Modeling and estimation of thrust force, torque, and surface roughness in indexable drilling of AA6061-T651 with Taguchi, ANN, and ANFIS

Creative Commons License

Akdulum A., Kayir Y.

SADHANA - ACADEMY PROCEEDINGS IN ENGINEERING SCIENCES, cilt.48, sa.143, ss.1-16, 2023 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 48 Sayı: 143
  • Basım Tarihi: 2023
  • Doi Numarası: 10.1007/s12046-023-02209-w
  • 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.1-16
  • Gazi Üniversitesi Adresli: Evet

Özet

In this study, the effect of drilling 6061-T651 aluminum alloy with different lengths of indexable insert drills, called U drills, on thrust force, torque, and surface roughness was investigated. As input parameters, length-to-diameter ratio, feed rate, and cutting speed were chosen for experimental works. The optimum values of the test parameters were determined by the ratio of signal to noise. In addition, output responses were modeled and compared with Taguchi, artificial neural networks (ANN), and the adaptive neuro-fuzzy inference system (ANFIS) methods. Both the experimental results and the signal-to-noise ratios derived from the experimental results were employed in the modeling process. The models with the highest accuracy were created using ANN when the predicted results from the models were compared to the experimental findings. The MAPE values of the ANN model created with the SN ratio were obtained as 0.18% for thrust force, 0.17% for torque, and 1.79% for surface roughness. Converting the output responses to SN ratios and using them in the models enabled the estimation of thrust, torque, and surface roughness with less error and satisfactory reliability. With the method proposed in this study, output responses according to input variables can be predicted with higher precision, resulting in the efficiency and reliability required by the industry.