MSc.Thesis Defense:Gamze Balçık

IDENTIFICATION OF OPTIMUM MILLING PARAMETERS

THROUGH MACHINE LEARNING

Gamze Balçık
Industrial Engineering, MSc. Thesis, 2024

Thesis Jury

Prof. Erhan Budak (Thesis Advisor)

Assoc. Prof. Lutfi Taner Tunç

Assoc. Prof. Umut Karaguzel

Date & Time: 23rd July, 2024 –  9.30 – 10.30 AM

Place: G035

Keywords : Milling, Bayesian Optimization, Machine Learning

Abstract

Milling operations are commonly utilized in many industries. Productivity rate becomes prominent in the industries such as automotive due to the need for high-volume manufacturing or the requirement to produce large die casts, whereas the aerospace and electronics industry must focus on precise manufacturing that does not exceed tolerance bands. This condition results in different types of optimization equations such as maximizing material removal rate with respect to machining center limits or minimizing the tool deflection and chatter risk to achieve conforming parts. Both optimizations will indicate a major effect on the unit cost of the product, hence they should describe the trade-off between the machining time and tool cost and help the selection of the optimum cutting parameters and tool dimensions.

The increase in AI implementations and their promising accuracy levels were the main reasons to choose the supervised machine learning (ML) to investigate the optimum solution. In this thesis, Titanium alloy (Ti-6-4) workpiece material cutting process with carbide tool has been simulated for many different cutting tool and process parameter scenarios to calculate cutting forces, chatter status, surface form errors, machining time, tool life and tool breakage. Following the data preparation step, Gaussian Process Regression model has been computed for the optimization step with Bayesian approach.