Block structured based model reduction
|Project members:||O. Naeem, MSc. (Omar), ir. A.E.M. Huesman (Adrie)|
|Keywords:||Process technology, High purity distillation columns|
|Sponsored by:||Eurpean Commission, ProMATCH, Delft University of Technology|
Computational effort (simulation time) has been one of the concerns of modern systems and control research. Large scale industrial process models require lot of computational effort, which is vital, if the model has to be used for closed-loop control and optimization purposes. Model reduction has been considered as one of the method to achieve acceptable computational effort. There are different perspectives of model reduction for example, linear system theory, projection based model reduction, time scale based model reduction, identification based etc.
In this research, the focus has been on identification based model reduction. Models with Hammerstein structure have been used to identify and yield a reduced model. Hammerstein model consists of a static non-linear block, followed by linear dynamic block.
It is expected, that this methodology and structure will help to achieve the goals. The methodology has been applied on a benchmark. Satisfactory results have been accomplished as far as identification is concerned. Computational effort has to be investigated for more complex benchmarks in future (for example; high purity distillation column).