpalagan A. Woungang I. GRU-BERT for NILM: A hybrid deep learning architecture for load disaggregation AI 2025 6 238 10.3390/ai6090238
  11. He G. Huang Y. Zhang Y. Zhu Y. Leng Y. Shang N. Zeng J. Pu Z. Hybrid transformer-convolutional neural network approach for non-intrusive load analysis in industrial processes Energies 2025 18 2464 10.3390/en18102464
  12. Ouzine J. Marzouq M. Dosse Bennani S. Lahrech K. El Fadili H. New parallel hybrid PHCNN-GRU deep learning model for multi-output NILM disaggregation Energy Effic. 2025 18 56 10.1007/s12053-025-10308-2
  13. Shin C. Joo S. Yim J. Lee H. Moon T. Rhee W. Subtask gated networks for non-intrusive load monitoring Proc. AAAI Conf. Artif. Intell. 2019 33 1150 1157 10.1609/aaai.v33i01.33011150
  14. Çimen H. Çetinkaya N. Vasquez J.C. Guerrero J.M. A microgrid energy management system based on non-intrusive load monitoring via multitask learning IEEE Trans. Smart Grid 2021 12 977 987 10.1109/TSG.2020.3027491
  15. Dash S. Sahoo N.C. Attention-based multitask probabilistic network for nonintrusive appliance load monitoring IEEE Trans. Instrum. Meas. 2023 72 2513412 10.1109/TIM.2023.3273663
  16. de Diego-Otón L. Hernández Á. Fuentes D. Nieto R. Navarro V.M. Architectural strategies for enhanced NILM classification and anomaly detection: Addressing limited data scenarios Expert Syst. Appl. 2025 282 127756 10.1016/j.eswa.2025.127756
  17. Saha D. Bhattacharjee A. Chowdhury D. Hossain E. Islam M.M. Comprehensive NILM framework: Device type classification and device activity status monitoring using capsule network IEEE Access 2020 8 179995 180009 10.1109/ACCESS.2020.3027664
  18. Hu L. Wei J. Yin L. Convolutional Gated Power Prediction Combined with Multiscale Multilabel Classification for Nonintrusive Load Monitoring IEEE Trans. Instrum. Meas. 2026 in press 10.1109/TIM.2026.3654696
  19. Cheng Y. Zhong Y. Non-intrusive load monitoring based on a combination of transformer and CNN Proceedings of the 2024 4th International Conference on Energy, Power and Electrical Engineering (EPEE 2024) Institute of Electrical and Electronics Engineers Inc. Piscataway, NJ, USA 2024 417 421 10.1109/EPEE63731.2024.10875106
  20. McLaughlin S. Holbert B. Fawaz A. Berthier R. Zonouz S. A multi-sensor energy theft detection framework for advanced metering infrastructures IEEE J. Sel. Areas Commun. 2013 31 1319 1330 10.1109/JSAC.2013.130714
  21. Caruana R. Multitask learning Mach. Learn. 1997 28 41 75 10.1023/A:1007379606734
  22. Ismail Fawaz H. Forestier G. Weber J. Idoumghar L. Muller P.A. Deep learning for time series classification: A review Data Min. Knowl. Discov. 2019 33 917 963 10.1007/s10618-019-00619-1
  23. Kiranyaz S. Avci O. Abdeljaber O. Ince T. Gabbouj M. Inman D.J. 1D convolutional neural networks and applications: A survey Mech. Syst. Signal Process. 2021 151 107398 10.1016/j.ymssp.2020.107398
  24. Hochreiter S. Schmidhuber J. Long short-term memory Neural Comput. 1997 9 1735 1780 10.1162/neco.1997.9.8.1735 9377276
  25. Schuster M. Paliwal K.K. Bidirectional recurrent neural networks IEEE Trans. Signal Process. 1997 45 2673 2681 10.1109/78.650093
  26. Vaswani A. Shazeer N. Parmar N. Uszkoreit J. Jones L. Gomez A.N. Kaiser Ł. Polosukhin I. Attention is all you need Advances in Neural Information Processing Systems 30 Curran Associates, Inc. Red Hook, NY, USA 2017 10.48550/arXiv.1706.03762
  27. Kelly J. Knottenbelt W. The UK-DALE dataset, domestic appliance-level electricity demand and whole-house demand from five UK homes Sci. Data 2015 2 150007 10.1038/sdata.2015.7 25984347
  28. Kolter J.Z. Johnson M.J. REDD: A public data set for energy disaggregation research Proceedings of the Workshop on Data Mining Applications in Sustainability (SIGKDD) Citeseer San Diego, CA, USA 2011 59 62 Available online: https://www.researchgate.net/publication/266597071_REDD_A_Public_Data_Set_for_Energy_Disaggregation_Research (accessed on 11 February 2026)
  29. Irani Azad M. Rajabi R. Estebsari A. Nonintrusive load monitoring (NILM) using a deep learning model with a transformer-based attention mechanism and temporal pooling Electronics 2024 13 407 10.3390/electronics13020407
  30. Dash S. Sahoo N.C. A multi-task deep learning approach for non-intrusive load monitoring of multiple appliances IEEE Trans. Smart Grid 2024 15 3337 3340 10.1109/TSG.2024.3373258
  31. Zai Z. Zhao S. Zhang Z. Li H. Sun N. Non-intrusive load monitoring based on the combination of gate-transformer and CNN Electronics 2023 12 2824 10.3390/electronics12132824
  32. Petralia A. Charpentier P. Kadhi Y. Palpanas T. NILMFormer: Non-intrusive load monitoring that accounts for non-stationarity Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining ACM New York, NY, USA 2025 4761 4772 10.1145/3711896.3737251
  33. UKERC EDC: Data Available online: https://ukerc.rl.ac.uk/cgi-bin/dataDiscover.pl?Action=detail&dataid=b3baa0aa-05ac-4ec8-b608-da5751758698 (accessed on 2 February 2026)

Issue

Energies, vol. 19, 2026, Switzerland, https://doi.org/10.3390/en19132963

Copyright MDPI

Вид: статия в списание, публикация в издание с импакт фактор, публикация в реферирано издание, индексирана в Scopus