This paper proposes a multi-controller structure for a plant and develops associated stability results. The main theoretical result of the paper is a quite simple necessary and sufficient condition for the closed-loop stability of the scheme. The result also specializes as a multi-controller generalization of the single controller Youla-Kucera theory for the class of all stabilizing controllers for a plant, and to known results concerning simultaneous stabilization. The plants considered are recursively expressed in terms of a set of approximations of frequency-shaped plant-model errors. The controllers in successive loops can be designed based on increasing refinement in model approximation, perhaps via on-line identification.
An advantage of the proposed multi-controller strategy is that a number of various controller design techniques can be employed to achieve performance objectives which are not conveniently achievable by any one technique. A design example is included with first an LQG design applied for stabilization of a crude nominal model, and then an H∞ optimal design, and finally an H2 optimal design. The latter two designs are aimed at enhancing H∞H2 performances. There could even be a further adaptive loop for on-line performance enhancement.
[1]
C. Nett.
Algebraic aspects of linear control system stability
,
1985,
1985 24th IEEE Conference on Decision and Control.
[2]
John B. Moore,et al.
Indirect adaptive techniques for fixed controller performance enhancement
,
1989
.
[3]
B. Francis,et al.
A Course in H Control Theory
,
1987
.
[4]
A. Antoulas.
On recursiveness and related topics in linear systems
,
1986
.
[5]
Mathukumalli Vidyasagar,et al.
Reliable stabilization using a multi-controller configuration
,
1983,
The 22nd IEEE Conference on Decision and Control.