Design Case Representation And Reasoning For Mechanical Products Based On Hierarchy Design Structure

Confronting the challenges of globally increasingly competitive market, innovation and speedy design of mechanical products have become the focus of competition among enterprises, while, knowledge is the fundamental resource of product designs.The Enterprises have to transit their development focus to the management of product design information in response to the quick and innovation product design, i.e. effectively build up the product design information for reutilization.This thesis took the advantage of case-based reasoning, integrated mechanical products hierarchy design, conducted an intensive research on five aspects of mechanical product design information management, including the representation and integration of products lifecycle information, construction of hierarchy design case database, automatic recognition of product design structures, index and modification of similar cases. The significance of this research is to eliminate the "isolated information resources" phenomena essentially, enhance the reuse of accumulated information, and advance the mechanical product innovation and speedy design.To solve the problem of oversized granularity of information management in Product Lifecycle Management(PLM), we comprehensively analyzed the relations among the contents of product components, information carriers and information itself from three dimensions of lifecycle, structure hierarchy of designs, and disintegration of design structure, We then put forward the conception of design structure unit and function surface, It is product、mechanism、component、part、design structure unit and function surface which The cores information described, which provide the carrier for representation of product design intension and schematic information. Finally, we established unified3-D whole lifecycle information modeling.According to the mechanical product features, using ontology technology information representation method, we built the class-relation-attribute modeling which possesses mechanical product information characteristic. Furthermore, we brought out the product lifecycle information ontology that layed the foundation for information representation.we utilized the advantages of case-based reasoning and the feature of layering mechanical product design structure, and built up the mechanical product design hierarchal case database model. According to requirements and features of case library, we describe the case upon ontology and decision table, and design the hierarchical of structure mechanical products design case database. The information of mechanical products is gathered through data mart, and the hierarchal structure case library is established, for knowledge discovered and reused lays the foundation.In solving the problem of the lack of descriptive information for product design structure in the PLMS, we analyzed3-D mechanical product model descriptive procedure, extracted the surface, edge information from3-D product B-REP model and the structure of design and topological relations from the assemble model. The design contracture auto recognition system was developed by the technology of matching the unit, parts, component, mechanism and product level structure associated graph (AAG) and AAG sub-graph. During the process of sub-graph matching, we developed knowledge based matching algorithm, compressed the sub-graph matching space, and enhanced the matching quality and efficiency. We also structured the ontology design knowledge database and AAG sub-graph database, provided basic frameworks for expansion of the design structure database utilized that ontology dictates conception and open infrastructure of ontology.Through reduction based on rough set theory,we Remove redundant properties of mechanical product cases. For the successive value of partial attributes we come up with successive attributes discretization method based on the combination of rough set knowledge classification approach and expert experience, which ensured the compatibility of decision table and engineering significance of attributes. For fuzzy case decision table of condition attributes, we listed four model of case retrieval and applied fuzzy similarity calculation method. We come up with similarity calculation method of multiple fuzzy attribute cases that is a nearest neighbor algorithm by adding correlation parameters. The quality of similar fuzzy case retrieval is improved by using the algorithm.After fully considered the pros and cons of rule based reasoning (RBR) and case based Reasoning (CBR), we draw on each other’s merits and put forward a CBR-RBR integrated reasoning strategy which applied on the modification of similar mechanical structure and provided a new approach which can automatically modify CBR mechanical design.