Adaptive Chain Observers for Controlling Cooperating Clusters of Nonlinear Delayed-Output Mechanical Systems

摘要：本文将基础研究与硬件在环实验相结合，生成并验证了不确定机械系统合作集群的新控制方法。本研究的独特之处在于我们明确考虑了实际低带宽车对车通信网络中不可或缺的时延特性。研究计划的理论部分侧重于非线性动力系统的合作集群，这些系统是通过领导-追随者层次结构来指挥的。前导控制器通过跟踪由非线性参考模型生成的参考轨迹来实现舰队的绝对定位和定向。跟随器控制器依赖于从先导器接收到的延时参考轨迹信号，以确保维持所需的“地层几何形状”。随动器各自模型中的参数不确定性要求设计利用当前可用的局部状态信息的自适应控制器。考虑到通信延迟的影响，我们提出了一种创新的链观测器结构，其目标是在闭环意义上“估计”领导者非线性参考模型的无延迟瞬时状态。这种链式结构将通过新的自适应控制解决方案得到进一步加强，这些解决方案将来自于作者最近给出的非线性再参数化技术。包括链观测器在内的自适应控制框架避免了过度参数化，并提供了某些独特的流形吸引特性，显著提高了闭环性能。该提案的实验部分将在戈达德太空飞行中心的分布式航天器编队飞行试验台设施中实施和评估所有理论。

Adaptive Chain Observers for Controlling Cooperating Clusters of Nonlinear Delayed-Output Mechanical SystemsAbstract:This proposal combines basic research with hardware-in-the-loop experimentation to generate and validate new control methodologies for cooperating clusters of uncertain mechanical systems. Uniqueness of this study arises from our explicit consideration of time-delay characteristics that are integral within realistic low-bandwidth vehicle-to-vehicle communication networks. Theoretical part of the research plan focuses upon cooperating clusters of nonlinear dynamical systems that are commanded through a leader-follower hierarchy. The leader controller enforces absolute positioning and orientation of the fleet by tracking reference trajectories generated by a nonlinear reference model. The follower controllers rely upon time-delayed reference trajectory signals received from the leader to ensure that the desired "formation geometry" is maintained. Parameter uncertainty within the respective models of the followers necessitates the design of adaptive controllers that utilize currently available local state information. Duly taking the effects of communication delay into consideration, we propose the development of an innovative chain-observer structure onboard each follower whose objective would be to "estimate" in a closed-loop sense the undelayed instantaneous state of the nonlinear reference model of the leader. This chain structure would be further strengthened by new classes of adaptive control solutions that will be derived from nonlinear re-parameterization techniques recently given by the proposer. The adaptive control framework including the chain-observer avoids over-parameterization and provides certain unique manifold attractivity properties that significantly boost the closed-loop performance. The experimental component of this proposal will implement and evaluate all the theory within the distributed spacecraft formation flying testbed facility at the Goddard Space Flight Center.