Research And Application On Case-Based Reasoning Technology For Rapid Design

Nowadays, the trend of manufacturing sector globalization and the accelerating pace of product refresh rate shifts the focus of manufacturing industry competition in the market's rapid response and product innovation. The development and application of CAD system for rapid design is one of the key links to improve enterprises'rapid response and innovation capability. Supported by National 863 Projects and provincial High Tech Research Projects, intelligent design system's building approach and key technologies to achieve rapid design of mechanical products by CBR technology are emphatically studied, a system supporting the construction machinery product rapid design is developed. Main study results and innovations are as follow.1) A XML-based complex case expression model and case base organizational model based on tree structure are proposed. Using the object model, foundation class definition, xml schema and links technology to express the tree-structure-like complex case with dynamic characteristics, multi-level dynamic case base for dynamic case storage and management is proposed and built, Native XML database and relational database are utilized to organize the case and relational model properly, which laid the foundation for the building of on-line resource library and implementation of collaborative design;2) In order to enhance the system performance and operating efficiency, a case base maintenance policy based on fuzzy rough sets theory is put forward. Mining and reduction fuzzy adaptation rules by fuzzy discrepancy vector, updating case base by representative cases so as to maintain and simplify the case base, which reduces the complexity of maintenance work and the performance and advantages of this new strategy is tested in practical application.3) A composite similarity measurement is introduced to describe the matching extent of cases from the semantic and structural aspects, A multi-level retrieval model, which can enlarge the distinction between feature attributes by classifying them, meanwhile, a new retrieval algorithm based on bitwise indexing and similarity mapping list is presented, which raises the retrieval speed by bitwise operation and reusing similarity calculation results. Finally, the conciseness and efficiency of this algorithm is proved by example;4) On the basis of analyzing the mechanical product design features, take the CommonKADS modeling method as instruction, PCM methodology-based task-structure model and working-sheet for expressing and acquiring design knowledge are presented, and on this basis, a knowledge modeling of four-layer structure is proposed, product design task decomposing structure and expression method supporting product design knowledge organization and expression is analyzed, which provide a reliable method for acquisition and expression of product design knowledge in lamination and classification.5) Referring to the key issues of system framework and development process, the development architecture, data model design and system framework are analyzed, key functions are introduced, such as design process navigation, product design guide, etc. Finally, the research findings-construction machinery rapid design system is applied in several domestic machinery manufacturing enterprises and achieves satisfactory results.