Smooth Switching LPV Dynamic Output-feedback Control

Abstract

In this paper, we present an innovative design of smooth-switching LPV (Linear Parameter-Varying) dynamic output-feedback (DOF) controllers. For a given partition of scheduling parameter region, a family of LPV controllers are designed simultaneously with guaranteed system performance on each subregion and switching smoothness between adjacent subregions. The proposed control design, called smooth-switching mixed Input Covariance Constraint (ICC) and $${{\cal H}_\infty}$$ LPV control design, minimizes a combined cost of system output $${{\cal H}_2}$$ performance and smooth-switching index subject to $${{\cal H}_2}$$ constraints on control inputs and $${{\cal H}_\infty}$$ performance constraint. These stability and performance criteria are then formulated into a set of Parametric Linear Matrix Inequalities (PLMIs). In addition, a tunable coefficient is introduced in cost function to provide an optimal trade-off between system $${{\cal H}_2}$$ performance and switching smoothness, and therefore, the corresponding optimal LPV controllers can be derived iteratively by convex optimization. For illustration, an active magnetic bearing (AMB) example is used to show the effectiveness of the proposed simultaneous design approach by demonstrating significantly improved switching smoothness and an optimal trade-off between achievable output performance and switching smoothness.