Machine learning algorithms based advanced optimization of wire-EDM parameters: an experimental investigation into titanium alloy

Abstract

This paper aims to optimize the various process inputs in the wire electrical discharge machining of Ti-alloy, Ti-6Al-7Nb, using machine learning (ML) algorithms. The present investigations report on examining the behavior of response factors with changes to the machine-controllable parameters. Pulse on time (Ton), pulse off time (Toff), peak current (Ip), and servo voltage (Sv) are considered as process input controllable parameters along with fixed factors, and their effects on surface quality (Sr) and machining rate in terms of material removed (MRR) are investigated. A full factorial design is utilized for the experimental runs. The 2-D and 3-D graphs depict the behavioral study of the response factor in conjunction with machining input variations. The novelty of this work is the WEDM machining of Ti-6Al-7Nb and the machine learning-assisted parameter optimization. Using the genetic algorithm and the teacher learning-based optimization (TLBO) techniques, single- and multi-objective machine learning-based optimization of investigated responses was carried out. The parametric combination of inputs obtained for optimization of multiple responses (MRR and Sr) is Ton 114 mu, Toff 60 mu, Ip 80 A, and Sv 80; or Ton 114 mu, Toff 60 mu, Ip 140 A, and Sv 80 V, as per GA and TLBO, respectively. Graphical abstract: This paper aims to optimize the various process inputs in the wire electrical discharge machining of Ti-alloy, Ti-6Al-7Nb, using machine learning (ML) algorithms. The present investigations report on examining the behavior of response factors with changes to the machine-controllable parameters. Pulse on time (Ton), pulse off time (Toff), peak current (Ip), and servo voltage (Sv) are considered as process input controllable parameters along with fixed factors, and their effects on surface quality (Sr) and machining rate in terms of material removed (MRR) are investigated. A full factorial design is utilized for the experimental runs. The 2-D and 3-D graphs depict the behavioral study of the response factor in conjunction with machining input variations. The novelty of this work is the WEDM machining of Ti-6Al-7Nb and the machine learning-assisted parameter optimization. Using the genetic algorithm and the teacher learning-based optimization (TLBO) techniques, single- and multi-objective machine learning-based optimization of investigated responses was carried out. The parametric combination of inputs obtained for optimization of multiple responses (MRR and rS) is Ton 114 mu, Toff 60 mu, Ip 80 A, and vS 80; or Ton 114 mu, Toff 60 mu, Ip 140 A, and vS 80 V, as per GA and TLBO, respectively. (Figure presented.) © The Author(s), under exclusive licence to Springer-Verlag France SAS, part of Springer Nature 2023.