A knowledge-based approach for the extraction of machining features from solid models

Computer understanding of machining features such as holes and pockets is essential for bridging the communication gap between Computer Aided Design and Computer Aided Manufacture. This thesis describes a prototype machining feature extraction system that is implemented by integrating the VAX-OPS5 rule-based artificial intelligence environment with the PADL-2 solid modeller. Specification of original stock and finished part geometry within the solid modeller is followed by determination of the nominal surface boundary of the corresponding cavity volume model by means of Boolean subtraction and boundary evaluation. The boundary model of the cavity volume is managed by using winged-edge and frame-based data structures. Machining features are extracted using two methods : (1) automatic feature recognition, and (2) machine learning of features for subsequent recognition. [Continues.]