Methods On Supporting General Product Structure Design Based On Ontology

Mass customization is a kind of production pattern which could not only obtain mass production benefits and also meet user’s personalized and diversification needs,and it’s gradually occpy an increasingly important position in the market with the development of economy.Product structure is the basic unit of mass customization,beacause the personalized needs derive from implementation of design and configuration of product structure and great product structure could help develop series products radidly.Due to the various types of customized products and service environments,it’s difficult to provide a general design method for all product structure.However,the common features could be extracted to make some common methods aiding product structure design.At present,most of research on design and configuration of product structure centre on single product type,and few general methods are involved.Data storage and management systems of each department are relatviely independent beacause fo the lack of univeral and computer readable knowledge representation method.It also limits the implementation of general method.In this paper,universal product family knowledge representation,general product structure similarity and new product design evaluation are provided to assist and optimization of general product structure aiming at probloms above.This paper studies ontology expression model of the universal product family knowledge base.Model about key elements of the product family,such as main objects,different types of components,constraint relationship is built.The construction method of product family ontology is also provided based on the analysis of the important features and usage of ontology description language.A general method for determining similarity of product structure for CBR technology is proposed using the same judgement,which is based on the analysis of shortcomings of existing method.The main characteristics of general product structure are expressed in matrix form.Product structure is mapped into a low dimensional space by non-negative matrix factorization which simplifies the calculation process and enrich the content of management.The superiority of the method is analyzed by a specific example.Aiming at the kind of product fuzzy overall attribute that can’t been analyzed by accurate calculation,This paper studies the relationship between fuzzy overall attribute and product structure.Operations during product design process are refined to edit operations.Fuzzy overall attribute is given by expert evaluation.In order to ensure objectivity and rationality of expert evaluation,consistentcy check is used.LM-BP neural network is used to connect fuzzy overall difference and local difference.The new design product is determined if meet usr’s need by this model.