| In this thesis.we proposed a genetic algorithm(GA) and Hm hybrid optimization approach for robust control system design.Genetic algorithms can efficiently solve the difficult mixed optimization problems and the complex functional optimization problems.H~∞ optimization can make the designed control system to have good robustness while LQG is not suited for the systems with model uncertainties.The proposed method translates the robust control system design requirements into the constrained conditions of genetic algorithm,which makes the designed system meet the design specifications both in the frequency domain and in the time domain, while the traditional H~∞ optimization can only meet the specifications hi the frequency domain.Further, a change-to-change strategy for genetic algorithm is developed in this thesis,which improves the genetic algorithm with good performance.A fuzzy-genetic algorithm(Fuzzy-GA) is also developed in this thesis based on the Fuzzy Set concept and theory.This thesis reviews the properties and the design requirements of the robust control system.The most popular robust control system design method H~∞ optirnization,particularly, the H~∞ loop-shaping design procedure based on the normalized coprime factor are summaried.Genetic algorithms are applied to optimize the loop-shaping weighting functions.The theories and the methods of the genetic algorithms are reviewed and a change-to-change genetic algorithm strategy is drawn out. The general procedure of the GA and H~∞ hybrid optimization method is proposed hi this thesis.A Fuzzy-GA approach is developed in order to deal with the complexities and the uncertainties of the MIMO robust control system disign problem. The effectiveness of the proposed methods is confirmed through design examples and computer simulations.The proposed methods will be applied easily to other designs on H00 robust control system,because these methods don't need a complicated optimization process such as a mathematical analysis.
|
PDF Full Text Request