Automatic loop-shaping of QFT robust controllers via genetic algorithms

One of the main problems when designing robust controllers using Quantitative Feedback Theory (QFT) is the great effort needed to make the loopshaping stage. In this paper, a new automatic loop-shaping procedure to design QFT controllers is developed. It is based on Genetic Algorithms (GAs). In contrast with typical approaches that require some restrictive assumptions, GAs use only the information provided by a cost function, without further assumptions about the search space. An appropriate cost function, taken into account both minimisation of high frequency gain and satisfaction of performance control specifications, is developed. Three illustrative examples are given to show the proposed methodology. Copyright (C) 2000 IFAC.