摘要
针对线性时不变系统在额外数据包注入攻击下的安全状态估计问题(受损传感器数量不限),提出了一种基于高斯混合模型的测量值估计算法。首先,采用期望最大化(EM)算法对受损测量值进行聚类和融合,并引入误差补偿器修正聚类过程中的误差,从而得到测量估计值。其次,设计了一种带补偿器的状态估计方法,有效提高了系统估计精度,使系统性能受损坏传感器数量的影响较小,在损坏传感器数量超过总数的一半时仍保持有效。仿真结果表明,当被攻击传感器的数量为3、7和10时,相比于直接丢弃被攻击传感器的测量值,所提方法的状态估计精度分别提升58%、88%和97%。
For the secure state estimation problem of linear time-invariant systems under extra packet injection attacks(without restricting the number of attacked sensors),a Gaussian mixture model-based measurement estimation algorithm is proposed.Firstly,the Expectation-Maximization(EM)algorithm is adopted to cluster and fuse the compromised measurements,and an error compensator is introduced to rectify errors generated during the clustering process,thereby obtaining the estimated measurements.Secondly,a state estimation method incorporating the compensator is designed,which effectively improved the estimation accuracy of the system.The system performance is less affected by the number of attacked sensors and remains effective even when the number of attacked sensors exceeds half of the total.The simulation results show that compared to directly discarding the measurements from the attacked sensors,the estimation accuracy of the proposed method is increased by 58%,88%and 97%when the number of attacked sensors is 3,7 and 10,respectively.
作者
仇海涛
王子乐
朱翠
QIU Haitao;WANG Zile;ZHU Cui(Beijing Key Laboratory of High Dynamic Navigation Technology,Beijing Information Science&Technology University,Beijing 102206,China;China United Network Communications Co.,LTD.Shaanxi Branch,Shaanxi 710075,China;School of Information and Communication Engineering,Beijing Information Science&Technology University,Beijing 102206,China)
出处
《中国惯性技术学报》
北大核心
2025年第7期736-742,共7页
Journal of Chinese Inertial Technology
基金
北京市自然科学基金(4222052)
北京信息科技大学“勤信人才”培育计划项目(QXTCP C202110)
北京信息科技大学促进高校分类发展重点研究培育项目(2121YJPY221)。
