Feature engineering for machine learning in automatic determination of cutting conditions

Abstract

In order to solve the labor shortage in manufacturing sites, full automation of machining, including NC program generation and other setup work, is required. In this study, an automatic cutting condition determination system was developed to achieve fully automated NC program generation. The cutting conditions that served as the teacher data were the conditions that calculated to make the machining time minimize through repeated simulations, and the relationship between the cutting conditions and the geometric features of the product shape (e.g., cross-sectional area and curvature of the curve) was machine learned to create a learner. Automatic determination of cutting conditions based on geometric features for an unknown product shape was performed, and its usefulness was confirmed by the rate of correct answers.