Research On Knowledge-based Method Of Rapid Design For Mechanical Products

With the rapid development of global manufacturing industry, the speed of product renewing is accelerating. Only by constantly improving their response to market demands, shortening the product design cycle and reducing design and manufacturing costs, can enterprises seek a place in the fierce competition. Under the support of the National Science and Technology Plan (2012BAF12B08) and the National Science and Technology of Small and Medium-sized Enterprise Technology Innovation Fund (10C26212110396),based on the analysis of KM and rapid design theory, method and application status at home and abroad, this dissertation addresses the problems in the process of rapid design of mechanical products by taking a typical product--the multi-spindle numerical control (NC) lathe as the research object, and conducts an in-depth study of the knowledge framework and implementation methods of rapid design theory by applying the idea of system and integration.Considering the customer requirement can not be fully expressed in dynamic and the law of creative design, the rapid design frame based on customer requirement is built for mechanical products. The basic element model and basic element implication tree model are used to describe and divide the customer requirements so as to obtain the common and individual requirements. The variable precision rough set (VPRS) is used for design rule extraction and knowledge reduction, which can extract the common design rules. The rapid design system frame is built on the condition that the customer requirements is regarded as input, the design rule is considered as standard and the final model of mechanical products is counted as output.Aiming at the knowledge construct of mechanical products is complex, the characteristics of the product heterogeneous information is multi-dimensional and multi-hierarchy, the method of knowledge acquisition and representation are studied. On the basis of axiomatic design, the product knowledge module is partitioned. The classification feature of knowledge module is extracted by adaptive genetic algorithm (AGA). In the model of knowledge representation, the web ontology language (OWL) is used to express the various feature information like product demands, functional requirements and design parameters etc. The case indicates that the method of knowledge representation and acquisition can effectively solve the bottleneck problems as well as realize product rapid response design and lays the foundation for the knowledge reuse in the process of rapid design.Since the mapping from the design space to performance space result that the existing knowledge cannot be applied to the new product design in the traditional design process, the product knowledge reuse model based on multi-hierarchy case decomposition is built. The model is guided by the inverse mapping from performance space to design space. In accordance with the design of grain size and sequence of decomposition, the mutual zigzagging between the function domain and structure domain is realized by multistage case decomposition technique, and the uncoupling or redundant instances are reused in the new design scheme, which enables the designers in different departments to share the knowledge, get the effective matching information and reuse it so as to make the design cost reduced and the efficiency of knowledge reuse improved.In regard to the problems of design defect and design iterative in each stage of product rapid design, a study of knowledge base evolution in design process is conducted, the established machine learning model by naive Bayes classifier is used to supervise learning, and applies it to knowledge mining to converse unstructured data to structured data for machine learning. Using domain ontology technology to extract information and similarity algorithm to evolve ontology, a cluster analysis of product structure is applied to knowledge mining. The continuous optimization and upgrading of product knowledge base and design rule base are realized. As a result, the repetitive work in the design process is reduced and thus the quality of rapid design is improved.To conclude, combining the research result with enterprise engineering practice, the multi-spindle NC lathe’s framework of rapid design support system based on the above theory and method is conducted. The research on framework, structure and specific process of rapid design system are accomplished. The use of software engineering technology allows carrying the prototype system, gradually achieves the general management of product rapid design for users, meeting enterprises’ demand for an automatic and intelligent rapid design. The application of rapid design system at the enterprise is discussed to prove feasibility and effectiveness of the proposed methods.