Call for participation
This is a one-day interactive course on model-based observer design. The aim is to determine which states and parameters of an observer model can be estimated with reasonable accuracy from available measurements. This can be established using observability theory. Once observability is established, development of Kalman filters will be discussed to estimate the unknown states and parameters. The grinding mill from the mineral processing industry will be used as the main illustrative example. At the end of the event, attendees will be equipped to assess the observability of the states and parameters of a process model, and apply an Extended Kalman Filter to estimate the unknowns.
The course is presented on request by Dr Derik le Roux, a lecturer at the University of Pretoria.
Contact him at email@example.com for further information regarding the course.
The course is accredited for 1 CPD point: SAIEE-1813-V
More information on this event
Who should attend
The course is intended for practitioners and academics involved with model-based design and implementation of observers. The course will have a strong theoretical component, but with adequate examples and problems.
08:00 – 8:30 Registration (Tea/Coffee)
Session 1: Background
- Discretisation of Non-linear and Linear systems
- Eigenvalues and Eigenvectors Lie Derivatives
Session 2: Observability Theory (1)
- Non-linear Observability
10:00 –10:15 Tea Break
Session 3: Observability Theory (2)
- Linear Observability
- Kalman Decomposition
12:15–13:00 Lunch Break
Session 3: Kalman Filters (1)
- Extended Kalman Filter
- Unscented Kalman Filter
14:30 – 14:45 Tea Break
Session 4: Kalman Filters (2)
16:30 Closing and Evaluation
SACAC members: R 2800
Non-members: R 4000
Bona Fide Students: R 1500
Fees include a manual, tea/coffee and a light lunch.