基于张量分解个性化联邦学习的网络-光伏爬坡协同攻击辨识方法研究

doi:10.19595/j.cnki.1000-6753.tces.240896

Abstract
Figure/Table
References (0)
Related Citation (2)

Download: PDF (1782 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract

Photovoltaic (PV) power has emerged as one of the most promising and rapidly developing renewable energy sources in recent years, significantly facilitating the green transition of power grids. However, the uncertainty of PV ramping events imposes higher demands on the accuracy and timeliness of cyber attack detection methods. Existing studies overlook the coupling characteristics between PV ramping events and cyber attack events, making it challenging to accurately identify cyber-ramping coordinated attacks. Besides, as the scale of the power grid continues to expand, strict privacy protection requirements between different grid partitions hinder data sharing. Traditional distributed algorithms are constrained by communication conditions, resulting in low training efficiency, which further limits the development of cyber-ramping coordinated attack detection methods. To address these issues, this paper proposes a cyber-ramping coordinated attack identification method based on Tensor Decomposition-based Personalized Federated Learning (TDPFed).
Firstly, the attack pathways of cyber-ramping coordinated attacks are analyzed, and a corresponding attack model is constructed by integrating common cyber attack methods with the characteristics of PV ramping to elucidate the attack mechanisms. Secondly, tensor decomposition is employed to enhance traditional federated learning by compressing the communication parameters, effectively reducing communication cost while preserving nearly all information. Thirdly, a two-layer objective model is constructed, consisting of a tensorized local model and a personalized model. The personalized model is retained to enhance the identification performance on local data, while the tensorized local model is transmitted to alleviate communication load and accelerate training. Fourthly, training strategies for the personalized model and tensorized local model are formulated, and two different global aggregation strategies are introduced to improve upon the traditional federated averaging method. Finally, the characteristics of PV power output are analyzed, and a dataset generation method is proposed.
To validate the effectiveness of the proposed method, a power grid-communication network model based on the IEEE 33-bus system was constructed. Cyber-ramping coordinated attack identification was performed using traditional Light Gradient Boosting Machine (LightGBM), Federated Averaging (FedAvg), Personalized Federated Learning with Moreau Envelopes (pFedMe), an improved FedAvg method, and the proposed TDPFed method. In terms of macro-averaged precision, the traditional methods achieved 0.8889, 0.9288, 0.9637, and 0.8945, respectively. The TDPFed algorithm, utilizing the aggregating factor matrix strategy and the aggregating composed tensor strategy, achieved 0.9999 and 0.9995, demonstrating its superior capability in accurately identifying various cyber-ramping coordinated attacks. When the compression ratio increased from 1.5 to 2, the macro-averaged precision of the improved FedAvg method decreased by 1.53%, whereas the proposed method exhibited only minor declines of 0.02% and 0.01%. This result indicates that the proposed approach effectively reduces the required transmission parameters while preserving nearly all information. In terms of communication cost, the four traditional methods required computation times that were two orders of magnitude higher than that of the proposed method. Furthermore, under adversarial attacks, the proposed method demonstrated the highest robustness among all evaluated approaches.
The following conclusions can be drawn from the above results: (1) The use of a two-layer objective model, appropriate training strategies, and aggregation strategies during model training effectively improves the accuracy of cyber-ramping coordinated attack identification and accelerates model convergence. (2) Tensor decomposition can compresses the parameters that need to be uploaded, reducing the communication resource requirements of federated learning. This improvement addresses the issue of insufficient communication resources in large-scale multi-node power grid systems, significantly shortening the training time. (3) By generating a cyber-ramping coordinated attack dataset using baseline curves and photovoltaic random components from historical data, the issue of insufficient training data is mitigated. (4) Through case analysis, the proposed TDPFed method demonstrates a significant improvement in overall identification performance compared to traditional methods such as LightGBM, FedAvg, and pFedMe. It maintains a high identification accuracy even in specific scenarios where traditional methods struggle.

Key words： PV ramping cyber-ramping coordinated attack false data injection attack tensor decomposition federated learning

Received: 24 October 2024

PACS:	TM73
	TM769

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Cui Peiran
	Cui Mingjian
	Wang Qing
	Zhang Jian

Cite this article:

Cui Peiran,Cui Mingjian,Wang Qing等. Cyber-Ramping Coordinated Attack Identification Method for PV Using a Tensor Decomposition based Personalized Federated Learning[J]. Transactions of China Electrotechnical Society, 0, (): 250428-.

URL:

https://dgjsxb.ces-transaction.com/EN/10.19595/j.cnki.1000-6753.tces.240896 OR https://dgjsxb.ces-transaction.com/EN/Y0/V/I/250428

[1] 陈艳波, 刘宇翔, 田昊欣, 等. 基于广义目标级联法的多牵引变电站光伏-储能协同规划配置[J]. 电工技术学报, 2024, 39(15): 4599-4612.
Chen Yanbo, Liu Yuxiang, Tian Haoxin, et al.Collaborative planning and configuration of photovoltaic and energy storage in multiple traction substations based on generalized analytical target cascading method[J]. Transactions of China Electrotechnical Society, 2024, 39(15): 4599-4612.
[2] 陈明昊, 朱月瑶, 孙毅, 等. 计及高渗透率光伏消纳与深度强化学习的综合能源系统预测调控[J]. 电工技术学报, 2024, 39(19): 6054-6071, 6103.
Chen Minghao, Zhu Yueyao, Sun Yi, et al.The predictive-control optimization method for park integrated energy system considering the high penetration of photovoltaics and deep reinforcement learning[J]. Transactions of China Electrotechnical Society, 2024, 39(19): 6054-6071, 6103.
[3] 孔令国, 王嘉祺, 韩子娇, 等. 基于权重调节模型预测控制的风-光-储-氢耦合系统在线功率调控[J]. 电工技术学报, 2023, 38(15): 4192-4207.
Kong Lingguo, Wang Jiaqi, Han Zijiao, et al.On-line power regulation of wind-photovoltaic-storage-hydrogen coupling system based on weight adjustment model predictive control[J]. Transactions of China Electrotechnical Society, 2023, 38(15): 4192-4207.
[4] 李俊涛, 贾科, 董学正, 等. 网侧故障下光伏直流并网系统不平衡功率快速平抑方法[J]. 电工技术学报, 2024, 39(5): 1340-1351.
Li Juntao, Jia Ke, Dong Xuezheng, et al.A fast method for suppressing unbalanced power in photovoltaic DC grid-connected system under grid-side faults[J]. Transactions of China Electrotechnical Society, 2024, 39(5): 1340-1351.
[5] 张波, 高远, 李铁成, 等. 计及IGBT结温约束的光伏高渗透配电网无功电压优化控制策略[J]. 电工技术学报, 2024, 39(5): 1313-1326.
Zhang Bo, Gao Yuan, Li Tiecheng, et al.Reactive voltage optimization control strategy for high penetration photovoltaic distribution network considering IGBT junction temperature constraint[J]. Transactions of China Electrotechnical Society, 2024, 39(5): 1313-1326.
[6] 戚永志, 刘玉田. 风光储联合系统输出功率滚动优化与实时控制[J]. 电工技术学报, 2014, 29(8): 265-273.
Qi Yongzhi, Liu Yutian.Output power rolling optimization and real-time control in wind-photovoltaic-storage hybrid system[J]. Transactions of China Electrotechnical Society, 2014, 29(8): 265-273.
[7] 乐健, 郎红科, 谭甜源, 等. 新型配电系统分布式经济调度信息安全问题研究综述[J]. 电力系统自动化, 2024, 48(12): 177-191.
Le Jian, Lang Hongke, Tan Tianyuan, et al.Review of research on information security problems in distributed economic dispatch for new distribution system[J]. Automation of Electric Power Systems, 2024, 48(12): 177-191.
[8] 韩晓, 王涛, 韦晓广, 等. 考虑阵列间时空相关性的超短期光伏出力预测[J]. 电力系统保护与控制, 2024, 52(14): 82-94.
Han Xiao, Wang Tao, Wei Xiaoguang, et al.Ultrashort-term photovoltaic output forecasting considering spatiotemporal correlation between arrays[J]. Power System Protection and Control, 2024, 52(14): 82-94.
[9] Cui Mingjian, Wang Jianhui.Deeply hidden moving-target-defense for cybersecure unbalanced distribution systems considering voltage stability[J]. IEEE Transactions on Power Systems, 2021, 36(3): 1961-1972.
[10] 郭庆来, 辛蜀骏, 孙宏斌, 等. 电力系统信息物理融合建模与综合安全评估: 驱动力与研究构想[J]. 中国电机工程学报, 2016, 36(6): 1481-1489.
Guo Qinglai, Xin Shujun, Sun Hongbin, et al.Power system cyber-physical modelling and security assessment: motivation and ideas[J]. Proceedings of the CSEE, 2016, 36(6): 1481-1489.
[11] Mohsenian-Rad A H, Leon-Garcia A. Distributed Internet-based load altering attacks against smart power grids[J]. IEEE Transactions on Smart Grid, 2011, 2(4): 667-674.
[12] 张程彬, 崔明建, 张梓枭, 等. 考虑攻击偏好的三相不平衡配电系统分布式FDIA检测[J]. 电力系统保护与控制, 2024, 52(24): 109-119.
Zhang Chengbin, Cui Mingjian, Zhang Zixiao, et al.Distributed FDIA detection for three-phase unbalanced distribution systems[J]. Power System Protection and Control, 2024, 52(24): 109-119.
[13] Živković N, Sarić A T.Detection of false data injection attacks using unscented Kalman filter[J]. Journal of Modern Power Systems and Clean Energy, 2018, 6(5): 847-859.
[14] Tian Jue, Tan Rui, Guan Xiaohong, et al.Enhanced hidden moving target defense in smart grids[J]. IEEE Transactions on Smart Grid, 2019, 10(2): 2208-2223.
[15] 刘鑫蕊, 常鹏, 孙秋野. 基于XGBoost和无迹卡尔曼滤波自适应混合预测的电网虚假数据注入攻击检测[J]. 中国电机工程学报, 2021, 41(16): 5462-5476.
Liu Xinrui, Chang Peng, Sun Qiuye.Grid false data injection attacks detection based on XGBoost and unscented Kalman filter adaptive hybrid prediction[J]. Proceedings of the CSEE, 2021, 41(16): 5462-5476.
[16] Mohammadpourfard M, Sami A, Weng Yang.Identification of false data injection attacks with considering the impact of wind generation and topology reconfigurations[J]. IEEE Transactions on Sustainable Energy, 2018, 9(3): 1349-1364.
[17] 李元诚, 曾婧. 基于改进卷积神经网络的电网假数据注入攻击检测方法[J]. 电力系统自动化, 2019, 43(20): 97-104.
Li Yuancheng, Zeng Jing.Detection method of false data injection attack on power grid based on improved convolutional neural network[J]. Automation of Electric Power Systems, 2019, 43(20): 97-104.
[18] Wang Huaizhi, Ruan Jiaqi, Wang Guibin, et al.Deep learning-based interval state estimation of AC smart grids against sparse cyber attacks[J]. IEEE Transactions on Industrial Informatics, 2018, 14(11): 4766-4778.
[19] 韩一宁, 崔明建, 罗光浩, 等. 基于自适应联邦学习的输配网动-静态综合状态估计方法研究[J/OL]. 中国电机工程学报, 1-15[2025-03-17]. http://kns.cnki.net/kcms/detail/11.2107.tm.20250124.1046.010.html.
Han Yining, Cui Mingjian, Luo Guanghao, et al. An integrated dynamic-static state estimation method for transmission and distribution networks based on self-adaptive federated learning[J/OL]. Proceedings of the CSEE, 1-15[2025-03-17]. http://kns.cnki.net/kcms/detail/11.2107.tm.20250124.1046.010.html.
[20] 王健宗, 孔令炜, 黄章成, 等. 联邦学习算法综述[J]. 大数据, 2020, 6(6): 64-82.
Wang Jianzong, Kong Lingwei, Huang Zhangcheng, et al.Research review of federated learning algorithms[J]. Big Data Research, 2020, 6(6): 64-82.
[21] Abdulrahman S, Tout H, Mourad A, et al.FedMCCS: multicriteria client selection model for optimal IoT federated learning[J]. IEEE Internet of Things Journal, 2021, 8(6): 4723-4735.
[22] Konečný J, Mcmahon H B, Yu F X, et al. Federated learning: strategies for improving communication efficiency[J]. arXiv preprint arXiv:1610. 05492, 2016.
[23] Sridhar S, Govindarasu M.Model-based attack detection and mitigation for automatic generation control[J]. IEEE Transactions on Smart Grid, 2014, 5(2): 580-591.
[24] Cui Mingjian, Wang Jianhui, Yue Meng.Machine learning-based anomaly detection for load forecasting under cyberattacks[J]. IEEE Transactions on Smart Grid, 2019, 10(5): 5724-5734.
[25] Yue Meng.An integrated anomaly detection method for load forecasting data under cyberattacks[C]//2017 IEEE Power & Energy Society General Meeting, Chicago, IL, USA, 2017: 1-5.
[26] 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[J]. 电力系统自动化, 2024, 48(2): 118-127.
Su Xiangjing, Deng Chao, Li Fengyong, et al.Interpretable detection method for false data injection attack on power grid based on multi-head graph attention network and time convolution network model[J]. Automation of Electric Power Systems, 2024, 48(2): 118-127.
[27] 杨玉泽, 刘文霞, 李承泽, 等. 面向电力SCADA系统的FDIA检测方法综述[J]. 中国电机工程学报, 2023, 43(22): 8602-8622.
Yang Yuze, Liu Wenxia, Li Chengze, et al.Review of FDIA detection methods for electric power SCADA system[J]. Proceedings of the CSEE, 2023, 43(22): 8602-8622.
[28] Reisizadeh A, Taheri H, Mokhtari A, et al.Robust and communication-efficient collaborative learning[J]. Advances in Neural Information Processing Systems, 2019, 32.
[29] Berahas A S, Bollapragada R, Keskar N S, et al.Balancing communication and computation in distributed optimization[J]. IEEE Transactions on Automatic Control, 2019, 64(8): 3141-3155.
[30] Lee K, Lam M, Pedarsani R, et al.Speeding up distributed machine learning using codes[J]. IEEE Transactions on Information Theory, 2018, 64(3): 1514-1529.
[31] 程礼临, 臧海祥, 卫志农, 等. 考虑多光谱卫星遥感的区域级超短期光伏功率预测[J]. 中国电机工程学报, 2022, 42(20): 7451-7465.
Cheng Lilin, Zang Haixiang, Wei Zhinong, et al.Ultra-short-term forecasting of regional photovoltaic power generation considering[J]. Proceedings of the CSEE, 2022, 42(20): 7451-7465.
[32] 夏泠风, 黎嘉明, 赵亮, 等. 考虑光伏电站时空相关性的光伏出力序列生成方法[J]. 中国电机工程学报, 2017, 37(7): 1982-1993.
Xia Lingfeng, Li Jiaming, Zhao Liang, et al.A PV power time series generating method considering temporal and spatial correlation characteristics[J]. Proceedings of the CSEE, 2017, 37(7): 1982-1993.
[33] Japkowicz N, Stephen S.The class imbalance problem: a systematic study1[J]. Intelligent Data Analysis, 2002, 6(5): 429-449.
[34] Kim M, Hwang K B.An empirical evaluation of sampling methods for the classification of imbalanced data[J]. PLoS One, 2022, 17(7): e0271260.
[35] Reisizadeh A, Mokhtari A, Hassani H, et al.Fedpaq: A communication-efficient federated learning method with periodic averaging and quantization[C]//International conference on artificial intelligence and statistics. Online, 2020: 2021-2031.
[36] 陈晓霖, 昝道广, 吴炳潮, 等. 面向纵向联邦学习的对抗样本生成算法[J]. 通信学报, 2023, 44(8): 1-13.
Chen Xiaolin, Zan Daoguang, Wu Bingchao, et al.Adversarial sample generation algorithm for vertical federated learning[J]. Journal on Communications, 2023, 44(8): 1-13.
[37] 黄冬梅, 丁仲辉, 胡安铎, 等. 低成本对抗性隐蔽虚假数据注入攻击及其检测方法[J]. 电网技术, 2023, 47(4): 1531-1540.
Huang Dongmei, Ding Zhonghui, Hu Anduo, et al.Low-cost adversarial stealthy false data injection attack and detection method[J]. Power System Technology, 2023, 47(4): 1531-1540.