Centralized and distributed moving horizon strategies for state estimation of networked control systems

This dissertation presents novel methods for the centralized and distributed state estimation of linear and nonlinear networked control systems. Thereby, the delays and losses caused by the packet-based data transmission between the individual elements of the control loop as well as the unsynchronized clocks of the event-based operating sensors are explicitly considered. To this end, the problem of reconstruction the full state is formulated as suitable optimization problems. Their solvability is analyzed within the observability framework and necessary conditions are derived. This forms the foundation for the presentation of a rigorous stability analysis. Moreover, several steps are presented for efficiently solving the optimization problems which enable a real-time implementation. The simulation and experimental test-rig results demonstrate the performance of the developed methods.