Research On Semantic Blend In Feature Modeling Systems

Feature modeling is product-oriented and it is an important milestone of CAD development, which makes integrated CAD/CAPP/CAM possible. It is also the theoretical and technological foundation of solving products problems. Semantic feature modeling is a declarative feature modeling approach that not only provides a well defined specification of feature semantics, but also effectively maintains semantics during the modeling process. In fact, one of the basic ideas of semantic feature modeling is the ability to associate functional information with shape information in a feature model, which forms the semantic of the feature together. In the course of reconstruction, the critical problems are how to maintain its semantic conformity, how to assure the stability and integrality of product information, and how to improve its efficiency.This dissertation researches the semantic blend feature from two aspects, theory and system design. In order to make the blend feature suffer less from the persistent naming problem, this dissertation deeply researches the blend surface and attachment of blend feature and presents an approach for constructing model and computing shape of semantic blend feature. The approach makes model with blends more intuitive and high-level, so the blend feature semantic can be maintained through validity maintenance. Using the approach, more complex blending is possible, and the blending in this dissertation is based on rolling-ball technology. Blend feature is smooth and sensitive. When 3D model is reconstructed, blend feature can not be shown accurately without considering its sensitivity. This dissertation presents a sampling approach about blend according to its particularity. The approach regard the feature sensitivity as a factor in order to compute the sampling density, and balance the distribution of sample points. The approach can generate high quality sample points and effectively reconstruct 3D model including blend feature. The research can be applied in HUST-CAID. The result shows that the approach can accelerate the designed speed and improve the designed efficiency and the universal property.