Practical, industrial use of nonlinear MPC
Note: This workshop has been cancelled!
Duration:
Sunday June 5th 08.30-12:30, lunch from 12:30-13:30.
Abstract:
This half-day workshop gives an overview of the main aspects of practical implementations of Nonlinear Model Predictive Control. The organizers from Cybernetica have long experience from developing, implementing and maintaining these types of controllers worldwide, in different industries. Parts of the workshop will be demonstrations based on Cybernetica's tools for NMPC. However, the focus will be on methods and principles, and not on specific software.
The target audience will be anyone who are familiar with the main principles of model predictive (or model based) control, and would like to learn more about
- Properties of, and possibilities with, nonlinear model predictive control
- Business cases for different industries
- Practical approaches versus theory
Agenda:
- NMPC
- Criteria formulations and optimization
- Horizons, parametrizations and controller tuning
- Nonlinear mechanistic models
- Model formulations
- Robustness, numerical aspects and simulation speed
- Substituting models and constraints into the NMPC criteria
- Model verification and offline parameter fitting
- Design of recursive estimators (EKF, MHE)
- Process noise models
- Interactions between controller tuning and estimator tuning
- Industrial installations – some project examples
- Practical issues
- Data communication
- Operator interfaces
- Safety and fallback strategies
- Operation
- Critical project phases
- Studies – specifications – development – commissioning and acceptance tests
- Simulation examples
- Online estimation and control of an offshore plant
- Online estimation and control of a polymer plant
- Summary and discussions
Presenters:
Svein Olav Hauger, and tentative Tor Steinar Schei / Peter Singstad, Cybernetica AS
Gold sponsor: Gemini Center PROST – Advanced Process Control, at NTNU and SINTEF. | |
![]() |
![]() |