摘要
针对配电物联网配电终端安全监测数据多源化、高度异构的特征以及配电物联网管理平台安全信息易受网络攻击和威胁的问题,文章提出一种跨模态数据聚合算法。该算法利用自适应分布嵌入阶段多源异构数据的跨模态来解决源域和目标域之间的欠适配和不平等问题,其对目标域数据是否产生网络威胁的判断可以比一些经典分类算法的更准确;通过构建各类型安全监测数据域的自适应方法来对网络威胁做出更加全面的描述,有利于进一步挖掘出潜在的未知网络威胁。实验结果表明,本文所提算法对参数的选择不敏感,且可收敛;与多视角学习支持向量机(SVM2K)、非监督域自适应法及同时使用特权信息和非监督域自适应法的方法相比,其识别准确率提高了10%~20%。
Aiming at the multi-source and highly heterogeneous characteristics of safety monitoring data of power distribution internet of things distribution terminal and the problem of security information of power distribution internet of things management platform being attacked and threatened by the network,this paper proposes a cross-modal data aggregation algorithm.Through the cross-modality of multi-source heterogeneous data in the adaptive distribution embedding stage,the problem of under fitting and inequality between source domain and target domain can be solved.Compared with some classic classification algorithms,this method can make a more accurate judgment on network threat caused by the data of target domain.By constructing adaptive methods of various types of security monitoring data domains to make a more comprehensive description of network threats,it is beneficial to further dig out potential unknown network threats.Experimental analysis results show that the algorithm proposed in this paper is insensitive to parameter selection,convergent,and its recognition accuracy is increased by 10%to 20%compared with the ones of SVM2K,adaptive method with non-monitoring data domain and the method with special authority information and non-monitoring data domain.
作者
陈娜
刘海姣
李今宋
邓清飞
田昕泽
张腾飞
CHEN Na;LIU Haijiao;LI Jinsong;DENG Qingfei;TIAN Xinze;ZHANG Tengfei(Nari Group Co.,Ltd.,Nanjing,Jiangsu 211106,China;College of Automation&College of Artificial Intelligence,Nanjing University of Posts and Telecommunications,Nanjing,Jiangsu 210023,China)
出处
《控制与信息技术》
2021年第4期1-7,共7页
CONTROL AND INFORMATION TECHNOLOGY
基金
贵阳供电局科技项目(060100KK52200003)。
关键词
网络安全
配电物联网
多源异构数据
安全监测数据
数据聚合
跨模态
network security
power distribution internet of things
multi-source heterogeneous data
security monitoring data
data aggregation
cross-modal
