Analysis of EDM machining parameters for keyway on Ti-6Al-4V alloy and modelling by artificial neural network and regression analysis methods


Creative Commons License

Çakıroğlu R.

SADHANA - ACADEMY PROCEEDINGS IN ENGINEERING SCIENCES, cilt.47, sa.150, ss.2-17, 2022 (SCI-Expanded)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 47 Sayı: 150
  • Basım Tarihi: 2022
  • Doi Numarası: 10.1007/s12046-022-01926-y
  • Dergi Adı: SADHANA - ACADEMY PROCEEDINGS IN ENGINEERING SCIENCES
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Communication Abstracts, INSPEC, Metadex, zbMATH, Civil Engineering Abstracts
  • Sayfa Sayıları: ss.2-17
  • Gazi Üniversitesi Adresli: Evet

Özet

Keyed joints are one of the most widely used methods of motion and power transmission. The high functionality and reliability of the joining elements depend on their resistance to the thermo-mechanical stresses that occur during operation and their high precision manufacturing. In this context, a widespread analysis of keyway opening and machining parameters (discharge current, pulse on time and pulse off time) was carried out according to DIN 6885 standard by a die-sinking electrical discharge machine on Ti-4Al-6V alloy with poor machinability. First, the effects of processing parameters on material removal rate (MRR), tool wear rate (TWR) and surface roughness (SR) were investigated by EDM experiments in a kerosene environment. It was seen that the discharge current and the pulse on time have a significant effect on the processing outputs MRR, TWR and SR. In addition, the processed surface and subsurface formations were comprehensively evaluated with SEM and EDS analyses. By reason of the low thermal conductivity of Ti6Al4V, it has been determined that the depth of the heat affected zone reaches 60 lm. In the second stage, mathematical models were developed for the prediction of processing outputs using artificial neural network (ANN) and regression analysis (RA) methods. When these two methods were compared, it was determined that the modelling results with ANN were closer to the experimental results.