Study On Product Design Knowledge Management And Reuse Techniques

Product Design is most creative stage in product development process, it can be regardedas a process that generating new knowledge based on the accumulated knowledge. In theproduct design field, around70%of product design is adaptive design and variant design,and around60%of new product design is based on the accumulated knowledge andexperience, therefore, reuse the accumulated knowledge is regarded as a better way toimprove the efficiency and quality of product design. However, the current research worksof knowledge management and reuse are focus on knowledge modeling while the researchworks about how to reuse them are inadequate. This situation directly results in seriousinformation overload during knowledge reuse process, and also makes the knowledge baseto be managed but not used. To solve the above problems and achieve the goal of improvingknowledge reuse level and alleviating information overload, several techniques are studiedincluding the framework of design knowledge management and reuse system, the modelingof design knowledge, the knowledge navigation method, the knowledge retrieval and theknowledge recommendation method. The following shows the detail of each chapter.(1) Based on the analysis of the characteristics of design process, this chaptersummarized four kinds of design knowledge reuse patterns including reference, guiding,automation and assistant. The function requirements of design knowledge management andreuse platform are discussed and the overall framework is constructed.(2) Aiming at the heterogeneous of design knowledge, this chapter proposed a multi-layer knowledge organization method. This method take the characteristics of designknowledge, knowledge management and knowledge reuse in to consideration, achieves thestructure and semantic consistency of design knowledge.(3) For the problem of information overload during interdisciplinary design knowledgenavigation, this chapter proposed a knowledge map based domain knowledge navigationmethod. The method uses natural language processing, latent semantic analysis and similarityanalysis techniques to automatically construct domain knowledge map and apply complexnetwork to analyze the internal structure of knowledge map. The experiments show that themethod can significantly reduce information overload.(4) For theproblem oftraditionalknowledgeretrieval does not reflect users’ preferences,which always leads to information overload, a multi-behaviors based knowledge retrievalmethod is proposed. A knowledge similarity calculation model is constructed considering theability to represent user’s preference of actions, the dependence between actions and theaffliction relations between users. The experiments shows that this method significantlyimproves the performance of traditional knowledge retrieval. (5) For the problem of semantic information is difficult to be extracted because of thecomplexity and diversity of design knowledge, this chapter proposed a method of buildinghybrid user model based on probability. This method combined user ratings and topics ofdesign knowledge to construct the user model and the user interest is defined as theprobability of a score is given to a specific topic. The method combines behavior-basedapproach and content-based approach to build hybrid user model, the experiments provedthat the model are superior to other methods in terms of predictive accuracy and coverage.(6) As engineering practice of the research results, a design knowledge management andreuse system is developed independently. The system allows designers to obtain designknowledge in the design environment and provides a personalized knowledge retrieval,knowledge recommendation and knowledge navigation services for designers. Theknowledge services provided by the system can significantly reduce the information overload.