【作者】 黄一鸣； 樊慧津； 刘磊； 王博；

【Author】 Yiming Huang;Huijin Fan;Lei Liu;Bo Wang;Huazhong University of Science and Technology;

【机构】 华中科技大学；

【摘要】 在非安全环境中,多智能体系统可能遭受如病毒注入、信道干扰等恶意攻击而导致某些个体失控,安全一致性控制的目标是确保系统中的正常个体免于受失控智能体污染信息的影响,从而达到状态一致。本文考虑到多智能体个体在不同类型恶意攻击下表现出的差异性,针对多类恶意攻击下包含多类异质节点的多智能体系统提出了相应的网络鲁棒性定义,并通过设计分布式信息处理方案建立安全控制算法,给出了该算法下系统实现安全一致性的充要条件。最后,通过数值仿真验证了结论的有效性。更多还原

【Abstract】 In the non-secure environment, some individuals of multi-agent system can spin out of control due to malicious attacks such as virus injection and channel interference. The goal of resilient consensus of multi-agent system is to ensure the normal individuals of system free from the influence of the malicious agents’ polluted information and achieve consensus. In this paper, we consider the individuals of multi-agent system show the differences in weaknesses under malicious attacks. Based on the multi-agent system with multi-type heterogeneous nodes under multi-class malicious attack, the corresponding definition of robustness is proposed, and we establish a resilient control algorithm by designing distributed information processing scheme. With the algorithm above, we provide the necessary and sufficient condition to achieve resilient consensus of heterogeneous network. Finally, several numerical examples are provided to validate our results.更多还原