Research On Organization Of Mechanical Assembly Structure And Application In Conceptual Design

Intelligence design of mechanical product is the ideal state pursued by designers and the target for community’s efforts. Conceptual design is the conncetion between the abstract requirement and the concrete structure. This is why conceptual design has strong complexity, experience, fuzziness and dependence to environment. At present we can only depend on designers’experience during the stage of conceptual design without the aid of a reliable computer software. With the wide spread of modularization and standardization of mechanical products, how to integrate the quantized information and designers’experience information is increasingly becoming an urgent question. Thus, the idea of realizing intelligent computer-aided conceptual design system based on ontology knowledge base and self-organization reasoning mechanism is put forward. Based on this idea, complex network, ontology, XML and self-organization are adopted as key technologies to realize intelligent conceptual design of mechanical products. This method may offer theory and technology sustenance to improve design automatization.The main research works and achievements in this paper are as follows:1. Analysis of Static Characteristics and Dynamic Characteristics of Mechanical Products with Part GranularityOn the basis of extensive analyses of typical mechanical assembly structure network, we innovatively proposed a new concept called assembly network which is high level of clustering and partially consistent with the scale-free network, and has other6obvious characteristics. These results prove that mechanical products with part granularity are system of high organization. By the analysis on static characteristics and dynamic characteristics of assembly network, we find that although mechanical parts themselves do not have the ability of self-determination, they have some common characteristics in comparison with the system of self-determination.2. Ontology-based Knowledge Model of Mechanical Assembly StructureFunction is realized by assembly. According to the types of functions, we proposed a concept called mechanical assembly structure, categorized them and analyzed their structural features. We proposed a method to construct knowledge ontology called MFSO, which divided the process of construction and description of ontology into5stages. We used a formalizing description based on Frame Representation and XML to represent knowledge ontology of mechanical assembly structure.3. Analysis of Organization of Mechanical Products with Mechanical Assembly Strucuture GranularityFirstly, a contrast is made between mechanical products with mechanical assembly structure and biology focusing on structure, life cycle and evolution to prove that mechanical products are system with organization. Secondly, we analyzed and found3characteristics of mechanical assembly structure network. Thirdly, we made a definition to mechanical assembly structure organization, and analyzed3factors, choice uncertainty, feature uncertainty and fit uncertainty, which may affect its organization. At last, we proposed a measure method to calculate the uncertainty of mechanical assembly structure using the Entropy Theory.4. Construction of Self-Organization Evolutionary Model of Mechanical Assembly StructureFirstly, mechanical products with mechanical assembly structure granularity having the necessary condition and environment of self-organization has been proved in four aspects:system openness, conditions far from equilibrium, non-linear interaction and fluctuation. At last, competition and cooperation mechanism of mechanical assembly structure is analyzed to find their self-organization dynamic characteristics.5. Knowledge Application and Improvement Based on Multi-Dimension Sandpile Space and Improved ACT FrameworkMulti-dimension sandpile space is put forward to represent and reason experience knowledge which is characteristic of fuzziness. Knowledge base using multi-dimension sandpile space can not only deal with uncertain experience knowledge, but also keep knowledge base in an acceptable size when learning new knowledge. And it has an efficient self-correcting mechanism to update knowledge in system. Anderson’s ACT theory is improved to construct knowledge base, which can manage both procedural or production knowledge and experience.At last, three prototype systems, Analysis and Generation of Mechanical Parts Network System (AGMPNS), Knowledge Ontology of Mechanical Assembly Structure System (KOMASS) and Self-Organization Based Computer Aided Concept Design System (SOBCACDS) were developed to prove the method and mechanism proposed in this paper.