Model Predictive Control deconstructed

Prof. Robert Bitmead (University of California)

SYSTEMS AND CONTROL SERIES

DATE: 2012-11-13
TIME: 09:00:00 - 10:30:00
LOCATION: Design office DO211 LT North UNSW@ADFA
CONTACT: JavaScript must be enabled to display this email address.

ABSTRACT:
Model Predictive Control or MPC is an extraordinarily widespread applied control law based on a constrained finite-horizon optimal control law being applied in a receding-horizon fashion, somewhat similarly to how a monthly chequing account might be managed.

The emphasis is the augmentation of the optimal control by operational constraints on inputs, outputs, and states. The ability to include constraints has led to this wide application of MPC and also to a focus on recursive feasibility of the optimization. Here we shall simplify the study of MPC by reducing it to its component parts and then studying what each brings to the control picture, notably in terms of the performance of the MPC control law in the rejection of stochastic disturbances, its primary application.
BIO:
Robert Bitmead is a Professor of Department of Mechanical & Aerospace Engineering University of California, San Diego and Cymer Corporation