[1]李伟,霍雪松,张明,等.基于残差全连接神经网络的电力监控系统异常行为检测方法[J].东南大学学报(自然科学版),2020,50(6):1062-1068.[doi:10.3969/j.issn.1001-0505.2020.06.010]
　Li Wei,Huo Xuesong,Zhang Ming,et al.Abnormal behavior detection method for power monitoring system based on fully connected residual neural network[J].Journal of Southeast University (Natural Science Edition),2020,50(6):1062-1068.[doi:10.3969/j.issn.1001-0505.2020.06.010]
点击复制

基于残差全连接神经网络的电力监控系统异常行为检测方法()
分享到：

《东南大学学报(自然科学版)》[ISSN:1001-0505/CN:32-1178/N]

卷:
50
期数:
2020年第6期
页码:
1062-1068
栏目:
计算机科学与工程
出版日期:
2020-11-20

文章信息/Info

Title:
Abnormal behavior detection method for power monitoring system based on fully connected residual neural network
作者:
李伟1; 霍雪松2; 张明2; 朱红勤2
1 东南大学计算机科学与工程学院, 南京211189; 2 国网江苏省电力有限公司, 南京210000
Author(s):
Li Wei1;  Huo Xuesong2;  Zhang Ming2;  Zhu Hongqin2
1 School of Computer Science and Engineering, Southeast University, Nanjing 211189, China
2 State Grid Jiangsu Electric Power Co., Ltd., Nanjing 210000, China
关键词:
电力监控系统;  异常行为检测;  残差全连接神经网络;  集成学习
Keywords:
power monitoring system;  abnormal behavior detection;  fully connected residual neural network;  ensemble learning
分类号:
TP393
DOI:
10.3969/j.issn.1001-0505.2020.06.010
摘要:
为了提升电力监控系统异常行为检测能力,提出了一种基于残差全连接神经网络的电力监控系统异常行为检测方法.将深度学习模型与半监督学习方法相结合,构建了两级残差全连接神经网络,并将其作为核心分类模型.选取训练样本特征子空间、有标记训练样本子集、残差全连接层层数的多种不同组合,采用混合扰动的方法生成具有差异性的成员分类器.基于成员分类器的分类误差率,通过加权多数表决对无标记样本数据进行增量学习,生成分类识别能力较强的集成分类器.实验结果表明,在同等标记数据规模下,所提方法的检测准确率和模型训练收敛速度均优于现有方法,可快速、准确识别电力监控系统异常行为,同时降低了对训练样本数据进行标记的开销.
Abstract:
To improve the ability of abnormal behavior detection of the power monitoring system, an abnormal behavior detection method for the power monitoring system based on fully connected residual neural network was proposed. By combining the deep learning model with the semi-supervised learning method, a two-level fully connected residual neural network was constructed as a core classification model. A variety of different combinations of the training sample feature subspace, the labeled training sample subset and the number of the fully connection residual layers were selected to generate different member classifiers by using the hybrid disturbance method. Based on the classification error rate of member classifiers, the weighted majority voting method was applied to incremental learn the unlabeled sample data to generate an ensemble classifier with strong classification and recognition ability. The experimental results show that under the same label data scale, the detection accuracy and the model training convergence speed of the proposed method are better than those of the existing methods. It can detect the abnormal behaviors of the power monitoring system quickly and accurately, and reduce the cost of marking the training sample data.

参考文献/References:

[1] 田伟宏. 智能变电站网络异常检测方法的研究与实现[D]. 沈阳:中国科学院沈阳计算技术研究所,2020.
　　Tian H W. Research and implementation of network anomaly detection method for intelligent substation [D]. Shenyang: Shenyang Institute of Computing Technology, Chinese Academy of Sciences, 2020.(in Chinese)
[2] Zhu M Y, Ye K J, Xu C Z. Network anomaly detection and identification based on deep learning methods [C]// Proceedings of 2018 International Conference on Cloud Computing. Seattle, WA, USA, 2018: 219-234. DOI:10.1007/978-3-319-94295-7_15.
[3] Buczak A L, Guven E. A survey of data mining and machine learning methods for cyber security intrusion detection[J]. IEEE Communications Surveys and Tutorials, 2016, 18(2): 1153-1176. DOI:10.1109/comst.2015.2494502.
[4] Sharma S,Diarra A, Alvares F, et al. KDetect: unsupervised anomaly detection for cloud systems based on time series clustering[C]//Proceedings of the 3rd International Workshop on Systems and Network Telemetry and Analytics. New York, USA: ACM, 2020: 3-10. DOI:10.1145/3391812.3396271.
[5] Mao J W,Hu Y Q, Jiang D, et al. CBFS: A clustering-based feature selection mechanism for network anomaly detection[J]. IEEE Access, 2020, 8: 116216-116225. DOI:10.1109/access.2020.3004699.
[6] Chen C M, Guan D J, Huang Y Z, et al. Anomaly network intrusion detection using hidden Markov model [J]. International Journal of Innovative Computing, Information and Control, 2016, 12(2):569-580.
[7] Wang D G, Dong J C, Huang L, et al. Anomaly behavior detection based on ensemble decision tree in power distribution network [C]// Proceeding of 4th Annual International Conference on Network and Information Systems for Computers. Wuhan, China, 2018: 312-316.
[8] 陆悠, 李伟, 罗军舟, 等. 一种基于选择性协同学习的网络用户异常行为检测方法[J]. 计算机学报, 2014, 37(1): 28-40.
　　Lu Y, Li W, Luo J Z, et al. A network user’s abnormal behavior detection approach based on selective collaborative learning [J].Chinese Journal of Computers, 2014, 37(1): 28-40.(in Chinese)
[9] Gao X W, Shan C, Hu C Z, et al. An adaptive ensemble machine learning model for intrusion detection[J]. IEEE Access, 2019, 7: 82512-82521. DOI:10.1109/access.2019.2923640.
[10] Khonde S R, Ulagamuthalvi V. Ensemble-based semi-supervised learning approach for a distributed intrusion detection system[J]. Journal of Cyber Security Technology, 2019, 3(3): 163-188. DOI:10.1080/23742917.2019.1623475.
[11] 黄立威,江碧涛,吕守业, 等. 基于深度学习的推荐系统研究综述[J]. 计算机学报, 2018, 41(7): 1619-1647.
　　Huang L W, Jiang B T,Lü S Y, et al. Survey on deep learning based recommender systems [J]. Chinese Journal of Computers, 2018, 41(7): 1619-1647.(in Chinese)
[12] Kwon D, Kim H, Kim J, et al. A survey of deep learning-based network anomaly detection[J].Cluster Computing, 2019, 22(S1): 949-961. DOI:10.1007/s10586-017-1117-8.
[13] He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition [C]// Proceedings of the 29th IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA, 2016: 770-778.
[14] Ioffe S. Batch renormalization: Towards reducing minibatch dependence in batch-normalized models [C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Long Beach, CA, USA, 2017: 1946-1954.
[15] Li M, Zhang T, Chen Y Q, et al. Efficient minibatch training for stochastic optimization [C]// Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York, USA, 2014: 661-670. DOI:10.1145/2623330.2623612.

备注/Memo

备注/Memo:
收稿日期: 2020-06-13.
作者简介: 李伟(1978—),男,博士,副教授,xchlw@seu.edu.cn.
基金项目: 国家重点研发计划资助项目(2017YFB1003000)、国网江苏省电力有限公司科技资助项目(J2018039).
引用本文: 李伟,霍雪松,张明,等.基于残差全连接神经网络的电力监控系统异常行为检测方法[J].东南大学学报(自然科学版),2020,50(6):1062-1068. DOI:10.3969/j.issn.1001-0505.2020.06.010.
更新日期/Last Update: 2020-11-20