Versatile feature extraction for machining planning.

摘要

Feature extraction for machining planning has been a weak link in most implementations of integrated CAD/CAM systems. Current research is often focused on extracting simple shaped form features such as holes, slots, rectangular pockets etc. Such form features are useful since they reduce the downstream tasks of machining planning (selection of tools, tool paths, and machining conditions) to relatively trivial ones. However, the approach encounters algorithmic as well as technical problems. The requirement of simple shaped features makes it necessary to recognize intersecting feature instances, resulting in prohibitively large search spaces. The application of canned routines to such intersecting features also results in non-optimal machining plans. This research addresses these two issues by developing a feature recognition method that could reduce the search space by using generic shaped pocket features. A novel strategy, called depth direction, is proposed to provide guidelines for the feature extraction process. The developed system makes sure that all the alternative feature interpretations are valid in terms of machining accessibility. We further demonstrate that the increased difficulty of machining planning arising from this choice can be tackled relatively efficiently. In doing so, a new methodology of machining of generalized pockets using multiple tools is developed. It is believed that the method of dividing one prohibitive search problem into two less complex problems will lead to more robust process planning systems for practical parts.;The issues mentioned previously are two important components of any automated process planning system. It is important to understand how these and other functions such as setup planning, fixture planning, etc. can co-operate within an integrated manufacturing planning system. The thesis presents a new architecture to achieve such integration. The main points of the architecture include an open-architecture, and a distributed functional model. The framework of an open-architecture system for computer-aided process planning (OSCAP) is proposed in this thesis.