This thesis deals with event-triggered model predictive control (MPC)
strategies for constrained networked and distributed control systems. A
networked control system usually consists of spatially distributed
sensors, actuators and controllers that communicate over a shared
communication network. Event-triggered control approaches consider the
network utilization in the controller design to provide a compromise
between control performance and communication effort. In this thesis a
holistic output-based MPC scheme for constrained linear systems with
event-triggered communication over the sensor-to-controller and
controller-to-actuator channels of a network is presented. The proposed
approach can be applied to centralized as well as decentralized setups
and handles bounded time-varying sampling intervals and transmission
delays for the control of constrained sampled-data systems. In
distributed control set-ups the overall plant is decomposed into
subsystems which are controlled by local controllers. Different
distributed model predictive control (DMPC) approaches with reduced
communication effort are presented in this thesis. The first approach is
non-iterative and uses event-triggered communication for the exchange of
state measurements. In the second approach, an event-triggered
cooperation strategy for DMPC based on distributed optimization is
introduced. Finally, an economic DMPC scheme for linear periodically
time-varying systems which is motivated by two real-world applications,
the control of a water distribution network and a medium voltage power
grid, is presented.
