This paper addresses a resilient bipartite output consensus issue for high-order heterogeneous multi-agent systems (MASs) with Byzantine attacks. Output regulator equations as well observers are introduced for bipartite leader-following issue with heterogeneous dynamics. For the security concerns, a multidimensional-bipartite-absolute-mean-subsequence-reduced (MBA-MSR) algorithm is developed for each component of received information from neighbors. This algorithm sorts and trims a set of structures, which judge the absolute value of the difference between the output of the current follower observer and that of the signed neighbor observer. It is able to exclude bounded paralyzed signals as the graph is strongly robust. Based on the filtered information via the algorithm, a resilient adaptive observer is designed to for individual follower. A resilient control strategy is then proposed for the MAS to achieve bipartite output consensus under -local/total attack. For reduction of number of sort, a conditional MBA-MSR (CMBA-MSR) algorithm is also developed. Finally, simulation examples are given to illustrate the effectiveness of the theoretical results.

中文翻译：

---

拜占庭攻击下高阶异构多智能体系统的弹性二方共识


本文解决了具有拜占庭攻击的高阶异构多代理系统（MAS）的弹性二方输出共识问题。引入输出调节器方程和观测器来解决具有异质动力学的二分领导者跟随问题。出于安全考虑，针对从邻居接收到的信息的每个分量开发了多维二分绝对均值子序列缩减（MBA-MSR）算法。该算法对一组结构进行排序和修剪，判断当前跟随观察者的输出与有符号邻居观察者的输出之间的差异的绝对值。由于图形非常稳健，因此它能够排除有界瘫痪信号。基于通过算法过滤的信息，为个体跟随者设计了弹性自适应观察器。然后为 MAS 提出弹性控制策略，以在局部/总体攻击下实现双方输出共识。为了减少排序次数，还开发了条件MBA-MSR（CMBA-MSR）算法。最后给出仿真算例说明理论结果的有效性。