虚假数据注入攻击下多智能体系统动态事件触发双向编队

Dynamic event-triggered bipartite formation for multi-agent systems with false data injection attacks

中文关键词  双向编队  数据驱动控制  虚假数据注入攻击  动态事件触发

英文关键词  bipartite formation  data-driven control  false data injection attacks  dynamic event-triggered

中文摘要

      针对未知动力学模型的多输入多输出非线性离散时间多智能体系统的虚假数据注入攻击问题, 本文设计 了一种基于径向基函数神经网络的攻击识别策略, 并针对其通讯受限问题, 设计了一种动态事件触发控制策略. 首 先, 利用伪偏导技术, 在智能体的每个工作点上建立了一种关于被控系统输入输出数据的紧格式动态线性化数据模 型, 并给出了该模型相应参数的估计法则. 此外, 利用符号图论分析了多智能体系统的双向编队控制问题, 设计了一 种组合测量误差方程, 将双向编队控制问题转化为一致性控制问题, 并设计了一种动态事件触发的无模型自适应双 向编队控制算法. 最后, 给出了双向编队跟踪误差的收敛性证明, 并通过仿真实验验证了该算法的有效性.

英文摘要

      To address the issue of false data injection attacks in multi-input-multi-output nonlinear discrete-time multiagent systems (MASs) with unknown dynamics models, this research presents a radial basis function neural networkbased attack recognition scheme. Additionally, it proposes a dynamic event-triggered control strategy to address its communication-constrained problem. Firstly, a compact form dynamic linearization data model for the controlled system’s input-output data is established at each working point of the agent using pseudo-derivative technology, and the estimation rule of the corresponding parameters of the model is provided. In addition, the bipartite formation control problem of the MASs is analyzed using signed graph theory, a combined measurement error equation is proposed to transform the bipartite formation control problem into a consensus control problem, and a dynamic event-triggered model-free adaptive bipartite formation control algorithm is proposed. Finally, the convergence of the bipartite formation tracking error is demonstrated, and the effectiveness of the algorithm is verified by simulation experiments.