Research Of The Collaborative Optimization Based On Particle Swarm Optimization

The achievement and capability of product designing are key factors to the innovative and competitive power of a nation's or an area's industry. Designing complex product is a multidisciplinary-crossed, integrated designing and optimization process. Multidisciplinary Design Optimization (MDO) meets the requirements of the design optimization and development of the engineering system. MDO is a kind of design methodology integrating design method, design knowledge and modern information technology which can solve the designing problem of the cosmically complex engineering system.The purpose of this paper is to analyze and improve the method of the MDO - the Collaborative Optimization (CO), and to apply the CO to the engineering projects.Firstly, I reviewed all the work and achievements of scholars around the world in the research of MDO, and described the content of the MDO technology in detail.Secondly, this thesis analyzed the computation characteristics of the original CO method and deeply discussed the computation difficulties of the CO method from different aspects. And by dint of math example, it also pursued the sources of the computing hardness of CO methods and makes a simple evaluation and improving direction. Furthermore, Considered the computing complexity caused by the original CO method and the improvement direction, this article applied a new creationary evolutionary method - Particle Swarm Optimization(PSO)to the traditional CO method. By modifying the system frame, this paper presents a new structure named improved Collaborative Optimization which is based on Particle Swarm Optimization (PSO-CO), this novel method is tested by the MDO standard examples. The computation results demonstrate that the proposed PSO-CO has better calculation ability.Finally, according to the four designing principles, by transforming the compute structure of CO, this section describes a novel CO mode which involves the PSO evolutionary algorithm as an optimizer, and the new-model method proposed in this thesis is also tested by MDO standard examples. The computation result illustrates that the new-mode CO method is robust and has its validity in applying to the practical design.