Identification for Robust Control: System Modeling for Synthesis of Control Laws
Final progress rept. 1 May 96-30 Apr 97
HUNGARIAN ACADEMY OF SCIENCES BUDAPESTHUNGARY COMPUTER AND AUTOMATION RESEARCH INST
Pagination or Media Count:
One of the major theoretical contribution of our project is a new two-degree of freedom controller design approach based on a generic optimal control scheme. This is a new structure how to design optimal pole placement controllers. The scheme named as a generic two-degree of freedom G2DF system is based on a special Keviczky-Banyasz, or shortly K-B parametrization. It was proved that the optimality of this scheme in H2 andor H sub infinity spaces can be reached by special selection of two serial filters obtained from the solution of low order Diophantine equations andor Navenlina-Pick approximation paradigm. A new controller refinement technique was introduced which allows to determine the reachable maximum bandwidth under an amplitude constraint for the control action by iteratively redesigning the applied reference model as a new step in the basic iterative scheme. It succeeded to derive a new uncertainty relationship limiting the product of control performance and robustness. In the generic scheme where the investigation was performed the control and identification errors are the same, so this inequality limits the product of the model accuracy and a robustness measure of the closed loop control system. The different separate phases of identification for and design of robust control can properly be handled in a new approach combining the classical minimum variance like control with the a concept of maximum variance input design for robust identification for control. This triple control approach gradually iteratively or recursively, depending on the applied scheme improves the frequency spectrum of an initial reference input signal excitation approaching and concentrating on the vital medium frequency domain around the cross-over frequency.