Research And Application On Approximate Model Design Method For Multivariable System Based On Genetic Algorithm

Considering uncertainty of mathematical model for the control system, robust control appears and becomes one of the important branches and research hotspots of control theory, and several analyzing and design methods for multivariable robust control systems are formed. Multivariable robust control system theory and design method based on the approximate model has advantage of clear physical concept and easy engineering implementation. However, it is difficult to select approximate model in this method. This paper introduces Niche Genetic Algorithm into designing approximate model, thereby overcomes this difficulty applying intelligent optimization method and obtains preferable robust control effect.This paper summarizes the research work of the author, which includes 6 chapters.Based on studying and improving a niche identification technique, this paper proposes Niche Genetic Algorithm using Improved Niche Identification Technique (INIT-NGA) and obtains better performance of the algorithm. On this basis, an approximate model design method based on INIT-NGA is proposed. This new method is easier to use than traditional method, and the optimal approximate model obtained can reflect the dynamic characteristic of the control plant better.Based on the above work, this paper develops successfully robust design CAD platform for multivariable system, including approximate model design module, precompensation matrix design module, main controller design module, robust stability test module and control system simulation module, etc. This platform optimizes the parameters of control system with Genetic Algorithm and realizes intelligent design for multivariable robust control system. It has few adjustable parameters and is convenient to use. Moreover, this paper verifies the design effect of the platform by liquid level control experiment on double-holding water tank laboratory system.Finally, control and optimization problem of combustion process of the power boiler are studied in this paper. The multivariable robust control system for a power boiler combustion process is designed and simulated with multivariable system robust design CAD platform. On the basis of this, an intelligent optimization method for the oxygen concentration in f1ue gas is proposed. This method is realized and integrated into SCON-2000 advanced control software platform, and thereby realizes the automatic optimization of the set value of the oxygen concentration in f1ue gas.