IEEE Trans Cybern. 2018 Feb;48(2):675-688. doi: 10.1109/TCYB.2017.2651026. Epub 2017 Jan 23.
This paper focuses on the average consensus of double-integrator networked systems based on the asynchronous periodic edge-event triggered control. The asynchronous property lies in the edge event-detecting procedure. For different edges, their event detections are performed at different times and the corresponding events occur independently of each other. When an event is activated, the two adjacent agents connected by the corresponding link sample their relative state information and update their controllers. The application of incidence matrix facilitates the transformation of control objects from the agent-based to the edge-based. Practically, due to the constraints of network bandwidth and communication distance, agents usually cannot receive the instantaneous information of some others, which has an impact on the system performance. Hence, it is necessary to investigate the presence of communication time delays. For double-integrator multiagent systems with and without communication time delays, the average state consensus can be asynchronously achieved by designing appropriate parameters under the proposed event-detecting rules. The presented results specify the relationship among the maximum allowable time delays, interaction topologies, and event-detecting periods. Furthermore, the proposed protocols have the advantages of reduced communication costs and controller-updating costs. Simulation examples are given to illustrate the proposed theoretical results.
本文主要研究基于异步周期性边事件触发控制的双积分网络系统的平均一致性。异步特性在于边事件检测过程。对于不同的边，它们的事件检测在不同的时间进行，相应的事件彼此独立发生。当事件被激活时，通过相应链路连接的两个相邻代理会采集它们的相对状态信息并更新它们的控制器。关联矩阵的应用促进了控制对象从基于代理到基于边的转换。实际上，由于网络带宽和通信距离的限制，代理通常无法接收其他代理的瞬时信息，这会影响系统性能。因此，需要研究通信时间延迟的存在。对于具有和不具有通信时间延迟的双积分多智能体系统，通过设计适当的参数，可以在所提出的事件检测规则下实现异步平均状态一致性。所提出的结果指定了最大允许时间延迟、交互拓扑和事件检测周期之间的关系。此外，所提出的协议具有降低通信成本和控制器更新成本的优点。给出了仿真示例来说明所提出的理论结果。
