Scholarlink

<p>The process of Electric Discharge Machining (EDM) for additively manufactured Ti-6Al-4&#xa0;V alloy presents substantial difficulties because of its intricate thermal properties and the requirement to achieve multiple performance objectives. The research investigates how to optimize Direct Metal Laser Sintering (DMLS) Ti-6Al-4&#xa0;V components through two factors, material removal rate and surface roughness. The researchers created their experimental design through a Taguchi L9 orthogonal array, which includes three essential process parameters: current and pulse-on duration and pulse-off duration. The researchers used a hybrid multi-criteria decision-making approach that combines CRITIC and TOPSIS to identify the best parameter settings through the assessment of objective weights and rank order. The best combination of MRR and SR efficiency was achieved with a current of 7&#xa0;A and a pulse-on duration of 300 µs and a pulse-off duration of 90 µs which resulted in a maximum closeness coefficient (CCi) of 0.9807. The system demonstrates performance capabilities that include a maximum material removal rate (MRR) of 0.3884&#xa0;mm³ per minute and a minimum surface roughness (SR) of 2.33 micrometers. Statistical analysis showed that current control affects 62.18% of material removal rate (MRR) while pulse-on time affects 92.56% of surface roughness (SR) which shows their major effects on machining efficiency. The development of supervised machine learning models, which included Decision Tree and Random Forest and AdaBoost, aimed to improve predictive ability through their various learning methods. The AdaBoost model outperformed other models because it achieved a high coefficient of determination R² value which reached approximately 0.998 through its ability to model nonlinear relationships between process parameters and responses. The proposed integrated statistical–MCDM–machine learning framework provides a strong efficient method for EDM process optimization. The system shows high predictive accuracy together with consistent optimization results which demonstrate its ability to control processes in real time and support intelligent manufacturing. These findings show great importance because they help develop better post-processing techniques for additively manufactured titanium components used in aerospace and biomedical fields.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hybrid CRITIC–TOPSIS and machine learning modelling for EDM optimization of DMLS Ti6Al4V alloy

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月16日

查看全文
  • Prithivirajan R.,
  • Parrthipan B. K.,
  • Mohammed Raffic N.,
  • Perumal A.,
  • Senthil Kumar S.,
  • Narayanasamy P.,
  • Jayabal S.,
  • Muthuperumal S.

摘要

The process of Electric Discharge Machining (EDM) for additively manufactured Ti-6Al-4 V alloy presents substantial difficulties because of its intricate thermal properties and the requirement to achieve multiple performance objectives. The research investigates how to optimize Direct Metal Laser Sintering (DMLS) Ti-6Al-4 V components through two factors, material removal rate and surface roughness. The researchers created their experimental design through a Taguchi L9 orthogonal array, which includes three essential process parameters: current and pulse-on duration and pulse-off duration. The researchers used a hybrid multi-criteria decision-making approach that combines CRITIC and TOPSIS to identify the best parameter settings through the assessment of objective weights and rank order. The best combination of MRR and SR efficiency was achieved with a current of 7 A and a pulse-on duration of 300 µs and a pulse-off duration of 90 µs which resulted in a maximum closeness coefficient (CCi) of 0.9807. The system demonstrates performance capabilities that include a maximum material removal rate (MRR) of 0.3884 mm³ per minute and a minimum surface roughness (SR) of 2.33 micrometers. Statistical analysis showed that current control affects 62.18% of material removal rate (MRR) while pulse-on time affects 92.56% of surface roughness (SR) which shows their major effects on machining efficiency. The development of supervised machine learning models, which included Decision Tree and Random Forest and AdaBoost, aimed to improve predictive ability through their various learning methods. The AdaBoost model outperformed other models because it achieved a high coefficient of determination R² value which reached approximately 0.998 through its ability to model nonlinear relationships between process parameters and responses. The proposed integrated statistical–MCDM–machine learning framework provides a strong efficient method for EDM process optimization. The system shows high predictive accuracy together with consistent optimization results which demonstrate its ability to control processes in real time and support intelligent manufacturing. These findings show great importance because they help develop better post-processing techniques for additively manufactured titanium components used in aerospace and biomedical fields.