Convergence and Stability of Iterative Controller Refinement in the Presence of Constraints

J. Kovács and U. Kortela (Finland)


Identification, input non-linearity, control, constraint.


In iterative controller refinement schemes, certain process parameters are identified and then used to update on-line the controller in closed-loop. The controller is usually tuned based on the certainty equivalence principle, therefore biased parameter estimation, for example due to input constraint, may lead to control performance degradation. The paper investigates the convergence of the identification in constrained control case, where both level and rate constraints occur. It is shown that improvement in convergence and performance stability can be achieved by applying simple anti-windup compensation. The stated constrained control problem is investigated in a certain two-degree-of-freedom structure developed for iterative controller refinement to eliminate conflict between the convergence of the identification and the control errors.

Important Links:

Go Back