The state-dependent Riccati equation (SDRE) approach to nonlinear system stabilization relies on representing a nonlinear system's dynamics in a manner to resemble linear dynamics, but with state-dependent coefficient matrices that can then be inserted into state-dependent Riccati equations to generate a feedback law. Although stability of the resulting closed loop system need not be guaranteed a priori, simulation studies have shown that the method can often lead to suitable control laws. In this paper, we consider the non-uniqueness of such a representation. In particular, we show that if there exists any stabilizing feedback leading to a Lyapunov function with star-shaped level sets, then there always exists a representation of the dynamics such that the SDRE approach is stabilizing. The main tool in the proof is a novel application of the S-procedure for quadratic forms.
[1]
Mario Sznaier,et al.
On the recoverability of nonlinear state feedback laws by extended linearization control techniques
,
1999,
Proceedings of the 1999 American Control Conference (Cat. No. 99CH36251).
[2]
Kelly D. Hammett,et al.
Controllability Issues in Nonlinear State-Dependent Riccati Equation Control
,
1998
.
[3]
J. Cloutier.
State-dependent Riccati equation techniques: an overview
,
1997,
Proceedings of the 1997 American Control Conference (Cat. No.97CH36041).
[4]
Stephen P. Boyd,et al.
Linear Matrix Inequalities in Systems and Control Theory
,
1994
.
[5]
Wei-Min Lu,et al.
Nonlinear optimal control: alternatives to Hamilton-Jacobi equation
,
1996,
Proceedings of 35th IEEE Conference on Decision and Control.