基于样本过滤-标签聚合极端树集成的电力信息物理系统虚假数据注入攻击定位检测

席磊; 李宗泽; 王文卓; 白芳岩; 董璐

doi:10.16383/j.aas.c250275

基于样本过滤-标签聚合极端树集成的电力信息物理系统虚假数据注入攻击定位检测

doi: 10.16383/j.aas.c250275 cstr: 32138.14.j.aas.c250275

席磊^{1, 2,},
李宗泽^1,,
王文卓^{1, 3,},
白芳岩^4,,
董璐^5,

1.
三峡大学电气与新能源学院宜昌 443002
2.
梯级水电站运行与控制湖北省重点实验室(三峡大学) 宜昌 443002
3.
长江三峡通航管理局宜昌 443000
4.
河北省送变电有限公司石家庄 050000
5.
国网湖北超高压公司宜昌运维分部宜昌 443000

基金项目: 国家自然科学基金(52477104), 湖北省高等学校优秀中青年科技创新团队计划(T2020006)资助

详细信息

作者简介:
席磊：三峡大学电气工程与新能源学院教授.2016年获华南理工大学电气工程专业博士学位. 主要研究方向为电力系统稳定与控制和信息物理系统安全防御研究. E-mail: xilei2014@163.com

李宗泽：三峡大学电气与新能源学院博士研究生. 主要研究方向为电力系统网络攻击与防御. 本文通信作者. E-mail: lizongze0608@163.com

王文卓：长江三峡通航管理局见习生. 2025年获得三峡大学电气工程专业硕士学位. 主要研究方向为信息物理系统安全防御研究. E-mail: wangwz030930@163.com

白芳岩：河北省送变电有限公司见习生. 2025年获得三峡大学电气工程专业硕士学位. 主要研究方向为信息物理系统安全防御研究. E-mail: bfyana@163.com

董璐：国网湖北省电力有限公司超高压公司宜昌运维分部见习生. 2025年获得三峡大学电气工程专业硕士学位. 主要研究方向为信息物理系统安全防御研究. E-mail: lu15275425765@163.com

计量
- 文章访问数: 19
- HTML全文浏览量: 12
- 被引次数: 0
出版历程
- 收稿日期: 2025-06-28
- 录用日期: 2025-09-03
- 网络出版日期: 2025-09-22

Location Detection of False Data Injection Attacks in Cyber-Physical Power Systems Based on Sample Filtering-Label Powerset Extreme Trees Ensemble Boost

Xi Lei^{1, 2
,},
Li Zong-Ze^1
,,
Wang Wen-Zhuo^{1, 3
,},
Bai Fang-Yan^4
,,
Dong Lu^5
,

1.
College of Electrical Engineering and New Energy, China Three Gorges University, Yichang 443002
2.
Hubei Provincial Key Laboratory for Operation and Control of Cascaded Hydropower Station, China Three Gorges University, Yichang 443002
3.
Three Gorges Navigation Authority, Yichang 443000
4.
Hebei Power Transmission and Transformation Co., Ltd, Shijiazhuang 050000
5.
Yichang Operation and Maintenance Division of State Grid Hubei Extra High Voltage Company, Yichang 443000

Funds: Supported by National Natural Science Foundation of China (52477104) and Excellent Young and Middle-aged Science and Technology Innovation Team Project of Hubei Universities (T2020006)

More Information

Author Bio:
XI Lei　Professor at the School of Electrical Engineering and New Energy, Three Gorges University.He received his Ph.D. degree in Electrical Engineering from South China University of Technology.His research interest cover power system stability control and the security defense of cyber-physical systems

LI Zong-Ze　Ph.D. candidate at the School of Electrical Engineering and New Energy, China Three Gorges University. His research interest cover the security defense of cyber-physical systems. Corresponding author of this paper

WANG Wen-Zhuo　Intern at Three Gorges Navigation Authority. He received his master's degree in Electrical Engineering from China Three Gorges University in 2025. His research interest cover the security defense of cyber-physical systems

BAI Fang-Yan　Intern at Hebei Power Transmission and Transformation Co., Ltd. He received his master＇s degree in Electrical Engineering from China Three Gorges University in 2025. His research interest cover the security defense of cyber-physical systems

DONG Lu　Intern at Yichang Operation and Maintenance Division of State Grid Hubei Extra High Voltage Company. She received her master＇s degree in Electrical Engineering from China Three Gorges University in 2025. Her research interest cover the security defense of cyber-physical systems

摘要

摘要: 虚假数据注入攻击行为严重威胁电力信息物理系统的安稳运行. 然而, 针对虚假数据注入攻击的现有检测未充分考虑海量量测数据的不平衡性和网络拓扑的关联性, 导致检测方法广泛存在定位性能差的问题, 提出一种基于样本过滤-标签聚合极端树集成的电网虚假数据注入攻击定位检测方法. 所提方法在基于深度学习的卷积神经网络中引入交叉验证思想, 用于过滤海量量测数据中代表性弱、重复率高的不平衡样本; 通过利用标签聚合将关联的网络拓扑融入到极端树中, 并在与各节点状态对应的所有极端树上进行集成, 继而加权输出得到各节点状态的最终检测概率, 以实现对受攻击位置的精确定位. 在IEEE-14、IEEE-57系统上进行大量仿真, 验证了所提方法的有效性, 且与多种已有定位检测方法进行充分对比, 验证其在准确率、精度、召回率、F1值和AUC值上具备更优性.
- 虚假数据注入攻击 /
- 深度学习 /
- 集成学习 /
- 标签聚合极端树集成
Abstract: False data injection attacks seriously threaten the safe and stable operation of the Cyber-Physical Power Systems. However, the existing detection for false data injection attacks does not fully consider the imbalance of massive quantitative measurement data and the correlation of network topology, resulting in a wide range of detection methods that suffer from poor localization performance. In this paper, a Sample Filtering-Label Powerset Extreme Tree Ensemble Boost-based localization and detection method is proposed for false data injection attacks in power grid. The proposed method combines five-fold cross-validation with deep learning-based Convolutional Neural Network for filtering unbalanced samples with weak representativeness and high repetition rate in massive quantitative measurement data; integrates the associated network topology into the extreme tree by utilizing label aggregation and integrates it over all the extreme trees corresponding to the states of each node, and then weights the outputs to obtain the final detection probability of each node＇s state, in order to achieve accurate localization of the attacked precise localization of the location. A large number of simulations on IEEE-14 and IEEE-57 systems are conducted to verify the effectiveness of the proposed method, and a full comparison with multiple existing localization detection methods verifies its superiority in terms of accuracy, precision, recall, F1 value and AUC value.
- False data injection attack /
- Deep learning /
- Ensemble learning /
- Sample filtering-label powerset extra trees ensemble boost

HTML全文

图 1 FDIA行为示意图

Fig. 1 Schematic diagram of FDIA behavior

下载: 全尺寸图片幻灯片

图 2 样本过滤示意图

Fig. 2 Schematic diagram of sample filtering

下载: 全尺寸图片幻灯片

图 3 IEEE-14节点系统标签聚合示例图

Fig. 3 Diagram of IEEE-14 node system label powerset

下载: 全尺寸图片幻灯片

图 4 日内攻击前后残差示意图((a)IEEE-14小节点系统, (b)IEEE-57大节点系统)

Fig. 4 Residual diagram before and after intraday attack((a)IEEE-14 node system, (b)IEEE-57 node system)

下载: 全尺寸图片幻灯片

图 5 IEEE-14节点系统各状态检测准确率示意图

Fig. 5 Schematic diagram of the accuracy of various state detection in IEEE-14 node system

下载: 全尺寸图片幻灯片

图 6 IEEE-14节点系统ROC曲线图

Fig. 6 ROC curve of IEEE-14 node System

下载: 全尺寸图片幻灯片

图 7 IEEE-14节点系统定位检测指标对比图

Fig. 7 Comparison chart of positioning detection indicators for IEEE-14 node system

下载: 全尺寸图片幻灯片

图 8 IEEE-57节点系统各状态检测准确率示意图

Fig. 8 Schematic diagram of the accuracy of various state detection in IEEE-57 node system

下载: 全尺寸图片幻灯片

图 9 IEEE-57节点系统ROC曲线图

Fig. 9 ROC curve of IEEE-57 node System

下载: 全尺寸图片幻灯片

图 10 IEEE-57节点系统定位检测指标对比图

Fig. 10 Comparison chart of positioning detection indicators for IEEE-57 node system

下载: 全尺寸图片幻灯片

表 1 量测单元信息部署表

Table 1 Deployment information of measurement units

节点系统	IEEE-14	IEEE-57
SCADA 配置节点编号	1−14	1−57
PMU 配置节点编号	4, 6	4, 10, 15, 20, 24, 29, 32, 37, 41, 48, 54
SCADA量测量	节点有功/无功功率注入; 节点有功/无功功率流
PMU量测量	节点电压幅值、相角; 电流相量实部、虚部

下载: 导出CSV

表 2 标签编码对照表

Table 2 Tag code comparison table

编码规则	1	2	3	4	5	6	7	8
[9-4]
[9-7]	[0 0]	[0 1]	[1 0]	[1 1]
[9-10]
[9-14]
[9-4-7]	[0 0 0]	[0 0 1]	[0 1 0]	[0 1 1]	[1 0 0]	[1 0 1]	[1 1 0]	[1 1 1]

下载: 导出CSV

参考文献(26)

[1]	龚立, 王先培, 田猛, 等. 电力信息物理系统韧性的概念与提升策略研究进展. 电力系统保护与控制, 2023, 51(14): 169−187 doi: 10.19783/j.cnki.pspc.221536 Gong L, Wang X P, Tian M, Li X X, Zhu Z Y. Concepts and research progress on enhancement strategies for cyber physical power system resilience. Power System Protection and Control, 2023, 51(14): 169−187 doi: 10.19783/j.cnki.pspc.221536
[2]	汤奕, 李梦雅, 王琦, 等. 电力信息物理系统网络攻击与防御研究综述(二)检测与保护. 电力系统自动化, 2019, 43(10): 1−9+18 Tang Y L, Meng Y, Wang Q, Ni M. A Review on Research of Cyber-attacks and Defense in Cyber Physical Power Systems Part Two Detection and Protection. Automation of Electric Power Systems. 2019, 43(10): 1-9+18. 43(10): 1−9+18.
[3]	Chen B Y, Li H B, Zhou B. Real-Time Identification of False Data Injection Attacks: A Novel Dynamic-Static Parallel State Estimation Based Mechanism. in IEEE Access, 2019(7): 95812−95824 doi: 10.1109/ACCESS.2019.2929785
[4]	刘增稷, 王琦, 薛彤, 等. 电力系统中数据驱动算法安全威胁分析及应对方法研究. 中国电机工程学报, 2023, 43(12): 4538−4554 doi: 10.13334/j.0258-8013.pcsee.220853 Liu Z J, Wang Q, Xue T, Tang Y. Research on Security Risks and Defense Methods of Data-driven Algorithms in Power Systems. Proceedings of the CSEE, 2023, 43(12): 4538−4554 doi: 10.13334/j.0258-8013.pcsee.220853
[5]	李卓, 谢耀滨, 吴茜琼, 等. 基于深度学习的电力系统虚假数据注入攻击检测综述. 电力系统保护与控制, 2024, 52(19): 175−187 doi: 10.19783/j.cnki.pspc.231632 Li Z, Xie Y B, Wu Q Q, Zhang Y W. Review of deep learning-based false data injection attack detection in power systems. Power System Protection and Control, 2024, 52(19): 175−187 doi: 10.19783/j.cnki.pspc.231632
[6]	席磊, 董璐, 程琛, 等. 基于混合黑猩猩优化极限学习机的电力信息物理系统虚假数据注入攻击定位检测. 电力系统保护与控制, 2024, 52(14): 46−58 doi: 10.19783/j.cnki.pspc.240042 Xi L, Dong L, Chen C, Tian X L, Li Z Z. Location detection of a false data injection attack in a cyber-physical power system based ona hybrid chimp optimized extreme learning machine. Power System Protection and Control, 2024, 52(14): 46−58 doi: 10.19783/j.cnki.pspc.240042
[7]	席磊, 王艺晓, 何苗, 等. 基于反向鲸鱼-多隐层极限学习机的电网FDIA检测. 中国电力, 2024, 57(9): 20−31 Xi L, Wang Y X, He M, Cheng C, Tian X L. FDIA Detection in Power Grid Based on Opposition-Based Whale Optimization Algorithm and Multi-layer Extreme Learning Machine. Electric Power, 2024, 57(9): 20−31
[8]	席磊, 何苗, 周博奇, 等. 基于改进多隐层极限学习机的电网虚假数据注入攻击检测. 自动化学报, 2023, 49(04): 881−890 doi: 10.16383/j.aas.c211127 Xi L, H eM, Zhou B Q, Li Y Y. esearch on False Data Injection Attack Detection in Power System Based onImproved Multi Layer Extreme Learning Machine. Acta Automatica Sinica, 2023, 49(04): 881−890 doi: 10.16383/j.aas.c211127
[9]	陶磊, 罗萍萍, 林济铿. 基于深度学习的直流微电网虚假数据注入攻击二阶段检测方法. 中国电力, 2024, 57(09): 11−19 doi: 10.11930/j.issn.1004-9649.202311112 Tao L, Luo P P, Lin J K. Two-stage Detection Method for DC Microgrid False Data Injection AttackBased on Deep Learning. Electric Power, 2024, 57(09): 11−19 doi: 10.11930/j.issn.1004-9649.202311112
[10]	席磊, 程琛, 田习龙. 基于改进卷积神经网络的电网虚假数据注入攻击定位方法. 南方电网技术, 2025, 19(01): 74−84 doi: 10.13648/j.cnki.issn1674-0629.2025.01.008 Xi L, Cheng C, Tian X L. Improved Convolutional Neural Network-Based Localization Method for False Data Injection Attacks on Power Grids. Southern Power System Technology, 2025, 19(01): 74−84 doi: 10.13648/j.cnki.issn1674-0629.2025.01.008
[11]	陈柏任, 夏候凯顺, 李梦诗. 基于数据驱动的电力系统虚假数据注入攻击防御框架的研究. 电测与仪表, 2024, 61(12): 10−16 doi: 10.19753/j.issn1001-1390.2024.12.002 Chen B R, Xiahou K S, Li M S. Research on defense framework for false data injection attacks in power system based on data-driven algorithm. Electrical Measurement & Instrumentation, 2024, 61(12): 10−16 doi: 10.19753/j.issn1001-1390.2024.12.002
[12]	苏向敬, 邓超, 栗风永, 等. 基于MGAT-TCN模型的可解释电网虚假数据注入攻击检测方法. 电力系统自动化, 2024, 48(02): 118−127 Su X J, Deng C, Li F Y, Fu Y, Xiao S Q. Interpretable Detection Method for False Data Injection Attack on Power Grid Based on Multi-head Graph Attention Network and Time Convolution Network Model. Automation of Electric Power Systems, 2024, 48(02): 118−127
[13]	Xue W L, Wu T. Active learning-based XGBoost for cyber physical system against generic AC false data injection attacks. in IEEE Access, 2020(8): 144575−144584 doi: 10.1109/ACCESS.2019.2929785
[14]	席磊, 田习龙, 余涛, 等. 基于相关特征-多标签级联提升森林的电网虚假数据注入攻击定位检测. 南方电网技术, 2024, 18(05): 39−50 doi: 10.13648/j.cnki.issn1674-0629.2024.05.005 Xi L, Tian X L, Yu T, Cheng C. Locational Detection of False Data Injection Attack in Power Grid Based onRelevant Features Multi-Label Cascade Boosting Forest. Southern Power System Technology, 2024, 18(05): 39−50 doi: 10.13648/j.cnki.issn1674-0629.2024.05.005
[15]	席磊, 王文卓, 白芳岩, 等. 基于最大信息系数-双层置信极端梯度提升树的电网虚假数据注入攻击定位检测. 电网技术, 2025, 49(02): 824−833 doi: 10.13335/j.1000-3673.pst.2024.0298 Xi L, Wang W Z, Bai F Y, Chen H J, Peng D M, Li Z Z. Grid False Data Injection Attack Localization Detection Based onMIC-double-deck Confidence XGBoost Tree. Power System Technology, 2025, 49(02): 824−833 doi: 10.13335/j.1000-3673.pst.2024.0298
[16]	王琦, 邰伟, 汤奕, 等. 面向电力信息物理系统的虚假数据注入攻击研究综述. 自动化学报, 2019, 45(01): 72−83 doi: 10.16383/j.aas.2018.c180369 Wang Q, Tai W, Tang Y, Ni M. A Review on False Data Injection Attack Toward Cyber-physical Power System. Acta Automatica Sinica, 2019, 45(01): 72−83 doi: 10.16383/j.aas.2018.c180369
[17]	Wu T, Xue W L, Wang H Z. Extreme Learning Machine-Based State Reconstruction for Automatic Attack Filtering in CyberPhysical Power System. IEEE Transactions on Industrial Informatics, 2021, 17(03): 1892−1904 doi: 10.1109/TII.2020.2984315
[18]	王彩强, 张青, 李晨, 等. 含有限PMU的配电网故障区域在线辨识算法. 南方电网技术, 2024, 18(12): 42−50 doi: 10.13648/j.cnki.issn1674-0629.2024.12.006 Wang C Q, Zhang Q, Li C, Huang Z Y, Liang C Y, He S M. Fault Section On-Line Identification Algorithm with Limited PMU forDistribution Network. Southern Power System Technology, 2024, 18(12): 42−50 doi: 10.13648/j.cnki.issn1674-0629.2024.12.006
[19]	Wang H, Wen X, Xu Y, Zhou B, Peng J, Liu W. Operating State Reconstruction in Cyber Physical Smart Grid for Automatic Attack Filtering. IEEE Transactions on Industrial Informatics, 2022, 18(05): 2909−2922 doi: 10.1109/TII.2020.3000172
[20]	Yadav S, Shukla S. Analysis of k-Fold Cross-Validation over Hold-Out Validation on Colossal Datasets for Quality Classification. 2016 IEEE 6th International Conference on Advanced Computing (IACC), 201678−83 doi: 10.1109/IACC.2016.25
[21]	申超波, 王志海, 孙艳歌. 基于标签聚类的多标签分类算法. 软件, 2014, 35(08): 16−21 doi: 10.3969/j.issn.1003-6970.2014.08.004 Shen C B, Wang Z H, Sun Y G. A Multi-Label Classification Algorithm Based on Label Clustering. Software, 2014, 35(08): 16−21 doi: 10.3969/j.issn.1003-6970.2014.08.004
[22]	陈柏任, 夏候凯顺, 李梦诗. 基于数据驱动的电力系统虚假数据注入攻击防御框架的研究. 电测与仪表, 2024, 61(12): 10−16 doi: 10.19753/j.issn1001-1390.2024.12.002 Chen B R, Xiahou K S, Li M S. Research on defense framework for false data injection attacks in power system based on data-driven algorithm. Electrical Measurement & Instrumentation, 2024, 61(12): 10−16 doi: 10.19753/j.issn1001-1390.2024.12.002
[23]	徐先峰, 李芷菡, 刘状壮, 等. 基于半监督学习标签传播-极端随机树算法的光伏阵列故障诊断及定位. 电网技术, 2023, 47(03): 1038−1047 doi: 10.13335/j.1000-3673.pst.2022.0992 Xu X F, Li Z H, Liu Z Z, Wang K, Ma Z X, Yao J J, Cai L L. Fault Diagnosis and Localization of Photovoltaic Arrays Based on Semi-supervised LearningLabel Propagation-extra Tree Algorithm. Power System Technology, 2023, 47(03): 1038−1047 doi: 10.13335/j.1000-3673.pst.2022.0992
[24]	李刚, 李天琦, 程晓荣, 等. 大数据可信性度量方法. 计算机工程与设计, 2017, 38(03): 652−658 doi: 10.16208/j.issn1000-7024.2017.03.018 Li G, Li T Q, Cheng X R, Wang H Y. Credibility measurement method of big data. Computer Engineering and Design, 2017, 38(03): 652−658 doi: 10.16208/j.issn1000-7024.2017.03.018
[25]	林峰, 梅勇, 朱益华, 等. 网络攻击对电力系统典型场景全过程影响综述. 南方电网技术, 2023, 17(11): 61−75 doi: 10.13648/j.cnki.issn1674-0629.2023.11.007 Lin F, Mei Y, Zhu Y H, Chang D X, Liu R, Guo H D. Overview of the Entire Process Influence of Cyber Attack on Typical Scenarios of Power Systems. Southern Power System Technology, 2023, 17(11): 61−75 doi: 10.13648/j.cnki.issn1674-0629.2023.11.007
[26]	Qu Z W, Yang J C, Wang Y J, Georgievitch P M. Detection of False Data Injection Attack in Power System Based on Hellinger Distance. IEEE Transactions on Industrial Informatics, 2020, 20(02): 2119−2128 doi: 10.1109/TII.2023.3286895