Safe optimization for feedrate scheduling of power-constrained milling processes by using Gaussian processes

Abstract

To set up a rough milling process with a maximized stock removal rate, an optimal feedrate schedule has to be determined without violating a given spindle power constraint. A method is presented, to achieve this by adapting the values of the feedrate schedule iteratively with every production cycle, based on predictions of a Gaussian Process model, and is validated on a face milling process of grey cast iron (EN-GJL-300). The Gaussian Process model is trained with data, measured in previous production cycles, and is used to ensure satisfaction of the given spindle power constraint during the optimization procedure.

Topics & Concepts