Form feature generation model for features technology

Abstract

The essence of CAD driven automation lies in modelling the interaction between geometry and domain knowledge. The concept of features and the area of features technology evolved to model this interaction. Feature definitions play an important role in features technology; problems such as extracting interacting and intersecting features, feature based editing, and feature conversion across domains can be attributed to the lack of rigour in feature definitions. Unlike earlier approaches, this thesis enumerates a Domain Independent Form (DIF) feature space through a Feature Generation Model (FGM). The enumerated DIF feature space is classified into a finite set of generic feature types based on the interaction of the feature creating solid (feature solid) with the solid on which the feature is created (base solid). Definitions for generic feature types are derived in their nascent states as they are created on the base solid. The completeness of the finite set of generic features is argued through an intuitive proof. Keeping feature definitions free from the subjectivity of application domains enables the design of general methods for feature extraction and manipulation. Feature extraction is conceptually modelled as the reverse of the feature generation process. The completeness of the DIF feature set ensures the presence of at least one DIF feature in its nascent state on the part solid. Shape independent characteristics of faces (Dynamic Topological Status) are used to identify and extract the faces associated with this feature. The nascent feature is recognised and removed, bringing other feature(s) into their nascent state. This process continues until all features are extracted. The extracted DIF features serve as a consistent high level intermediate representation between geometric representation and the application’s feature space. Geometric and knowledge based reasoning is employed to map the extracted DIF features to a particular application. The feature extraction algorithm has been implemented and tested for solids with both polyhedral and exact boundary representation. Mapping to the machining domain has also been implemented and tested. Next, the thesis discusses the use of DIF feature definitions-owing to their evaluative nature-and the inherent localisation in feature based modelling. A Feature Based Modelling (FBM) scheme is proposed in which the B rep can be directly manipulated through features for model construction and editing. The separation of domain specific features from DIF features, the definition of DIF features through the feature generation model, the extraction of generic features, and the use of DIF features as a high level intermediate representation for feature based reasoning distinguish the present work from earlier efforts in this area.