Comparative optimization of wire-cut EDM parameter for enhancing surface finish and machining time on stainless steel: a machine learning, genetic algorithms, teaching–learning-based optimization, and multi-objective Jaya approach

• Published in: International journal of advanced manufacturing technology Vol. 137; no. 9; pp. 5339 - 5362
• Main Authors: Siyoum, Yitayal Belew, Kindie, Fikir Gashaw, Gebeyehu, Mebratu Assefa, Chanie, Sewale Enyew, Yeshiwas, Teshager Awoke, Zelalem, Yilkal Azene
• Format: Journal Article
• Language: English
• Published: London Springer London 01.04.2025; Springer Nature B.V
• Subjects: Advanced manufacturing technologies; Algorithms; Application; Artificial neural networks; CAE) and Design; Computer-Aided Engineering (CAD; Electric discharge machining; Engineering; Feed rate; Genetic algorithms; Industrial and Production Engineering; Machine learning; Mechanical Engineering; Media Management; Multiple objective analysis; Optimization; Parameter identification; Parameter robustness; Stainless steel; Stainless steels; Surface finish; Surface properties; Wire; TLBO; Surface roughness; Artificial neural network; Parameter optimization; Genetic algorithm; MO-Jaya algorithm
• ISSN: 0268-3768; 1433-3015
• DOI: 10.1007/s00170-025-15450-w

This study focuses on optimizing wire electric discharge machining (WEDM) parameters to enhance surface finish (Ra) and reduce machining time (Mt) for stainless steel. The performance of three algorithms, including the genetic algorithm (GA), teaching–learning-based optimization (TLBO), and multi-objective Jaya (MO-Jaya) algorithms, was compared to identify the most efficient parameter tuning methods. The experimental array was designed using a Taguchi L9 matrix, utilizing controlled parameters including pulse-off time (Toff), peak current (Ip), pulse-on time (Ton), and wire feed rate (WFR). The experimental results were validated by artificial neural network modeling and found a 1.97% error and a mean squared error of 0.113. The optimal parameters determined by MO-Jaya were Toff of 14.29 µs, Ip of 2.438 A, Ton of 8.391 µs, and WFR of 21.274 mm/s, achieving a Ra of 2.193 µm and Mt of 183.469 s. MO-Jaya confirmed greater performance over TLBO, reducing Ra by 1.65% and Mt by 1.69%. Confirmatory tests further improved, enhancing Ra by 1.57% and Mt by 1.17%. The integration of machine learning models provided accurate predictions of surface quality and machining efficiency, while GA, TLBO, and MO-Jaya offered robust frameworks for parameter optimization. Comparative analysis revealed significant improvements in surface finish and reduced machining time using these approaches. These findings contribute to the precision manufacturing of stainless-steel parts and establish a methodological framework for multi-objective optimization in advanced machining processes.