Prediction of Life of Compound Die Punch Using Machine Learning

Salunkhe S. S., Vasarla P., Naranje V., Dharmateja M.

International Conference on Computational Intelligence and Knowledge Economy (ICCIKE), Dubai, Birleşik Arap Emirlikleri, 11 - 12 Aralık 2019, ss.647-650, (Tam Metin Bildiri) identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/iccike47802.2019.9004375
  • Basıldığı Şehir: Dubai
  • Basıldığı Ülke: Birleşik Arap Emirlikleri
  • Sayfa Sayıları: ss.647-650
  • Gazi Üniversitesi Adresli: Hayır

Özet

In this paper, machine learning (ML) model is developed for prediction of life of punches of compound die. The life of punches of compound die depends on sheet thickens, sheet material and perimeter area of punch. Three-dimensional (3D) explicit finite element (FE) model is used to determination of maximum and minimum principal stresses of punch. Based on FE model results, machine learning model is developed using linear regression, XGBoost and Support Vector Regression algorithms for prediction of life of punch. The predictive model are built using data from a FEM (maximum and minimum principal stress) software. The results revealed that the proposed machine learning model is higher prediction accuracy more than 99.5% compared to artificial neural network (ANN) and adaptive neuro fuzzy inference systems (ANFIS) model. A comparative results is made between these three models (ML, ANN and ANFIS) and the output showed the superiority between machine learning, ANN and ANFIS model. The approach presented is tested and validated using data from FE on different compound die punches.