Advanced

Flexible Implementation of Model Predictive Control Using Sub-Optimal Solutions

Henriksson, Dan LU and Åkesson, Johan LU (2004) In Report TFRT 7610.
Abstract
The on-line computational demands of model predictive control (MPC) often prevents its application to processes where fast sampling is necessary. This report presents a strategy for reducing the computational delay resulting from the on-line optimization inherent in many MPC formulations. Recent results have shown that feasibility, rather than optimality, is a prerequisite for stabilizing MPC algorithms, implying that premature termination of the optimization procedure may be valid, without compromising stability. The main result included in the report is a termination criterion for the on-line optimization algorithm giving rise to a sub-optimal, yet stabilizing, MPC algorithm. The termination criterion, based on an associated... (More)
The on-line computational demands of model predictive control (MPC) often prevents its application to processes where fast sampling is necessary. This report presents a strategy for reducing the computational delay resulting from the on-line optimization inherent in many MPC formulations. Recent results have shown that feasibility, rather than optimality, is a prerequisite for stabilizing MPC algorithms, implying that premature termination of the optimization procedure may be valid, without compromising stability. The main result included in the report is a termination criterion for the on-line optimization algorithm giving rise to a sub-optimal, yet stabilizing, MPC algorithm. The termination criterion, based on an associated delay-dependent cost index, quantifies the trade-off between successively improved control profiles resulting form the optimization algorithm and the potential performance degradation due to increasing computational delay. It is also shown how the cost index may be used in a dynamic scheduling application, where the processor time is shared between two MPC tasks executing on the same CPU. (Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Book/Report
publication status
published
subject
keywords
Feedback Scheduling, Model Predicive Control, Delay Compensation
in
Report TFRT
volume
7610
publisher
Department of Automatic Control, Lund Institute of Technology (LTH)
ISSN
0280-5316
language
English
LU publication?
yes
id
c9c414b6-a22d-479c-9027-d5f64b1120eb (old id 8602626)
date added to LUP
2016-02-15 18:36:13
date last changed
2016-04-16 04:09:08
@techreport{c9c414b6-a22d-479c-9027-d5f64b1120eb,
  abstract     = {The on-line computational demands of model predictive control (MPC) often prevents its application to processes where fast sampling is necessary. This report presents a strategy for reducing the computational delay resulting from the on-line optimization inherent in many MPC formulations. Recent results have shown that feasibility, rather than optimality, is a prerequisite for stabilizing MPC algorithms, implying that premature termination of the optimization procedure may be valid, without compromising stability. The main result included in the report is a termination criterion for the on-line optimization algorithm giving rise to a sub-optimal, yet stabilizing, MPC algorithm. The termination criterion, based on an associated delay-dependent cost index, quantifies the trade-off between successively improved control profiles resulting form the optimization algorithm and the potential performance degradation due to increasing computational delay. It is also shown how the cost index may be used in a dynamic scheduling application, where the processor time is shared between two MPC tasks executing on the same CPU.},
  author       = {Henriksson, Dan and Åkesson, Johan},
  institution  = {Department of Automatic Control, Lund Institute of Technology (LTH)},
  issn         = {0280-5316},
  keyword      = {Feedback Scheduling,Model Predicive Control,Delay Compensation},
  language     = {eng},
  series       = {Report TFRT},
  title        = {Flexible Implementation of Model Predictive Control Using Sub-Optimal Solutions},
  volume       = {7610},
  year         = {2004},
}