| Summary: | 碩士 === 國立中正大學 === 化學工程研究所 === 90 === The purpose of this disseration is to design an ideal controller controling efficiently many plants. In this,we choose Integral of Sequared Error (ISE) as our cost function and minimize sums of ISEs of plants,besides,in order to avoiding parameters we finding to leave out boundary that this can cause controller gains so highly that occur highly oscillation on the system,so we include error variable rate among our performance index.As for looking for optimum controller parameters,we apply differential evolution algorithm
(DEA) to perform the work so as to avoid dropping the local region.Besides,we also try using linear matrix inequality to stabilize many plants simultanously.
in this disseration,we use quadratic stability criteria to contruct a BMI problem.Here,we abstract some definitions and theorems that some scholar provided,which are Hurwitz stable,Hermite-Fujiwara matrix,etc.Then,convert BMI problems into standard LMI problem by potential reduction method and
ellipsoid intersection method,and use interior-point method to solve LMI problem. Last, geting optimum controller parameters.
In fifth chapter,we apply DEA to optimial controller design of interval system.Here,we use test of robust stability as our stable ascertain,and also satisfy that 32 edge polynomials stabilize simultaneouslywhich Chapellatm,etc privided.In this disseration,we perfrom multi-model method and find controller parameters by two layerDEA method, and compare difference of them. Expect to be able to look for the best efficiency but not lost accuracy.
|
|---|
