Model predictive control for nonlinear networked control systems

Developments in communication technologies, and the diffusion of Internet, have brought to light a new control paradigm which goes commonly under the name of networked control systems (NCSs). While in standard control system and regulator are directly connected, in NCSs the control loop is closed over a digital network, e. g. LAN, Internet. This provides many advantages to traditional control, such as adding components on-the-fly, redundancy, fault-tolerance, and reduced invasivity. On the other hand, new challenges arise, since the presence of a communication network in the control loop introduces additional uncertainties that result, for example, from random delays, information losses, jitters, or quantization. This work presents a predictive control approach and a model-based compensator that can cope with several aspects of NCSs. These results are not limited to NCSs, but they can be extended to every system with input/ouput delays or information losses due, for example, to computation, information processing, slow sensor- and actuator-dynamics, or component failures. A distributed dissipativity-based solution for large-scale systems, e. g. networks of power-plants, is also introduced.