Research On Principle And Method Of Evolutionary Design Oriented To Mechanism Configuration Design

Globalization economy makes competition in market more and more drastic. To increase competitive ability of products, it is necessary to improve product quality and shorten developing cycle for factories and enterprises. As we known, mechanism configuration design is the key step for new products, but Up-Down and sequent decision-making mode is used in the traditional design method. It costs long design time and makes efficiency low. The local modifying design will make influence on overall design course, so that many problems are caused, such as bad agile and flexible ability. Therefore, it is difficult using traditional design methods to meet the need of modern complex products. Based on natural evolution principle, the evolutionary design method oriented to mechanism configuration desgn is proposed to improve design ability, such as flexibility, agility, quick response, design automation and intelligence. The existing theory and methods of evolutionary design methodology and its application for complex mechanism configuration have been investigated. Based on that, the following work is done as follow, the confused concept of evolution is clarified by comparing evolutionary design with evolutionary computation; the basic principle of evolution design is proposed by simulating nature evolution phenomena and process; the relevant definitions are described; the uniform expression of evolution design is presented, which indicates the key problems in evolutionary design are to obtain the evolution routine and design the evolutionary operations. Finally, the framework of this method is established. An approach to acquire evolution routine of design based-on constraint network is proposed. The composite process of design is considered as the constraint satisfactory problem. The features of constraint about design are discussed in detail. The design constraints are classified according to their roles, and then the design constraint system is constructed. With the valid connection among constraints, the constraint network of evolutionary design can be got and run through design gene. So the design constraint chains are obtained. Design knowledge and methods play an important role in the resolution of constraint chains, so that the constraint chains can be transformed into the evolution routine of design. By changing one or more constraint networks, several valid constraint chains can be got and the multiple evolutionary routines can be provided for product creative design schemes or the development of new products. Two modes of gene operation are proposed according to different design need. One is the genetic model of evolutionary design, and the other is the evolutionary mechanism of scheme design based-on MAS. The former emphasizes on the functional construction and optimal configuration. It realizes function description in the formal way, builds up the functional construction matrix of design product, then extracts design gene on the basis of that matrix. The genetic operation is given as evolution operation mechanism. Evolutionary design gene model is built up, which can realize design with organization-self characteristic. The latter investigates that the way of evolution operation of design gene in creative design of product scheme. The research on agent is one of focal problems in distributed artificial intelligence because of its adaptation, sociality, organization and other features. Those characteristics are similar to evolutionary design, so evolution system can be consider as Multi-Agent System (MAS). The design agent model and classification is given. The principle of MAS and integrated evolution mechanism of gene agent are also investigated. It can support for complex evolutionary design in the way of information feedback. Therefore, the prototype system oriented to mechanism configuration design composed of three function modules has been developed, which is mechanism configuration design of evolutionary system. Two relevant researches on evolutionary design are developed. The Approach to identification of link curves transfers atlas match into model recognition, and then uses the genetic algorithm to deal with features of curves to identify atlas. There is chaos in evolutionary design activity. The resource is investigated and local dynamics is used to control the chaos by introducing outer information. It can make evolutionary design realize the expected result. Furthermore, the shearing mechanism of shearing machine has been designed by using the developed prototype system. The design case indicates that the proposed method is feasible in its application to mechanical design.