Process planning by recognizing and learning machining features

We present two methods for process planning of 2.5D machined parts. The first method is based on feature recognition from a 3D model. We embedded the shape and the machining method of two generic classes of machining features in a set of OPS5 rules to form a machining feature recognizer. When successfully recognizing a machining feature, machining instructions, in terms of the tool entrance face, drive face and part face, for cutting the machining feature will be generated and further processed to produce NC codes. The second method is based on learning the shape and the machining method of the machining feature. When a machining feature cannot be recognized by the former feature recognizer, the user can use the machining feature as a positive training example to instruct the system about the tool entrance face, drive face and part face of the machining feature. The system then builds a new rule, using the boundary shape of the unrecognized machining feature as the rule's matching condition and the acquired machining instruction as the rule's action. The new rule can be used subsequently for process planning of machining features that have shapes similar to the memorized one.