Performance in Oblique Cutting using Conventional Methods and Neural Networks

Abstract

The oblique cutting process has been extensively studied in the past, and was shown to be a better representative of many practical machining operations than orthogonal cutting. In this paper, an existing conventional mechanics of cutting approach to forces and power, as a direct modelling methods for performance prediction, is discussed. The number of process variables required for this process of performance prediction are highlighted, and the limitations of the model are shown. A neural network architecture is developed for use as a direct modelling method, to predict forces and power in a single-edged oblique cutting operation. Oblique cutting experiments covering a comprehensive range of tool geometrical features were carried out to verify the predictive nature of both the traditional and neural network models. The quantitative predictive capability of using traditional and neural network approaches are compared using statistical routines, which showed that neural network predictions for three force components were ± 3% close to the experimental values, compared to ± 15% using conventional methods.