qLPV modeling and mixed-sensitivity L 2 control for a magnetic levitation system

ABSTRACT This paper proposes a comprehensive mixed-sensitivity control design for an experimental magnetic levitation (Maglev) system. The control strategy can be seen as an extension of the loop-shaping procedure for discrete-time linear parameter-varying (LPV) systems using linear-fractional representation (LFR). By making use of an efficient quadratic approach given in the form of linear matrix inequalities (LMIs), a functional and computationally attractive gain-scheduling technique is achieved. Despite the rigorous mathematical considerations to obtain the controller, the guidelines to its practical implementation are presented as a straightforward method using LMIs. A detailed modeling of the Maglev plant manufactured by Quanser is carried out to illustrate the procedure, including a description of the nonlinear equations embedding process to obtain a discretized quasi-LPV (qLPV) model. Experimental results demonstrate the effectiveness of the proposed control design.