基于金字塔网络的非侵入式负荷辨识及其隐私保护方案

doi:10.19678/j.issn.1000-3428.0067381

摘要/Abstract

摘要： 智能电网融合了信息系统,能够为能源供应提供更有效的解决方案。智能电表是智能电网的关键部分,对智能电表数据的深入研究有助于为智能电网的管理和决策提供有效支持。非侵入式负荷辨识(NILM)技术为需求侧管理提供了技术支撑,但现有方式需要用户和NILM服务端进行数据交互,在这个过程中泄露了隐私信息。针对上述问题,设计了基于2D-卷积神经网络(2D-CNN)金字塔网络的NILM,并采用同态加密和安全多方计算技术进行隐私保护,针对金字塔网络的卷积、全连接、批标准化、平均池化、ReLU和上采样等算子设计隐私保护协议,组合隐私保护算子构建隐私保护的2D-CNN金字塔网络。整个过程没有还原数据和中间结果的原始信息,从而保护了双方隐私。在UK-DALE数据集上的实验结果表明,基于2D-CNN的金字塔网络能够表现出良好的效果,准确率达到95.81%,并且隐私保护的2D-CNN金字塔网络能够在保护客户端数据和服务端模型参数隐私性的情况下保持2D-CNN金字塔网络的推理效果,精确率、召回率和准确率等保持一致。同时,隐私保护的2D-CNN金字塔网络在广域网中计算时间不到5 s,在局域网中不到0.5 s,并且通信量仅需4.79 MB,能够适用于NILM任务的现实场景。

关键词: 智能电网, 非侵入式负荷辨识, 金字塔网络, 同态加密, 安全多方计算

Abstract: Smart grids integrate information systems to provide more effective energy-supply solutions. Smart electricity meters are a key part of smart grids, and in-depth research on smart electricity meter data can provide effective support for smart grid management and decision-making. Non-Intrusive Load Monitoring (NILM) technology provides technical support on the demand-side management; however, existing methods require data interaction between users and NILM servers, resulting in the disclosure of private information during this process. To solve these problems, an NILM based on a pyramid network with a two-dimensional Convolutional Neural Network (2D-CNN) is designed, and privacy is protected by homomorphic encryption and secure multiparty computation technology. Privacy-preserving protocols are designed for operators of pyramid networks, such as convolution, full connection, batch normalization, average pooling, ReLU, and upsampling, and are combined to construct a privacy-preserving 2D-CNN pyramid network. The entire process does not restore the original information contained in the data or the intermediate results, thereby protecting the privacy of both parties. The experimental results on the UK-DALE dataset show that the pyramid network based on 2D-CNN can perform well, with an accuracy of 95.81%, and that the privacy-preserving 2D-CNN pyramid network can maintain the inference performance of the 2D-CNN pyramid network while protecting the privacy of the client data and server model parameters with consistent accuracy and recall. At the same time, the privacy-preserving 2D-CNN pyramid network requires a computation time of less than 5 s in a Wide Area Network(WAN) and less than 0.5 s in a Local Area Network(LAN), with a communication volume of only 4.79 MB, making it suitable for real-world scenarios with NILM tasks.

Key words: smart grid, Non-Intrusive Load Monitoring (NILM), pyramid network, homomorphic encryption, secure multi-party computation

中图分类号:

TP309.7

王以良, 周鹏, 叶卫, 戚伟强. 基于金字塔网络的非侵入式负荷辨识及其隐私保护方案[J]. 计算机工程, 2024, 50(5): 182-189.

WANG Yiliang, ZHOU Peng, YE Wei, QI Weiqiang. Non-Intrusive Load Monitoring and Its Privacy-Preserving Scheme Based on Pyramid Network[J]. Computer Engineering, 2024, 50(5): 182-189.

https://www.ecice06.com/CN/Y2024/V50/I5/182

参考文献

[1] CONNOR P M, BAKER P E, XENIAS D, et al. Policy and regulation for smart grids in the United Kingdom[J]. Renewable and Sustainable Energy Reviews, 2014, 40:269-286.
[2] LAMNATOU C, CHEMISANA D, CRISTOFARI C. Smart grids and smart technologies in relation to photovoltaics, storage systems, buildings and the environment[J]. Renewable Energy, 2022, 185:1376-1391.
[3] HART G W. Nonintrusive appliance load monitoring[J]. Proceedings of the IEEE, 1992, 80(12):1870-1891.
[4] CARRIE ARMEL K, GUPTA A, SHRIMALI G, et al. Is disaggregation the holy grail of energy efficiency? The case of electricity[J]. Energy Policy, 2013, 52:213-234.
[5] MEIDAN Y, BOHADANA M, MATHOV Y, et al. N-BaIoT-network-based detection of IoT botnet attacks using deep autoencoders[J]. IEEE Pervasive Computing, 2018, 17(3):12-22.
[6] HART G W. Residential energy monitoring and computerized surveillance via utility power flows[J]. IEEE Technology and Society Magazine, 1989, 8(2):12-16.
[7] ZHANG C Y, ZHONG M J, WANG Z Z, et al. Sequence-to-point learning with neural networks for non-intrusive load monitoring[EB/OL].[2023-03-08]. https://arxiv.org/pdf/1612.09106.pdf.
[8] SHIN C, JOO S, YIM J, et al. Subtask gated networks for non-intrusive load monitoring[EB/OL].[2023-03-08]. https://arxiv.org/pdf/1811.06692.pdf.
[9] BEJARANO G, DEFAZIO D, RAMESH A. Deep latent generative models for energy disaggregation[C]//Proceedings of the AAAI Conference on Artificial Intelligence[S.l.]:AAAI Press, 2019:850-857.
[10] KONG W C, DONG Z Y, MA J, et al. An extensible approach for non-intrusive load disaggregation with smart meter data[J]. IEEE Transactions on Smart Grid, 2018, 9(4):3362-3372.
[11] XIA D, BA S S, AHMADPOUR A. Non-intrusive load disaggregation of smart home appliances using the IPPO algorithm and FHM model[J]. Sustainable Cities and Society, 2021, 67:102731.
[12] SALEM H, SAYED-MOUCHAWEH M, TAGINA M. Unsupervised Bayesian non parametric approach for non-intrusive load monitoring based on time of usage[J]. Neurocomputing, 2021, 435:239-252.
[13] DESAI S, ALHADAD R, MAHMOOD A, et al. Multi-state energy classifier to evaluate the performance of the NILM algorithm[J]. Sensors, 2019, 19(23):5236.
[14] KELLY J, KNOTTENBELT W. Neural NILM:deep neural networks applied to energy disaggregation[C]//Proceedings of the 2nd International Conference on Embedded Systems for Energy-Efficient Built Environments. New York, USA:ACM Press, 2015:55-64.
[15] 张丽, 张涛, 张宏伟, 等. 一种基于多参量隐马尔可夫模型的负荷辨识方法[J]. 电力系统保护与控制, 2019, 47(20):81-90. ZHANG L, ZHANG T, ZHANG H W, et al. Research on a method of load identification based on multi parameter hidden Markov model[J]. Power System Protection and Control, 2019, 47(20):81-90.(in Chinese)
[16] REN Z R, TANG B, WANG L F, et al. Household appliance identification based on a novel load signature processing framework[C]//Proceedings of the 3rd Conference on Energy Internet and Energy System Integration (EI2). Washington D. C., USA:IEEE Press, 2019:2076-2080.
[17] 张玉天, 邓春宇, 刘沅昆, 等. 基于卷积神经网络的非侵入负荷辨识算法[J]. 电网技术, 2020, 44(6):2038-2044. ZHANG Y T, DENG C Y, LIU Y K, et al. Non-intrusive load identification algorithm based on convolution neural network[J]. Power System Technology, 2020, 44(6):2038-2044.(in Chinese)
[18] 李莉, 黄友金, 熊炜, 等. 基于改进卷积神经网络的非侵入负荷辨识方法研究[J]. 电测与仪表, 2024,61(1):125-130,156. LI L, HUANG Y J, XIONG W, et al. Non-intrusive load identification based on inproved convolutional neural network[J]. Electrical Measurement & Instrumentation, 2024,61(1):125-130,156.(in Chinese)
[19] DAI S, MENG F L, WANG Q, et al. DP$^2$-NILM:a distributed and privacy-preserving framework for non-intrusive load monitoring[EB/OL].[2023-03-08]. http://arxiv.org/abs/2207.00041.
[20] ROUHANI B D, RIAZI M S, KOUSHANFAR F. DeepSecure:scalable provably-secure deep learning[C]//Proceedings of the 55th ACM/ESDA/IEEE Design Automation Conference (DAC). New York, USA:ACM Press, 2018:1-6.
[21] JUVEKAR C,VAIKUNTANATHAN V,CHANDRAKASAN A. GAZELLE:a low latency framework for secure neural network inference[C]//Proceedings of the 27th USENIX Conference on Security Symposium. New York, USA:ACM Press, 2018:1651-1668.
[22] MISHRA P, LEHMKUHL R, SRINIVASAN A, et al. Delphi:a cryptographic inference system for neural networks[C]//Proceedings of the Workshop on Privacy-Preserving Machine Learning in Practice. New York, USA:ACM Press, 2020:27-30.
[23] RATHEE D, RATHEE M, KUMAR N, et al. CrypTFlow2:practical 2-party secure inference[C]//Proceedings of the ACM SIGSAC Conference on Computer and Communications Security. New York, USA:ACM Press, 2020:325-342.
[24] HUANG Z, LU W, HONG C, et al. Cheetah:Lean and fast secure {two-party} deep neural network inference[C]//Proceedings of the 31st USENIX Security Symposium.[S.l.]:AAAI Press, 2022:809-826.
[25] ZHAO H S, SHI J P, QI X J, et al. Pyramid scene parsing network[C]//Proceedings of Conference on Computer Vision and Pattern Recognition. Washington D. C., USA:IEEE Press, 2017:6230-6239.
[26] MASSIDDA L, MARROCU M, MANCA S. Non-intrusive load disaggregation by convolutional neural network and multilabel classification[J]. Applied Sciences, 2020, 10(4):1454.
[27] 杜宇, 严萌, 武昕. 基于上采样金字塔结构的卷积神经网络的非侵入负荷辨识算法[J]. 计算机应用, 2022, 42(10):3300-3306. DU Y, YAN M, WU X. Non-intrusive load identification algorithm based on convolutional neural network with upsampling pyramid structure[J]. Journal of Computer Applications, 2022, 42(10):3300-3306.(in Chinese)
[28] KELLY J, KNOTTENBELT W. The UK-DALE dataset, domestic appliance-level electricity demand and whole-house demand from five UK homes[J]. Scientific Data, 2015, 2:150007.

选择文件类型/文献管理软件名称

选择包含的内容