Stochastic MPC with applications to process control

Abstract

This paper presents a model predictive control formulation for Networked Control Systems subject to independent and identically distributed delays and packet dropouts. The design takes into account the presence of a communication network in the control loop, resorting to a buffer at the actuator side to store and consistently apply delayed control sequences when fresh control inputs are not available. The proposed approach uses a statistical description of transmissions to optimise the expected future control performance conditioned upon the current system state, previously calculated control packets and transmission acknowledgements. Experimental studies using a quadruple tank process illustrate the applicability of the method to process control.

Es la versión aceptada del artículo. Se puede consultar la versión final en: https://doi.org/10.1080/00207179.2014.975845