Non-intrusive load disaggregation technology offers significant potential for power resource optimization in smart grids. However, several challenges are faced by existing methods. Long-term temporal dependencies are inadequately modeled. Diverse appliance features are captured imprecisely. Predictions between power and state are also inconsistent. A multi-task non-intrusive load disaggregation architecture was proposed to address these limitations. The architecture was based on a transformer and a multi-scale attention convolutional network. Two parallel branches were contained within the architecture. These branches were designated as power disaggregation and state recognition. A transformer structure was used in the power disaggregation branch. A multi-head self-attention mechanism was employed by this structure to capture long-range dependencies and dynamic patterns in the load sequence. In the state recognition branch, a dedicated convolutional neural network was designed. Multi-scale convolutional modules and a channel-spatial attention mechanism were utilized by this network to extract key features of appliance switching events. Inter-task consistency was enhanced through a specific fusion mechanism. Element-wise multiplication was performed between predicted power values and state probabilities. The power output was directly constrained by the use of state information. As a result, appliances were identified in off-states. In these states, power false positives were significantly suppressed. The proposed model was evaluated on the public REDD (reference energy disaggregation dataset) and UK-DALE (UK domestic appliance-level electricity) datasets. Superior performance over existing mainstream methods is achieved. This performance is demonstrated across key evaluation metrics. These metrics are MAE (mean absolute error), SAE (signal aggregate error), and the F1-score.
CHENG Peng-ju, FAN Yan-fang, LIU Jiang-feng, LIN Hong.
Multi-task Non-intrusive Load Disaggregation Method Based on Transformer and Multi-scale Attention Convolutional Network[J].
Science Technology and Engineering, 2026, 26(6): 2442-2452 DOI:10.12404/j.issn.1671-1815.2503008
SiDajun, YaoYiming, LiLingfang, et al. Optimal configuration method of grid-side hybrid energy storage for renewable energy consumption[J]. New Technology of Electrical Engineering and Electric Power, 2024, 43(6): 90-100.
ZhuoZhenyu, ZhangNing, XieXiaorong, et al. Key technologies and development challenges of high proportion renewable energy power systems[J]. Automation of Electric Power Systems, 2021, 45(9): 171-191.
LiYifei, WangFang, ZhangYajing, et al. Non-intrusive load identification method for multi-energy equipment in integrated energy systems considering spatio-temporal coupling characteristics[J]. Science Technology and Engineering, 2024, 24(26): 11283-11293.
ZhangRuiqi, LiuBo, WeiZun. Joint perception method of source-load state in non-invasive integrated energy systems[J]. Procee-dings of the CSEE, 2024, 44(8): 3105-3115.
RuanYingjun, HouZequn, QianFanyue, et al. Operation optimization of distributed energy systems based on deep reinforcement learning[J]. Science Technology and Engineering, 2022, 22(17): 7021-7030.
LuanWenpeng, WeiZun, LiuBo, et al. Test and evaluation method of non-intrusive load monitoring algorithm[J]. Power System Technology, 2022, 46(11): 4568-4579.
LeiYiqin, SunZhaolong, YeZhihao, et al. Research on non-invasive monitoring method for power system load[J]. Transactions of China Electrotechnical Society, 2021, 36(11): 2288-2297.
YangJinnan, LiuBo, LuanWenpeng, et al. Optimization matching method of nilm load events in multi-electrical stack scenarios[J]. Power System Technology, 2022, 46(2): 691-698.
LinShunfu, ZhanYinfeng, LiYi, et al. Non-invasive residential load monitoring method based on CNN-BiLSTM and DTW[J]. Power System Technology, 2022, 46(5): 1973-1981.
ZhouDongguo, ZhangHeng, ZhouHong, et al. Non-intrusive load event detection method based on state feature clustering[J]. Transactions of China Electrotechnical Society, 2020, 35(21): 4565-4575.
[24]
BozorgM, BracaleA, CaramiaP, et al. Bayesian bootstrap quantile regression for probabilistic photovoltaic power forecasting[J]. Protection and Control of Modern Power Systems, 2020, 5(3): 218-229.
[25]
KolterJ Z, JohnsonM J. REDD: a public data set for energy disaggregation research[C]// Workshop on Data Mining Applications in Sustainability (SIGKDD). San Diego: SIGKDD, 2011: 59-62.
[26]
MakoninS, PopowichF, BajicI V, et al. Exploiting HMM sparsity to perform online real-time nonintrusive load monitoring[J]. IEEE Transactions on Smart Grid, 2015, 7(6): 2575-2585.
[27]
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 Environments. New York: ACM, 2015: 55-64.
[28]
VermaS, SinghS, MajumdarA. Multi-label LSTM autoencoder for non-intrusive appliance load monitoring[J]. Electric Power Systems Research, 2021, 199. DOI: 10.1016/j.epsr.2021.107414.
AnQi, WangZhanbin, AnGuoqing, et al. Non-intrusive load identification method based on random forest-genetic algorithm-extreme learning machine[J]. Science Technology and Engineering, 2022, 22(5): 1929-1935.
[31]
MassiddaL, MarrocuM, MancaS. Non-intrusive load disaggregation by convolutional neural network and multilabel classification[J]. Applied Sciences, 2020, 10(4).DOI: 10.3390/app10041454.
[32]
PanG B, WangH P, TianT, et al. Research on non-intrusive load decomposition model based on parallel multi-scale attention mechanism and its application in smart grid[J]. Energy and Buildings, 2024, 312. DOI: 10.1016/j.enbuild.2024.114210.
WangDanyu, LiuJun, ZhouYatong, et al. Feature fusion and transformer model for electricity load disaggregation[J]. Proceedings of the CSU-EPSA, 2024, 36(6): 129-136.
[37]
ShinC, JooS, YimJ, et al. Subtask gated networks for non-intrusive load monitoring.[C]// Proceedings of the AAAI Conference on Artificial Intelligence. Washington: AAAI, 2019: 1150-1157.
YangXiu, LiAn, SunGaiping, et al. Research on a non-intrusive load monitoring method based on improved GMM-CNN-GRU hybrid[J]. Power System Protection and Control, 2022, 50(14): 65-75.
YangXiu, WuJihai, SunGaiping, et al. Non-intrusive load decomposition of public buildings based on deep learning and transfer learning[J]. Power System Technology, 2022, 46(3): 1160-1167.
[42]
邵帅. 不同场景下工商业非侵入式负荷监测技术研究[D]. 淮南: 安徽理工大学, 2023.
[43]
ShaoShuai. Research on non-intrusive load monitoring technology for commercial and industrial loads in different scenarios[D]. Huainan: Anhui University of Science and Technology, 2023.
[44]
YueZ R, WitzigC R, JordeD, et al. BERT4NILM: a bidirectional transformer model for non-intrusive load monitoring[C]// Proceedings of the 5th International Workshop on Non-Intrusive Load Monitoring. Washington: AAAI, 2020: 89-93.
[45]
何敏瑶. 非侵入式家庭用户负荷监测算法研究[D]. 西安: 西安理工大学, 2021.
[46]
HeMinyao. Research on non-intrusive load monitoring algorithms for household users[D]. Xi’an: Xi’an University of Technology, 2021.
JiangYouhua, YeMengdou, ZhaoLe, et al. Non-intrusive user abnormal behavior detection based on seq2seq and SVM dual-layer fusion[J]. Computer Applications and Software, 2024, 41(9): 97-105.
ChenJiawei, JiTianyao, MeiGuang, et al. Non-intrusive load monitoring method based on multi-scale feature fusion and multi-task learning framework[J]. Power System Technology, 2024, 48(5): 2074-2083.
[51]
ZhangC Y, ZhongM, WangZ Z, et al. Sequence-to-point learning with neural networks for non-intrusive load monitoring[C]// Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence. New Orleans: AAAI Press, 2018: 2604-2611.
[52]
ÇimenH, WuY, WuY P, et al. Deep learning-based probabilistic autoencoder for residential energy disaggregation: an adversarial approach[J]. IEEE Transactions on Industrial Informatics, 2022, 18: 8399-8408.