多智能体系统的事件驱动一致性控制与多Lagrangian系统的分布式协同

发布时间：2018-05-07 04:28

本文选题：多智能体系统 + 一致性控制　；参考：《北京理工大学》2015年博士论文

【摘要】：本文重点针对线性/非线性多智能体系统的一致性问题,开展了基于事件驱动和分布式协同的一致性控制算法研究。针对线性连续时间同构多智能体系统一致性控制中的连续通信问题,研究了基于事件触发预估状态反馈控制策略的分布式协同控制算法。提出了仅依赖于少数离散点通信的一致性控制律,使得智能体网络的状态趋于相同,且不发生Zeno现象,将传统的智能体间连续通信改善为某些离散时间点通信。提出了自触发预估状态反馈控制器,消除了智能体连续监测自身状态的限制。提出了基于观测器的事件触发预估输出反馈一致性控制器,解决了输出反馈情形下的一致性控制问题。针对线性离散时间同构多智能体系统一致性控制中的周期性通信问题,研究了事件触发预估状态反馈一致性控制算法。提出了仅依赖于少数采样点通信的一致性控制律,使得智能体网络的状态趋于相同,且不发生类Zeno现象,将传统智能体间的周期性通信改善为某些采样时间点通信。提出了自触发预估状态反馈控制器,消除了智能体周期性监测自身状态的限制。同样为状态不可测的情形,提出了基于观测器的事件触发预估输出反馈一致性控制器。研究了线性异构多智能体系统的事件驱动一致性控制算法。基于内参考模型和观测器的思想,提出了无外部干扰情形下的事件触发预估反馈一致性控制器,使得智能体的输出趋于相同。针对存在匹配外部干扰的情形,基于状态反馈提出一种切换控制律,使得智能体在抑制干扰的同时实现事件驱动输出一致,且不发生Zeno现象。研究了多Lagrangian系统的事件驱动一致性控制问题。基于辅助状态测量误差和触发函数,提出了事件驱动一致性控制律,使得多Lagrangian系统的广义位置能够趋于一致,广义速度趋于零,并且不会发生Zeno现象,很大程度上节省了通信资源。此外,对于常值参数未知的情况,研究了多机器人系统的事件驱动集结一致问题,使得仅通过有限离散点的通信,机器人系统的位置趋于一致,速度趋于零。研究了基于Lagrangian模型的多航天器分布式姿态跟踪控制问题。从航天器姿态系统出发,针对领导者的控制输入可能为非零,有界且对跟随者未知的情形,对每个跟随者提出分布式非连续自适应控制器,使得跟随者和领导者的姿态误差收敛到零。针对非连续控制器造成的抖振问题,基于边界层理论和σ修正提出了分布式连续自适应控制器。针对外部扰动和未建模动态问题,提出了分布式鲁棒姿态跟踪控制器,使得领导者航天器和跟随者航天器之间的组合误差渐近收敛到零。将航天器模型转化为广义的Lagrangian模型,研究了分布式自适应姿态跟踪控制问题。研究了多Lagrangian系统的分布式包含控制问题。基于分布式滑模观测器提出非完全和完全两种分布式自适应滑模控制算法,其中后者去除了需要每个跟随者知道领导者状态导数的上界和其他全局信息的限制,解决了无向图下考虑参数不确定性和外部干扰的具有多个领导者的多Lagrangian系统分布式包含控制问题。
[Abstract]:This paper focuses on the consistency problem of linear / nonlinear multi-agent systems, and develops a conformance control algorithm based on event driven and distributed collaboration. In order to solve the problem of continuous communication in the unified control of linear continuous time isomorphic multi intelligence system, the point based state feedback control strategy based on event triggering is studied. A conformance control law which only relies on a few discrete point communication is proposed, which makes the state of the intelligent body network tend to be the same, without Zeno phenomenon, and improves the traditional inter body continuous communication to some discrete time point communication. A self triggered pre estimation state feedback controller is proposed to eliminate the continuity of the agent. An observer based event trigger prediction output feedback conformance controller is proposed, which solves the conformance control problem in the output feedback situation. The state feedback consistency of event triggered predictive state feedback is studied for the periodic communication problem in the unified multiple intelligence system of linear discrete time isomorphic multi intelligence system. A conformance control law, which only relies on a few sampling points communication, makes the state of the intelligent network tend to be the same and does not have the class Zeno phenomenon. The periodic communication between the traditional agents is improved to some sampling time communication. A self triggering predicate state feedback controller is proposed to eliminate the periodic monitoring of the agent. The limit of self state is measured. Also, an observer based event trigger prediction output feedback conformance controller is proposed. The event driven conformance control algorithm for linear heterogeneous multi-agent systems is studied. Based on the idea of the internal reference model and observer, an event touch without external interference is proposed. The output of the feedback conformance controller makes the output of the agent tend to be the same. Based on state feedback, a switching control law is proposed based on state feedback, which makes the agent realize the event driven output consistent while the interference is suppressed, and does not occur Zeno image. The event driven consistency of the multi Lagrangian system is studied. The problem of sexual control. Based on the auxiliary state measurement error and the trigger function, the event driven conformance control law is proposed, which makes the generalized position of the multi Lagrangian system tend to be consistent, the generalized velocity tends to zero, and the Zeno phenomenon will not occur, and the communication resources are saved to a great extent. In addition, the case of unknown constant value parameters is studied. The event driven aggregation problem of multi robot system makes the position of the robot system tend to be consistent and the speed tends to zero through the communication of the finite discrete points. The distributed attitude tracking control problem of multi spacecraft based on Lagrangian model is studied. From the attitude system of the spacecraft, the control input for the leader may be not. A distributed non continuous adaptive controller is proposed for each follower to make the follower and leader's attitude error converge to zero. For the buffeting problem caused by the discontinuous controller, a distributed continuous adaptive controller is proposed based on the boundary layer theory and the sigma correction. The distributed robust attitude tracking controller is proposed, which makes the combination error between the leader spacecraft and the follower spacecraft converge to zero asymptotically. The spacecraft model is transformed into a generalized Lagrangian model, and the distributed adaptive attitude tracking control problem is studied. The distribution of the multi Lagrangian system is studied. It includes control problem. Based on distributed sliding mode observer, two distributed adaptive sliding mode control algorithms are proposed, in which the latter eliminates the limit of the upper bound and other global information that requires each follower to know the leader's state derivative, and solves the uncertainty of parameters and the external interference in the undirected graph. A leader's multi Lagrangian system contains distributed control problems.

【学位授予单位】：北京理工大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TP13;TP18

【共引文献】