Turning: Autonomous process set-up through Bayesian optimization and Gaussian process models

Abstract

In turning, process parameters such as feed rate and cutting speed are normally selected by an operator and adapted depending on the process requirements. In this study, autonomous process set-up of the turning process is investigated based on Bayesian optimization and Gaussian process models. Unconstrained Bayesian optimization is compared to an implementation with added constraint with simulated machining parameters. It is shown that constraint Bayesian optimization has a better performance in the turning process set-up. The method is suitable for measurements with uncertainty and typically results in reduced number of experiments.