15th Triennial World Congress of the International Federation of Automatic Control
  Barcelona, 21–26 July 2002 
LINEAR OUTPUT SELECTION FOR FEEDBACK LINEARIZATION
G. Szederkényi* K. M. Hangos* J. Bokor* T. Vámos**
* Systems and Control Laboratory, Computer and Automation
Research Institute,
H-1518 Budapest P.O. Box 63, Hungary
** AI Research Group, Computer and Automation Research
Institute, Hungary

The problem of selecting a linear output Kx for a SISO nonlinear dynamic system given in input-affine form is considered in this paper. In the general case, when the zero dynamics is hard to investigate analytically, K is suggested to be the resulting feedback gain of an LQR-design problem for the linearized model. The advantageous properties of the LQR design (gain and phase margins, etc.) known for linear systems enable to obtain a locally asymptotically stable, yet simple nonlinear controller if the linear output Kx is used for feedback linearization. With this output selection, the open loop system is of relative degree 1 at the desired operating point and it possesses at least locally asymptotically stable zero dynamics. It is shown on examples that the resulting closed loop nonlinear system can be stable in a wide neighborhood of the operating point. The concepts are illustrated on two characteristic nonlinear systems of two different application domain: the inverted pendulum and a continuous fermenter.
Keywords: nonlinear systems, output variables, feedback linearization, zero dynamics
Session slot T-Th-M07: Linearization of Nonlinear Systems/Area code 2c : Non-linear Systems