Research Of Multidisciplinary Design Optimization Base On Cooperative Co-evolutionary Genetic Algorithms

With the development of the aerospace industry,A new engineering disciplinewas developing since the1990s,that is called Multidisciplinary Design Optimization(MDO) methodology.It can solve the problem of large-scale complex engineeringsystem’s optimization design.Collaborative Optimization (CO) algorithms is one ofthe MDO algorithms which is most potential. CO has been studied for a long time inthe abroad and be going deep in our country. This paper compares and summarizes theMDO algorithms.And the emphasis are on the analysis, research and perfecting of COmethod.The main contents of this thesis are as follows:1) Firstly analyse the advantages and disadvantages of CO based on thesystematic study of CO framework.Then carried on deep research to theshortage of CO method. Based on the recent theory and existing researchstudy,this paper proposes that handle system level optimization byCooperative Co-evolutionary Genetic Algorithms(CCGA) which needn’tthe derivative information of the object function to deal with the additionalconstraint in the system level optimization of CO.2) Based on the basic principles of Genetic Algorithms(GA),the Frameworkand procedure of CCGA are introduced especially. Some results that CCGAis more efficient than Simple Genetic Algorithms (SGA) are obtained fromthe calculation of several testing functions. On the foundation that analysesof the computation resultst,the phenomenon that the good results are notpreserved and inherited to the next generation is found in CCGA. In order toimprove the CCGA, A improved algorithms which introduces the method ofoptimal individual preservation is applied.And the simulation resultsindicate that the effectiveness of optimal individual combinationpreservation method.3) In order to deal with the equality constraints and inequality constraints in theoptimization problems, The augmented lagrange multiplier method wasintroduced in CO. 4) Inorder to solve the problem that constraint functions are non-smooth evendiscontinuous, CCGA and augmented Augmented lagrange multipliermethod are introduced in the system level optimization of CO. Thesimulation results indicate that this method not only guarantees globalconvergence, but also improves the stability.