Non-Intrusive Load Monitoring by Multi-constraints Elastic-Net Regression

DOI: https://doi.org/10.62517/jiem.202603301

Author(s)

Yu Ran1, Zhixin Zhang2*

Affiliation(s)

1Basic Department, Southwest Jiaotong University Hope College, China 2School of Big Data and Statistics, Sichuan Tourism University, China *Corresponding Author

Abstract

Non-intrusive load monitoring (NILM) is a critical technology in smart energy management, enabling detailed appliance-level energy disaggregation from a single aggregated power signal without the need for individual sensors. In this paper, we introduce a novel regression-based approach for NILM that incorporates multiple physical constraints—sparsity, smoothness, and finite-state consistency—into a unified optimization framework. The proposed model enhances the classical Elastic Net method by integrating domain-specific knowledge to better represent the operating characteristics of household appliances. Specifically, the added constraints help to promote solutions that are not only sparse and stable but also physically plausible, leading to more ac- curate identification of appliance states and power consumption patterns. Evaluated on the widely-used AMPds and UK-DALE dataset, our method demonstrates a marked improvement over Elastic Net, Lasso, GSP and HMM, with an average increase in R2 score performance of approximately 5%-7% across a diverse set of appliances in AMPds and 2%-6% in UK-DALE. Additionally, some complex appliances will get significant improved results by our method. The results under- score the importance of incorporating structural prior knowledge into regression models for NILM and highlight the potential of the method for practical applications in energy analytic and residential load monitoring.

Keywords

Non-Intrusive Load Monitoring Multi-Constraint Regression; Energy Disaggregation; Elastic Net; Appliance Identification

References

[1] Araveeporn, A. (2021). The higher-order of adaptive Lasso and elastic net methods for classification on high dimensional data. Mathematics, 9(10), 1091. https://doi.org/10.3390/math9101091 [2] Azzini, H. A. D., Torquato, R., & da Silva, L. C. P. (2014). Event detection methods for nonintrusive load monitoring. In IEEE PES General Meeting Conference & Exposition, 1-5. https://doi.org/10.1109/PESGM.2014.6939797 [3] Breyer J, Alizai M H, Samant S, et al. Advanced Filtering of Unknown Devices in Event-Based NILM. ACM SIGENERGY Energy Informatics Review, 2024, 4(5): 8-16. https://doi.org/10.1145/3727200.3727204 [4] Cui, M., Han, D., Liu, H., Li, K.-C., Tang, M., Chang, C.-C., ..., & Guan, Y. L. (2024). Secure data sharing for consortium blockchain-enabled vehicular social networks. IEEE Transactions on Vehicular Technology, 73(12), 19682-19695. https://doi.org/10.1109/TVT.2024.3448207 [5] Dai, S., Meng, F., Wang, Q., & Chen, X. (2024). DP2-NILM: A distributed and privacy-preserving framework for non-intrusive load monitoring. Renewable and Sustainable Energy Reviews, 191, 114091. https://doi.org/10.1016/j.rser.2023.114091 [6] Dai, S., Wang, Q., & Meng, F. (2021). A telehealth framework for dementia care: An ADLs patterns recognition model for patients based on NILM. In International Joint Conference on Neural Networks, 1-8. https://doi.org/10.1109/IJCNN52387.2021.9534058 [7] Du, Z., Yin, B., Zhu, Y. et al. A NILM load identification method based on structured V-I mapping. Sci Rep 13, 21276 (2023). https://doi.org/10.1038/s41598-023-48736-8 [8] Ghosh S, Chatterjee A, Chatterjee D. Extraction of statistical features for type-2 fuzzy NILM with IoT enabled control in a smart home. Expert Systems with Applications, 2023, 212: 118750.https://doi.org/10.1016/j.eswa.2022.118750 [9] Ghosh S, Mitra A, Chakrabarti S, et al. Decomposition-transformation assisted optimized heterogeneous classification strategy in NILM. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 1-11.https://doi.org/10.1109/TIM.2023.3291008 [10] Han, D., Zhu, Y., Li, D., Liang, W., Souri, A., & Li, K.-C. (2022). A blockchain-based auditable access control system for private data in service-centric IoT environments. IEEE Transactions on Industrial Informatics, 18(5), 3530-3540. https://doi.org/10.1109/TII.2021.3114621 [11] Hao P, Yan Z, Liang J, et al. Self-Supervised Event Detection and Labeling Method for Non-Intrusive Load Monitoring. IEEE Transactions on Instrumentation and Measurement, 2025.https://doi.org/10.1109/TIM.2025.3613888 [12] Huang, P., Chen, X., Feng, X., Wang, Q., Zhang, D., Han, L., & Han, J. (2026). DecoupleNet: Domain-specific task decoupling network for low-light image enhancement. Pattern Recognition, 171, 112203. https://doi.org/10.1016/j.patcog.2025.112203 [13] Huang, Y., Liu, K., Chen, T., Xu, Y., & Ji, H. (2026). Deep intrinsic image decomposition via physics-aware neural networks. Pattern Recognition, 172, 112299. https://doi.org/10.1016/j.patcog.2025.112299 [14] Ibrahem, M. I., & Fouda, M. M. (2024). A lightweight privacy-preserving load forecasting and monitoring scheme supporting dynamic billing for smart grids: No KDC required. IEEE Internet of Things Journal, 11(19), 32160-32171. https://doi.org/10.1109/JIOT.2024.3426486 [15] Jiang, T., Li, Y., Li, Y., Xing, W., Yu, M., Xie, F., & Ta, D. (2026). A segmentation knowledge-based global-local attention network for tumor classification in breast ultrasound images. Pattern Recognition, 171, 112152. https://doi.org/10.1016/j.patcog.2025.112152 [16] Kaselimi, M., Doulamis, N., Doulamis, A., Voulodimos, A., & Protopapadakis, E. (2019). Bayesian-optimized bidirectional LSTM regression model for non-intrusive load monitoring. In 2019 IEEE International Conference on Acoustics, Speech and Signal Processing, 2747-2751. https://doi.org/10.1109/ICASSP.2019.8683110 [17] Lam, H. Y., Fung, G., & Lee, W. (2007). A novel method to construct taxonomy electrical appliances based on load signaturesof. IEEE Transactions on Consumer Electronics, 53(2), 653-660. https://doi.org/10.1109/TCE.2007.381742 [18] Li, X., Zhao, B., Luan, W., & Liu, B. (2022). A training-free non-intrusive load monitoring approach for high-frequency measurements based on graph signal processing. In Asia Conference on Power and Electrical Engineering, 859-863. https://doi.org/10.1109/ACPEE53904.2022.9783715 [19] Lu, J., Yu, Z., Zhang, J., Zhao, R., Li, K., & Zeng, K. (2023). Identification of appliance V-I trajectory based on hash image retrieval method. In Power System and Green Energy Conference, 1146-1151. https://doi.org/10.1109/PSGEC58411.2023.10255957 [20] Luan, W., Yang, F., Zhao, B., & Liu, B. (2022). Industrial load disaggregation based on Hidden Markov Models. Electric Power Systems Research, 210, 108086. https://doi.org/10.1016/j.epsr.2022.108086 [21] Ma, R., Kang, W., Sha, J., Liu, J., & Zhu, D. (2024). NILM load identification based on CNN and V-I trajectory. In International Conference on Power Science and Technology, 2079-2084. https://doi.org/10.1109/ICPST61417.2024.10601881 [22] Mitra, A., Ghosh, S., Mohapatra, A., & Chakrabarti, S. (2024). Appliance identification via combinatorial fusion analysis-assisted Bayesian-optimized classifier. IEEE Transactions on Smart Grid, 15(1), 1165-1176. https://doi.org/10.1109/TSG.2023.3286949 [23] Musib, L., Coletti, R., Lopes, M. B., Mouriño, H., & Carrasquinha, E. (2024). Priority-Elastic net for binary disease outcome prediction based on multi-omics data. BioData Mining, 17(1), 45. https://doi.org/10.1186/s13040-024-00401-0 [24] Pereira, L., & Nunes, N. (2018). Performance evaluation in non-intrusive load monitoring: Datasets, metrics, and tools—A review. WIREs Data Mining and Knowledge Discovery, 8(6), e1265. https://doi.org/10.1002/widm.1265 [25] Pérez-Lombard, L., Ortiz, J., & Pout, C. (2008). A review on buildings energy consumption information. Energy and Buildings, 40(3), 394-398. https://doi.org/10.1016/j.enbuild.2007.03.007 [26] Pötter, H., Lee, S., & Mossé, D. (2021). Towards privacy-preserving framework for non-intrusive load monitoring. In Proceedings of the Twelfth ACM International Conference on Future Energy Systems, 259-263. https://doi.org/10.1145/3447555.3464873 [27] Regan, J., Saffari, M., & Khodayar, M. (2022). Deep attention and generative neural networks for nonintrusive load monitoring. The Electricity Journal, 35(5), 107127. https://doi.org/10.1016/j.tej.2022.107127 [28] Reka, S. S., Dragicevic, T., Venugopal, P., Ravi, V., & Rajagopal, M. K. (2024). Big data analytics and artificial intelligence aspects for privacy and security concerns for demand response modelling in smart grid: A futuristic approach. Heliyon, 10(15), e35683. https://doi.org/10.1016/j.heliyon.2024.e35683 [29] Song, D., Yan, Q., Wang, H., Yang, J., & Luo, X. (2026). Wavelet-guided diffusion enhancement network with directional learning for single-pixel imaging. Pattern Recognition, 172, 112322. https://doi.org/10.1016/j.patcog.2025.112322 [30] Tanoni, G., Principi, E., & Squartini, S. (2024). Non-intrusive load monitoring in industrial settings: A systematic review. Renewable and Sustainable Energy Reviews, 202, 114703. https://doi.org/10.1016/j.rser.2024.114703 [31] Tibshirani, R. (1996). Regression shrinkage and selection via the Lasso. Journal of the Royal Statistical Society Series B: Statistical Methodology, 58(1), 267-288. https://doi.org/10.1111/j.2517-6161.1996.tb02080.x [32] Wang, A. L., Chen, B. X., Wang, C. G., & Hua, D. (2018). Non-intrusive load monitoring algorithm based on features of V–I trajectory. Electric Power Systems Research, 157, 134-144. https://doi.org/10.1016/j.epsr.2017.12.012 [33] Wang, H., Zhang, J., Lu, C., & Wu, C. (2021). Privacy preserving in non-intrusive load monitoring: A differential privacy perspective. IEEE Transactions on Smart Grid, 12(3), 2529-2543. https://doi.org/10.1109/TSG.2020.3038757 [34] Xiao, Y., & Xia, Y. (2026). A novel image enhancement method based on image decomposition and deep neural networks. Pattern Recognition, 172, 112371. https://doi.org/10.1016/j.patcog.2025.112371 [35] Wu, X., Guo, Y., Yan, M., Li, X., Yao, L., & Gong, G. (2023). Non-intrusive load monitoring using identity library based on structured feature graph and group decision classifier. IEEE Transactions on Smart Grid, 14(3), 1958-1973. https://doi.org/10.1109/TSG.2022.3209213 [36] Deng, X.-P., Zhang, G.-Q., Wei, Q.-L., Peng, W., & Li, C.-D. (2022). Fēi qīnrù shì fùhè jiāncè zòngshù [A survey on the non-intrusive load monitoring]. Acta Automatica Sinica, 48(3), 644-663. https://doi.org/10.16383/j.aas.c200270 [37] Yin, H., Zhou, K., & Yang, S. (2023). Non-intrusive load monitoring by load trajectory and multi-feature based on DCNN. IEEE Transactions on Industrial Informatics, 19(10), 10388-10400. https://doi.org/10.1109/TII.2023.3240924 [38] Yu, Z., Zhao, R., Jia, R., Zheng, W., Zhang, J., & Li, S. (2023). Appliance V-I trajectory detection for NILM based on conditional autoencoder. In IEEE Sustainable Power and Energy Conference, 1-5. https://doi.org/10.1109/iSPEC58282.2023.10402911 [39] Yu, Z., Yao, X., & Liu, H. (2025). Bayesian Elastic Net variable selection and application for spatial quantile panel autoregressive model. Communications in Statistics - Simulation and Computation. Advance online publication. https://doi.org/10.1080/03610918.2025.2450698 [40] Yuan, J., Jin, R., Wang, L., & Wang, T. (2024). A nonintrusive load identification method based on dual-branch attention GRU fusion network. IEEE Transactions on Instrumentation and Measurement, 73, 1-10. https://doi.org/10.1109/TIM.2024.3413200 [41] Zhang, X., Zhu, H., Li, X., Hou, B., Zhao, W., Yi, X., ..., & Jiao, L. (2026). Recurrent progressive fusion-based learning for multi-source remote sensing image classification. Pattern Recognition, 171, 112284. https://doi.org/10.1016/j.patcog.2025.112284 [42] Zhang, Y., Wu, H., Ma, Q., Yang, Q., & Wang, Y. (2023). A learnable image-based load signature construction approach in NILM for appliances identification. IEEE Transactions on Smart Grid, 14(5), 3841-3849. https://doi.org/10.1109/TSG.2023.3239598 [43] Zhao, J., Han, M., Ma, A., Jiang, F., Chen, R., Dong, Y., ..., & Chen, Y. (2024). A machine vision-assisted Argonaute-mediated fluorescence biosensor for the detection of viable Salmonella in food without convoluted DNA extraction and amplification procedures. Journal of Hazardous Materials, 466, 133648. https://doi.org/10.1016/j.jhazmat.2024.133648 [44] Zhao, F., Zhou, J., Ren, Y., Liu, H., Wang, Y., & Xia, W. (2023). Mean teacher for semi-supervised non-intrusive load monitoring based on V-I trajectory. In Asia Conference on Power and Electrical Engineering, 1623-1627. https://doi.org/10.1109/ACPEE56931.2023.10135645 [45] Zhao, R., Lu, J., Liu, B., Yu, Z., Ren, Y., & Zheng, W. (2023). Non-intrusive load identification method based on self-supervised regularization. IEEE Access, 11, 144696-144704. https://doi.org/10.1109/ACCESS.2023.3337385 [46] Zhao, R., Shi, B., Ruan, J., Pan, T., & Dong, B. (2024). Estimating noisy class posterior with part-level labels for noisy label learning. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, 22809-22819. https://doi.org/10.1109/CVPR52733.2024.02152 [47] Zhong, M., Goddard, N., & Sutton, C. (2015). Latent Bayesian melding for integrating individual and population models. In International Conference on Neural Information Processing Systems, 2, 3618-3626. [48] Zhou, K., Zhang, Z., & Lu, X. (2024). Non-intrusive load monitoring based on an efficient deep learning model with local feature extraction. IEEE Transactions on Industrial Informatics, 20(7), 9497-9507. https://doi.org/10.1109/TII.2024.3383521 [49] Zhu, Z., Wei, Z., Yin, B., Zhang, S., & Wang, X. (2017). A novel approach for event detection in non-intrusive load monitoring. In 2017 IEEE Conference on Energy Internet and Energy System Integration, 1-5. https://doi.org/10.1109/EI2.2017.8245715 [50] Zou, H., & Hastie, T. (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society Series B: Statistical Methodology, 67(2), 301-320. https://doi.org/10.1111/j.1467-9868.2005.00503.x