Observer-based linear parameter-varying control with guaranteed L2-gain and H2-type performance objectives
Artikel i vetenskaplig tidskrift, 2014
Synthesis of an observer-based linear parameter-varying controller is considered for a general linear
parameter-varying plant. The parameter vector and its derivative are both assumed to take values in known
bounded domains, whereas only the parameter vector is assumed to be measurable during online operation.
The synthesis problem is considered for L2-gain and H2-type performance objectives. Potentially conservative
parameter-dependent linear matrix inequality conditions are derived for the solvability of these two
problems. To facilitate the reduction of conservatism, the conditions are expressed in a way to have bilinear
dependence on an arbitrary positive scalar. In addition to employing a suitable relaxation scheme to reduce
these conditions into finitely many constraints, a line search hence needs to be performed over the positive
scalar to obtain the best achievable performance with an observer-based controller. The online implementation
of the observer-based controller is relatively simpler if compared with a controller of unrestricted
structure. Moreover, the observer-based controller will have no dependence on the parameter derivatives
irrespective of the choices of the design variables.
L2-gain minimization
observer-based control
linear parameter-varying systems
convex optimization
H2 synthesis
linear matrix inequalities