Automatic modeling of machining processes

Abstract

3 axis CNC milling is a ubiquitous manufacturing method in industry due to its versatility and precision. The fundamental parameters that dictate cutting performance ("speeds, feeds, and engagement") must be manually set by the machine programmer; proper operation therefore relies heavily on operator skill. In this thesis, an intelligent CNC controller is presented that uses low-cost sensors to fit an analytical model of cutting forces. The analytical nature of this model allows for favorable convergence characteristics and low computational costs. This is used to optimize cutting feeds with respect to process constraints for future movements; as more data is collected, the model continuously reinforced. This intelligent controller therefore abstracts out some of the complexities of machining and makes the process more approachable.