Mixed sensitivity control: a non-iterative approach

Abstract

Recent analytical solutions to Mixed Sensitivity Control (MSC) are developed and compared with standard MSC based on γ-iteration. The proposed MSC solution gives conditions for strong stability and overcomes the pole-zero cancellations between the plant and the controller of non-iterative solutions, keeping the low-computational effort advantage of non-iterative solutions. The proposed MSC is based on the minimization of the most common closed-loop sensitivity functions in low-frequencies and the free-parameters of the stabilizing-controllers solve an algebraic equation of restriction that assigns the same value to the infinity-norms of the sensitivity functions at low and high-frequencies, guaranteeing robust stability and robust performance. It is assumed that the plant state dimension is double the plant input dimension and that the linear time-invariant nominal plant has a stabilizable and detectable realization and is strongly stabilizable. This MSC problem is solved in a one-parameter observer-controller configuration and reference tracking-control of positions is realized on a two-degrees of freedom feedback-configuration. An approximated optimal value of the location of the closed-loop poles is proposed based on Glover and McFarlane's optimal stability margin [(1989)] which in turn is based on Nehari's Theorem. Simulations of a mechanical system illustrate the results.

Keywords

• robust control
• robust stability
• robust performance
• mixed sensitivity
• stabilizing controllers
• unstructured uncertainty
• sub-optimal control
• γ-iteration