Neural network modeling for tool path planning of the rough cut in complex pocket milling

Abstract

This paper presents a new approach to rough-cut tool path planning of pocket milling operations. The key idea is to formulate the tool path problem into a TSP (traveling salesman problem) so that the powerful neural network method can be effectively applied. Specifically, the method is composed of (1) digitization of the pocket area into a finite number of tool points, (2) a neural network method (called a SOM-self-organizing map) for path finding, and (3) postprocessing for path smoothing and feed rate adjustment. By the neural network procedure, an efficient tool path (in the sense of path length and tool retraction) can be robustly obtained for any arbitrary shaped pockets with many islands. In postprocessing, (1) the detailed shape of the path is partially modified by eliminating sharp corners of the path segments and (2) any crossovers between the path segments and islands. With the determined tool path, the feed rate adjustment is finally performed for legitimate motion without requiring excessive cutting forces. The validity and powerfulness of the algorithm is demonstrated through various computer simulations and real machining.