引用本文:张龙杰,陈勇,刘越智,潘成伟.网络攻击下异构网联系统的分布式自适应动态规划控制[J].控制理论与应用,2025,42(4):669~678.[点击复制]
ZHANGLong-jie,CHEN Yong?,LIU Yue-zhi,PAN Cheng-wei.Distributed adaptive dynamic programming for heterogeneous interconnected systems under cyber attacks[J].Control Theory & Applications,2025,42(4):669~678.[点击复制]
网络攻击下异构网联系统的分布式自适应动态规划控制
Distributed adaptive dynamic programming for heterogeneous interconnected systems under cyber attacks
摘要点击 3  全文点击 1  投稿时间：2022-11-21  修订日期：2025-03-05
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2023.21026
  2025,42(4):669-678
中文关键词  最优控制  网络攻击  异构网联系统  分布式控制  自适应动态规划  有限时间评价–执行网络算法
英文关键词  optimal control  cyber-attacks  heterogeneous interconnected systems  distributed control  ADP  finite-time critic-actor algorithm
基金项目  国家重点研发计划项目(2022YFE0120700),国家自然科学基金项目(61973331,61973257,61903064),四川省科学与技术支持项目基金项目 (2021YFG0079, 2021YFG0080, 2021YFG0082)资助.
作者单位E-mail
张龙杰 电子科技大学自动化工程学院 lizhang uestc@163.com 
陈勇* 电子科技大学自动化工程学院 ychencd@uestc.edu.cn 
刘越智 电子科技大学自动化工程学院  
潘成伟 电子科技大学自动化工程学院  
中文摘要
      本文考虑了节点注入攻击下异构网联系统的安全状态估计与控制问题,通过设计一种基于分布式远程状 态安全估计器的有限时间自适应动态规划控制策略,抑制节点注入攻击对分布式系统协同跟踪效果的影响,实现对 异构网联系统的安全控制.首先,为了实现对节点注入攻击下异构网联系统状态信息的重塑,融合最优攻击补偿策 略设计,设计基于无迹卡尔曼滤波的分布式远程状态安全估计器;然后,融合远程状态估计器的安全优化目标和协 同优化目标,基于哈密尔顿方程的最优控制理论,提出分布式安全优化控制策略;在此基础上,基于有限时间优化理 论, 提出基于策略迭代算法的有限时间评价–执行网络权重更新算法,实现对最优控制策略和值函数的有限时间趋 近;最后,利用仿真研究和对比分析验证了所提控制策略的有效性.
英文摘要
      This article considers the secure state estimation and control for the heterogeneous interconnected systems under node-injected attacks, and a finite-time adaptive dynamic programming based on the distributed secure remote estima tion is designed to improve the security of the control systems. Firstly, to recover the state information of the heterogeneous interconnected systems under the node-injected attacks, the distributed remote secure estimator is designed based on the unscented Kalman filter and the optimal attack compensation strategy. Then, by combining the optimal objective of the secure estimation and the optimal objective of the consensus, the distributed secure optimal control strategy is presented based on the solution of the Hamilton equation. Furthermore, to approximate the optimal controller and value function in the finite-time, the finite-time tuning laws of the critic-actor network are proposed based on the policy iteration algorithm and finite-time optimization. Finally, the effectiveness of the proposed method is verified by the comparison result analysis.