Study On Multi-method Collaborative Optimization Algorithm

With the development of the aerospace industry, a new engineering displine hasbeen developing since the1990s, that is Multidisplinary Design Optimization (MDO)methodology. It has become a new research area in international industry design,which is applied to solve the problem of large-scale complex engineering system’soptimization design. And MDO algorithm is an important research direction in MDOfield. Currently a hot spot in MDO algorithm study is that research new searchmechanism and discover new search strategy. Multi-method CollaborativeOptimization Algorithm (MCOA) is put forward by doctor Luo Wencai, which is anew optimization strategy, study and achievements on this fields are few. As anopportunity, this article has done deep research, analysis and improvement on MCOA.The main contents of this thesis are as follows:1) Firstly study the common optimization algorithm, which is includingtraditional algorithm and modern intelligent optimization algorithm, understand thebasic ideas of each kind of algorithm and every algorithm’s unique search mechanismand search process. The similarities and differences of various algorithm has beencompared, and the advantages and disadvantages of them has been summarized.2) Based on further research in MCOA, the convergence of the MCOA bydifferent cooperation form has been proved from the perspective of mathematicaltheory, so that the theory system of MCOA has been further improved.3) Choose the genetic algorithm and Nelder-of Mead algorithm, form serialcollaborative optimization method by serial cooperation form, optimize and calculatesix typical optimization test function with this method. And the results of theexperiment has been deeply analysed and summarized, the different cooperationstrategy especially iteration steps how to affect the optimization results has beenexplored. And the results show that, this method is feasible and effective under thereasonable parameters settings.4) The serial collaborative optimization method based on genetic algorithm andNelder-Mead algorithm was used to sovle a certain RLV wing plane parameters design optimization problem. The optimization model of weight discipline has beenoptimized with this methed, the results show this method’s feasibility and validityonce again as an practical example.