Adaptive Inferential Control

The use of model based controllers such as Inferential (Internal Model) Control has been increasing in the process industries. Controller design is straightforward when the process model is an accurate representation of the process. Since the model is always only an approximation to the physical process, for stable, robust performance the controller must be tuned for the mismatch between the model and process. The adaptive tuning of inferential controllers has been investigated where the response speed of the controller is modified based on the observed closed loop performance. The asymptotic behavior of the adaptation algorithm together with a stability analysis of the closed loop system show that stable robust control performance will be obtained even with large modeling errors present. The specification and design of the adaptation algorithm is presented along with the application to several systems.