Networked control of distributed-parameter systems
Final Report Abstract
The completed research project concerns itself with the development of new methods for networkbased control of multi-agent systems, in which the agents have distributed-parameter dynamics. The solution of this problem is based on the generalization of the cooperative output regulation theory for lumped-parameter agents to agents with distributed-parameter dynamics, which includes a multitude of cooperative control tasks. During the project new systematic cooperative regulator design methods for networks of hyperbolic agents with unknown leader and disturbance models were developed based on previous results, which all assume known dynamics of the leader agents and the disturbance models. This allows a fully distributed solution of leader-follower problems as well as the adaptation to changing disturbances by communicating updated disturbance models. For heterogeneous multi-agent systems the design of adaptive reference observers for ODE agents was generalized to hyperbolic agents. In addition, not only the reference model, but also the disturbance models were synchronized as well, which has not yet been considered in the literature both for ODE and PDE agents so far. New adaptive design methods for cooperative disturbance observers with hyperbolic PDE-ODE dynamics were developed to solve this problem, which reconstruct the agent and disturbance model states by means of a network communication. The associated feedforward controller is designed by solving the so-called regulator equations online using a novel sample-and-hold approach. This results in a systematic regulator design method for heterogeneous multi-agent systems with hyperbolic agent dynamics and unknown leader and disturbance models. This method, however, is not robustness w.r.t. model uncertainty. To achieve the latter, the cooperative internal model principle was generalized to unknown leader and disturbance models. Therein, the internal model is continuously updated by a cooperative signal model observer using a network communication. The associated networked controller design requires the challenging simultaneous stabilization of the hyperbolic agents extended by the internal model, for which new methods were developed. With the rigorous proof that the robust cooperative output regulation problem is solved even in the presence of non-destabilizing model perturbations, this results in a new general design methodology for robust leader-follower synchronization with unknown leader and disturbance models for distributed-parametric MAS. Thus, the cooperative output regulation theory was broadly extended to multi-agent systems with hyperbolic agents in this project, which can also be easily applied to hyperbolic PDE-ODE agents and also offers starting points for parabolic agents. In addition, robust synchronization for ODE agents with unknown signal models was also considered, because previous nonlinear approaches using adaptive control are only feasible under restrictive assumptions. The newly developed networked control methods were successfully validated experimentally for ODE agents using a test rig for mobile robots.
Publications
-
Tagung des GMA-Fachausschusses 1.40, 19.09.2022, Anif, T. Enderes: Synchronisierung von Netzwerken hyperbolischer Systeme mit unbekanntem Leader und Störmodell
T. Enderes
-
Cooperative output regulation for networks of hyperbolic systems using adaptive cooperative observers. Automatica, 162, 111506.
Enderes, Tarik; Gabriel, Jakob & Deutscher, Joachim
-
Cooperative Robust Output Regulation for Networks of Hyperbolic Systems With Unknown Signal Models. International Journal of Robust and Nonlinear Control, 35(4), 1355-1367.
Enderes, Tarik & Deutscher, Joachim
-
Tagung des VDI/VDE-GMA-Fachausschusses 2.15, 21.02.2024, Günzburg, T. Enderes: Robuste Synchronisierung von Netzwerken hyperbolischer Systeme mit unbekanntem Leader und Störmodell
T. Enderes
-
Robust cooperative output regulation for linear multi-agent systems with unknown leader and disturbances. Systems & Control Letters, 203, 106119.
Enderes, Tarik & Deutscher, Joachim
