Model Predictive Control with Applications
Jay H. Lee, Georgia Institute of TechnologyErik Ydstie, Carnegie Mellon University Joseph Lu, Honeywell Process Solutions
This one-day short-course is intended to give an introduction to the most widely applied model-based control techniques the process industries. The course describes a new model identification technique for large-scale systems and the use of the model in predictive control algorithms developed for linear (and nonlinear) continuous (and batch) processes. Certain exemplary commercial applications are described in some detail, to provide guidance on how to address the issues that commonly arise during controller design for industrial processes, including model identification, model uncertainties, and constraints on the actuators and states. This course is suitable for practicing engineers, students, instructors, and researchers interested in control engineering practice.
The workshop begins with an overview on model predictive control (MPC) describing the basic algorithms for linear and nonlinear model predictive control for continuous processes that are applied to industrial processes, including the coupling of state estimation with state feedback to derive output feedback controllers. This will be followed by an introduction of a model identification and adaptation technique suited for large-scale systems. Successful applications on power systems and large-scale refinery/chemical systems will be presented. The final section of the workshop discusses some commercial application cases highlighting the issues and benefits in implementing the MPC technology on real processes.
The presenters have extensive experience working with and in industry. Jay Lee held a visiting position at DuPont where he worked on the design and implementation of control algorithms to processes at DuPont, before joining the faculty at the Georgia Institute of Technology and now Korea Advanced Institute of Science and Technology. His main research foci have been in model identification, model predictive control, and approximate dynamic programming. Erik Ydstie worked in the area of adaptive control and machine learning for his and has continued research in the area for the pas 2 decades. He has a held short courses and graduate classes in adaptive control. He is leading a CMU spin-off company that commercializes adaptive control and identification for MPC. Joseph Lu has extensively designed and implemented model predictive control algorithms at Honeywell Process Solutions. The expertise of the three presenters in the area of advanced process control also has been recognized by numerous honors and awards.
- 1:00 pm - 1:45 pm: MPC Overview: Main Idea and History of Its Theoretical and Practical Developments - J. Lee
- 1:45 pm - 2:15 pm: Model Identification for Large Scale Systems - E. Ydstie
- 1:15 pm - 3:00 pm: Industrial Practice of Model Predictive Control - J. Lu
- Conference Chairs
- Christos Maravelias
- University of Wisconsin - Madison
- John Wassick
- The Dow Chemical Company
- Erik Ydstie
- Carnegie Mellon University
- Larry Megan
- Praxair Inc.
- CACHE Liaison
- B. Wayne Bequette
- Rensselaer Polytechnic Insitute
- Conference Information Center
- Robin Craven
- Conference Manager