In smart-meter based non-intrusive load disaggregation with low-frequency sampling, load switching events are sparse and the class distribution is imbalanced. The shortage of minority-class and boundary samples tends to cause missed detections and misclassifications during load ON states and transition stages. Existing oversampling methods still have limitations in precisely controlling the number of synthesized samples, selecting nearest neighbors, and defining boundary samples, and random interpolation strategies may further introduce redundant samples or cross-class mixed samples. To address these issues, this paper proposes an improved algorithm that combines K-means clustering with Borderline-SMOTE (KB-SMOTE): minority boundary samples are first extracted and clustered, and then centroid-guided within-cluster interpolation is performed to generate new samples, thereby reducing redundancy and enhancing boundary separability. At the model level, to overcome the limited capability of conventional sequence networks in capturing key transient and local features, a Bi-LSTM based load disaggregation model embedded with a convolutional block attention module is designed. By jointly leveraging channel and spatial attention, the model adaptively reweights features and strengthens key information relevant to load operating states. Experiments on the UK-DALE dataset show that, compared with baseline models including DAE, Seq2point, and the basic Bi-LSTM, the proposed method achieves better performance on multiple evaluation metrics, validating its effectiveness in imbalanced-load scenarios.
HartG W. Nonintrusive appliance load monitoring[J]. Proceedings of the IEEE, 1992, 80(12):1870-1891.
[2]
ChawlaN V, BowyerK W, HallL O, et al. SMOTE: synthetic minority over-sampling technique[J]. Journal of Artificial Intelligence Research, 2002, 16(1):321-357.
[3]
HanH, WangW Y, MaoB H. Borderline-SMOTE: a new over-sampling method in imbalanced data sets learning[C]∥International Conference on Intelligent Computing, Hefei,China, 2005: 878-887.
[4]
GuoY, XiongX, FuQ, et al. Research on non-intrusive load disaggregation method based on multi-model combination[J]. Electric Power Systems Research, 2021, 200: No.107472.
[5]
GalesM, YoungS. The application of hidden Markov models in speech recognition[J]. Foundations and Trends in Signal Processing, 2008, 1(3): 195-304.
[6]
UpadhyayA, SharmaS K, UpadhyayS. Face identification and verification using hidden Markov model with maximum score approach[J]. Indian Journal of Science and Technology, 2017, 10(47): 671-677.
[7]
KimH, MarwahM, ArlittM, et al. Unsupervised disaggregation of low frequency power measurements[C]∥Proceedings of the SIAM International Conference on Data, Mining, Mesa,USA, 2011: 747-758.
[8]
PattemS. Unsupervised disaggregation for non-intrusive load monitoring[C]∥The 11th International Conference on Machine Learning and Applications(ICMLA), Boca Raton,USA, 2012: 515-520.
[9]
ZohaA, GluhakA, NatiM, et al. Low-power appliance monitoring using factorial hidden Markov models[C]∥IEEE 8th International Conference on Intelligent Sensors, Sensor Networks and Information Processing,Melbourne,Australia, 2013: 527-532.
[10]
WangX N, WangJ H, ShiD, et al. A factorial hidden Markov model for energy disaggregation based on human behavior analysis[C]∥Proceedings of the IEEE Power & Energy Society General Meeting (PESGM), Portland, USA,2018: 1-5.
[11]
KolterJ Z, JaakkolaT. Approximate inference in additive factorial HMMs with application to energy disaggregation[C]∥Proceedings of the 5th International Conference on Artificial Intelligence and Statistics, La Palma,Spain, 2012: 1472-1482.
[12]
KellyJ, KnottenbeltW. Neural NILM: deep neural networks applied to energy disaggregation[C]∥Proceedings of the 2nd ACM International Conference on Embedded Systems for Energy-Efficient Built Enviroments, New York,USA, 2015: 55-64.
[13]
ZhangC, ZhongM, WangZ, et al. Sequence-to-point learning with neural networks for non-intrusive load monitoring[C]∥The 32nd AAAI Conference on Artificial Intelligence, New Orleans, USA, 2018: 2604-2611.
[14]
WuQ, WangF. Concatenate convolutional neural networks for non-intrusive load monitoring across complex background[J]. Energies, 2019, 12(8): 1572-1576.
De AguiarE L, Da Silva NolascoL, LazzarettiA E, et al. St-nilm: a wavelet scattering-based architecture for feature extraction and multilabel classification in nilm signals[J]. IEEE Sensors Journal, 2024, 24(7): 10540-10550.
[17]
WooS, ParkJ, LeeJ Y, et al. CBAM: convolutional block attention module[C]∥Proceedings of the European Conference on Computer Vision (ECCV),Munich,Germany, 2018: 3-19.
[18]
HuJ, ShenL, SunG,et al. Squeeze-and-excitation networks[C]∥IEEE/CVF Confernece on Computer Vision and Pattern Recognition,Salt Lake City, USA,2018:7132-7141.
[19]
SimonsenJ, JensenO S. Contact quality in participation: a "sensethic" perspective[C]∥Proceedings of the 14th Participatory Design Conference: Short Papers, Interactive Exhibitions, Workshops,Aarhus,Denmark, 2016: 45-48.
[20]
KellyJ, KnottenbeltW. The UK-DALE dataset, domestic appliance-level electricity demand and whole-house demand from five UK homes[J]. Scientific Data, 2015, 2(1): 1-14.
[21]
NalmpantisC, VrakasD. Machine learning approaches for non-intrusive load monitoring: from qualitative to quantitative comparation[J]. Artificial Intelligence Review, 2019, 52(1): 217-243.