Research On Semantic-based 3D Product Information Search In Variant Design

With the great advancements of industrial modernization and informationalization of the Internet, the rapid effective and accurate information search system is particularly important for variant design of mechanical products. Some datas show that about 40% is directly using the existing design resources, about 40% is using variant design to get the new designs, and only 20% is using new design.To search proper CAD models which meet engineers' requirements for design reuse, some more same or similar parts is selected for developing new products. The way of design reuse would cost reduction and efficiency improvement. The structural Ontology can be used to deseribe knowledge for shortage problem of semantic expression in existed 3D model retrieval systems. The following works of this study focus on the representation of product design knowledge and the method model of semantic similarity.Web Ontology Language(OWL) is adopted for representing the information of 3D model. Ontology knowledge abstracts the facts by concepts and relationship of concepts. Firstly, the knowledge structure of CAD product model is analyzed in the field level of variant design, and the semantic concepts woulde be classified by model design stage in knowledge. Secondly, knowledge of CAD model is effectively expressed by Ontology Language OWL DL(Description Logic) with strong logical Description and reasoning ability of decidability, and make the automatic semantic matching possible. Finally, the automatic or semi-automatic semantic reasoning and intelligent information searching would be realized by the deep of knowledge structural Description in Ontology techonology. The knowledge representation based on OWL DL would make master model of CAD product and the query model convert into structured models with concepts and attribute characters. The theory basis of chapter would be helped for similarity method in the way of collection.The method of model similarity by OWL-based semantic characteristics is designed for the core of intelligent searching. This new menthod make the evaluation benchmark switch from the object itself to the set of semantic description, and make out similarity evaluation by a quantitative or qualitative way in collection. Firstly, as similarity unit,the quantifiable set of semantic information are extracted from the OWL-based product model for similarity. Secondly, a matching of total weight combining subgraph isomorphism and Tversky algorithm was proposed for similarity calculation. Finally, comprehensive quantitative assessments of experiment show that this method could effectively give the similar degree of the two models, and its feasibility and effectiveness would satisfy the requirements of engineering application.The prototype system of product semantic search in variant design would be implemented by MFC framework based on VC++ 6.0 development tools. For ontology knowledge base, we use ontology language OWL DL in Protégé tool to describe product master model and user query model. Combining the technologies of ontology and information searching, ontology knowledge would be reasoned to semantic matching for model similarity in effective collection description. On the basis of traditional information search system keyword-based, we develop method of semantic search based on ontology knowledge of product design. Finally, an example of horizontal container customized products is used to verify the feasibility of our prototype system in product variant design.