[1]鄢仁武,田堃霖,吴慧敏,等.融合小波散射与注意力时间卷积的非侵入式负荷监测方法[J].福建理工大学学报,2026,24(01):1-10.[doi:10.3969/j.issn.2097-3853.2026.01.001]
　YAN Renwu,TIAN Kunlin,WU Huimin,et al.Non-intrusive load monitoring method integrating wavelet scattering and attention time convolution[J].Journal of Fujian University of Technology;,2026,24(01):1-10.[doi:10.3969/j.issn.2097-3853.2026.01.001]
点击复制

融合小波散射与注意力时间卷积的非侵入式负荷监测方法()
分享到：

《福建理工大学学报》[ISSN:2097-3853/CN:35-1351/Z]

卷:
第24卷
期数:
2026年01期
页码:
1-10
栏目:
出版日期:
2026-02-25

文章信息/Info

Title:
Non-intrusive load monitoring method integrating wavelet scattering and attention time convolution
作者:
鄢仁武; 田堃霖; 吴慧敏; 梁涔; 李培强
福建理工大学电子电气与物理学院
Author(s):
YAN Renwu;  TIAN Kunlin;  WU Huimin;  LIANG Cen;  LI Peiqiang
School of Electronic, Electrical Engineering and Physics, Fujian University of Technology
关键词:
非侵入式负荷监测; 多任务学习; 时间卷积; 小波散射; 注意力网络
Keywords:
non-intrusive load monitoring; multi-task learning; time convolution; wavelet scattering; attention network
分类号:
TP391
DOI:
10.3969/j.issn.2097-3853.2026.01.001
文献标志码:
A
摘要:
面对建筑能源消耗的持续增长和对精细用电管理的迫切需求,现有的负荷监测方法在处理复杂信号时的计算复杂度和存储成本较高。为此,提出一种经济高效且易于实现的非侵入式电力负荷监测方法。首先,采用融合小波散射与时间卷积的神经网络架构精确捕捉总负荷数据的上下文信息;通过残差连接实现多尺度特征的整合,从而增强模型对负荷信号细节和整体模式的识别能力。其次,引入高效通道注意力网络,提升模型对负荷关键特征的识别精度,并提高网络的训练效率。最后,通过结合多任务学习框架和阈值分析法技术,有效降低误判率,增强对电器特征的识别准确性。在UK?DALE 数据集上的比较实验结果验证了该方法相较于现有典型方法在负荷分解和负荷辨识任务上的优越性能。
Abstract:
The continuous rise in building energy consumption and the growing demand for precise electricity management have exposed the limitations of existing load monitoring methods, which incur high computational and storage costs. To address these issues, a cost-effective and easily implementable non-intrusive load monitoring method is proposed. First, a neural network that integrates wavelet scattering and time convolution is used to capture the contextual information of total load data accurately. The integration of multi-scale features is realized by residual connection, so as to enhance the model’s ability to identify both detailed and overall load patterns. Second, an efficient channel attention network is introduced to improve the recognition accuracy of the model for the key features of the load and improve the training efficiency of the network. Finally, the integration of a multi-task learning framework and threshold analysis method effectively reduces false detection rates and improves appliance recognition accuracy. Comparative experiments on the UK-DALE dataset demonstrate that the proposed model outperforms existing models in load disaggregation and identification tasks.

参考文献/References:

［1］ YANG Y D，ZHONG J，LI W，et al. Semisupervised multilabel deep learning based nonintrusive load monitoring in smart grids［J］. IEEE Transactions on Industrial Informatics，2020，16（11）：68926902. ［2］ GILLIS J M，MORSI W G. Nonintrusive load monitoring using semisupervised machine learning and wavelet design［J］. IEEE Transactions on Smart Grid，2017，8（6）：26482655. ［3］HE J L, LIU J M, ZHANG Z J, et al. MSDC: exploiting multistate power consumption in nonintrusive load monitoring based on a dualCNN model ［C］∥AAAI Conference on Artificial Intelligence. Menlo Park: AAAI Press, 2023： 50785086. ［4］KELLY J, KNOTTENBELT W. Neural NILM: deep neural networks applied to energy disaggregation ［C］∥In Proceedings of the 2nd ACM International Conference on Embedded Systems for EnergyEfficient Built Environments，NY：ACM，2015：55-64. ［5］ ZHANG C Y，ZHONG M J，WANG Z Z，et al. Sequencetopoint learning with neural networks for nonintrusive load monitoring［J］. Proceedings of the AAAI Conference on Artificial Intelligence，2018，32（1）：26042611. ［6］ JIANG J，KONG Q Q，PLUMBLEY M D，et al. Deep learningbased energy disaggregation and on/off detection of household appliances［J］. ACM Transactions on Knowledge Discovery from Data，2021，15（3）：121. ［7］ 袁博，葛少云，刘洪，等. 基于压缩感知的非侵入式负荷监测［J］. 中国电机工程学报，2025，45（3）：12051219. ［8］ 邓旭晖，陈中，杨凯，等. 基于多任务学习卷积网络的非侵入式负荷监测方法［J］. 电力系统自动化，2023，47（8）：189197. ［9］ PRECIOSO D，GMEZULLATE D. Thresholding methods in nonintrusive load monitoring［J］. The Journal of Supercomputing，2023，79（13）：1403914062. ［10］ HUBER P，CALATRONI A，RUMSCH A，et al. Review on deep neural networks applied to lowfrequency NILM［J］. Energies，2021，14（9）：2390. ［11］ JIA Z Y，YANG L F，ZHANG Z R，et al. Sequence to point learning based on bidirectional dilated residual network for nonintrusive load monitoring［J］. International Journal of Electrical Power & Energy Systems，2021，129：106837. ［12］ WANG Q L，WU B G，ZHU P F，et al. ECAnet：efficient channel attention for deep convolutional neural networks［C］∥2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle：IEEE，2020：1153111539. ［13］ KIM H, PARK J, BENNIS M, et al. Blockchained ondevice federated learing［J］IEEE Communications letters, 2019, 24(6): 12791283. ［14］ BRUNDAGE M，AVIN S，CLARK J，et al. The malicious use of artificial intelligence：forecasting，prevention，and mitigation［J］. ArXiv ePrints，2018：arXiv：1802.07228.

更新日期/Last Update: 2026-02-25