Study On The Decomposition Strategy Of Multidisciplinary Design Optimization And Its Cloud Service Platform Implementation

In order to solve the discipline interaction complexity and computational complexity in the process of multidisciplinary design optimization(MDO), the thesis researches the decomposition strategy of multidisciplinary design and its implementation technology of cloud service platform. On the one hand, it explores how to adopt different decomposition strategies to organize and manage the design process for different multidisciplinary optimizations. On the other hand, it discusses applying the idea of cloud manufacturing in the MDO process, and solving the MDO problem by means of cloud services. The main procedures are as follows:Firstly, it studies the optimization method of analytical target cascading, which decomposes hierarchical multidisciplinary optimization issues, disassembling a complex engineering system into a number of subsystems. To make full use of the advantages of analytical target cascading, it discusses the hybrid application of analytical target cascading and intelligent optimization algorithms through specific examples.Secondly, non-hierarchical multidisciplinary optimization problems are decomposed through sensitivity analysis, on the basis of which a multidisciplinary optimization framework is proposed and demonstrated with concrete examples. Due to the limitations of sensitivity analysis in solving large scale coupling problems and nonlinear problems, a new method of sequenced collaborative optimization is proposed, which applies design structure matrix to MDO problems. Through the clustering algorithm, Muiltidisciplines are aggregated into subgroups, which are made to be a tree structure without feedback loops by the dividing operation and adding constraints, and then parallel optimization analysis can be conducted.Thirdly, it studies the approximate modeling technology in the process of multidisciplinary design optimization, outlines some basic principles of it. Further, it conducts a comparative analysis of the approaches to collecting sample points in the process of approximate modeling and several approximate models. It combines evaluation examples of multidisciplinary optimization, uses a variety of approximate models to conduct approximate modeling analysis, and adopts different optimization algorithms for optimization comparison.Finally, in view of the advantages of cloud manufacturing in integrating the optimal resources, the idea of cloud manufacturing is applied to the process of multidisciplinary optimization. The prototype system of a multidisciplinary cloud service platform is proposed. The service mode of optimizing resources and the calling process of service resources are expounded. Also, on the basis of service invocation, the application processes of examples under different solution strategies are realized by means of workflow.