Multi-objective MIMO optimal control design without zero interpolation

MaRDI QIDQ4263644zbMATH OpenFDO

Authors Murti Salapaka, Mustafa Khammash

Publication date 21 November 1999

zbMATH Keywords

\(\ell_1/\mathcal H_2\) problemcombination problemconverging upper and lower boundsmulti-input multi-output (MIMO) nonlinear systemsoptimal controller designspure \(\ell_1\) problemYoula parameterization

Mathematics Subject Classification ID

(H^infty)-control (93B36) Design techniques (robust design, computer-aided design, etc.) (93B51) Multivariable systems, multidimensional control systems (93C35)

Review text

In this article two problems are presented. Given a plant \(G,\) let the Youla parameterization be characterized by the transfer function matrices \(H, U,\) and \(V\) so that a closed loop map is achievable via a stabilizing controller if it can be written as \[ \left(\begin{matrix} H^{11}&H^{12}\\H^{21}&H^{22} \end{matrix}\right)- \left(\begin{matrix} U^1\\U^2\end{matrix}\right)\ast Q\ast(V^1 V^2) \] for some \(Q\) in \(\ell_1^{n_u\times n_y}.\) The authors study the combination problem where a positive combination \(c_1\|H^{11}-U^1\ast Q\ast V^1\|_1+ c_2\|H^{22}-U^2\ast Q\ast V^2\|_2^2\) is minimized. Converging upper and lower bounds are obtained. It is established that solutions approximating the optimal one can be obtained, which converge in the two-norm. The relation to the pure \(\ell_1\) problem is also discussed. In the \(\ell_1/\mathcal H_2\) problem the \(\ell_1\) norm of a transfer function \(\|H^{11}-U^1\ast Q\ast V^1\|_1\) is minimized while keeping the two-norm of another transfer function below a prespecified level: \(\|H^{22}-U^2\ast Q\ast V^2\|_2^2\leq\gamma, Q\in\ell_1^{n_u\times n_y}.\) Results similar to the combination problem are obtained. The solution method includes the Youla parameter \(Q\) as an optimization variable. Thus obtaining the optimal controller from the optimal solution is immediate. The difficulties associated with the inversion of maps and the correct interpolation of zeros are therefore avoided. Also, for the method presented, there is no need to characterize the achievable closed-loop map via a linear constraint on the closed-loop's impulse-response sequence. Thus the computationally intensive task of finding the zeros and the zero directions is eliminated. For the entire collection see [Zbl 0918.00041].

Recommendations

• Multi-objective optimal design of nonlinear controls
• Multi-objective optimal control: a direct approach
• Multiobjective output-feedback control via LMI optimization
• Multi-objective control without Youla parameterization
• scientific article; zbMATH DE number 1054762
• scientific article; zbMATH DE number 850122
• Multiobjective \({\mathcal H}_2/{\mathcal H}_{\infty}\) control design
• MIMO optimal control design: the interplay between the ℋ/sub 2/ and the l/sub 1/ norms
• Multi-Objective Control-Structure Optimization via Homotopy Methods
• An efficient solution to multi-objective control problems with LMI objectives

Cited in

(7)

• MIMO l^ 1-optimization with a scalar control
• An efficient solution to multi-objective control problems with LMI objectives
• Multi-objective control without Youla parameterization
• Mixed control design: Numerical optimization approaches
• MIMO l 1 optimal control problems via the polynomial equations approach
• MIMO optimal control design: the interplay between the ℋ/sub 2/ and the l/sub 1/ norms
• Solution of MIMO \boldmath\mbox\mathcalH₂/\ell₁ Problem Without Zero Interpolation

This page was built for publication: Multi-objective MIMO optimal control design without zero interpolation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4263644)