Modeling Texture of Peripheral-Milled Surfaces Using a Neural Network and Fractal Method With Evidence of Chaos

Abstract

Modeling texture of milled surfaces using analytic methods requires explicit knowledge of a large number of variables some of which change during machining. These include dynamically changing tool runout, deflection, work-piece material properties, displacement of the workpiece within its fixture and others. Due to the complexity of all factors combined, an alternative approach is presented utilizing the ability of neural networks and fractals to implicitly account for these combined conditions. In the initial model, predicted surface points are first connected using splines to reconstruct 3D surface maps. Results are presented over varying several cutting parameters. Then, replacing splines, an improved fractal method is presented that determines fractal characteristics of milled surfaces to reconstruct more representative surface maps on a small scale. The fractal character of self-similarity within surfaces as manifested by the fractal dimension provides evidence of chaos in milling.