Process Planning Using Adjacency-Based Feature Extraction

The use of attributed adjacency (AA) feature recognition can encapsulate the engineering significance of a part and represent it as a matrix or an arc-node graph. This creates a feature recognition technique that involves scanning a matrix or graph for a combination of zeros (concave relationships/edges) and ones (convex), or smaller arc-node graphs that are predetermined to be features. AA techniques have often suffered problems when dealing with feature interactions, as a feature found must be exactly matched with those stored. This paper presents a modification of the AA matrix and employs a new concept for identifying features and the application in developing software for process planning. A unique feature taxonomy is described, which when combined with the new feature identification system creates a feature recognition and extraction system that includes curved surfaces and eliminates the need for separating interacting primitive features. The new system takes its input from neutral STEP files and produces a list of features with complete information for process planning.

cycle, and so automatic feature recognition has its place in manufacturing.