Robust event-triggered networked MPC with active set updates

摘要

We extend a recently proposed robust event-triggered MPC approach that exploits the piecewise affine structure of the solution. Robust optimal feedback laws and their regions of validity are calculated on a central compute node on demand. These regional laws are used by one or more local nodes in local feedback loops. Whenever one of the local systems detects that the region of validity of the current optimal control law has been left, it requests a new law from the central node. The present paper uses a particular variant of updating the control laws in a way that reduces the number of requests to the central node.