Centralized Robust Multivariable Controller Design Using Optimization

Abstract

The paper deals with the determination of controller parameters for multivariable systems by means of parameter optimization. In order to reduce the computational burden, the starting point in the optimization is determined through information provided by the feedback controller parameterized using the multivariable internal model controller (IMC) computed from the plant's p/q moment approximant and the original plant's multi-loop controller parameter magnitudes. Thereafter, controller parameter optimization proceeds to completion using the MATLAB optimization toolbox which is widely available. A potency of this technique is that all performance specifications and stipulated constraints are easily accommodated in the problem formulation thus facilitating complete problem solution in one go. The technique has been found to be effective for all plants considered so far and generally produces closed loop systems with favorable characteristics when compared to systems designed by similar methods.