Contact problem analysis in functionally graded layers with machine learning method

In this study, the solution is realized using contact analysis finite element method (FEM) and Machine Learning techniques in functionally graded layers loaded with two circular punches. The problem consists of two layers of functional grade loaded with two rigid punches of circular geometry. The external loads P and Q were transferred to the sheet via two rigid punches. All surfaces are considered actionless. The finite element model of functionally graded layers was created using the ANSYS package program, and a 2-dimensional problem analysis was performed. The contact lengths obtained from both punches were transferred to the Python environment and taught to the computer. By means of Machine Learning algorithms suitable for the data, the contact lengths were estimated without the need for an additional solution. Linear Regression, Random Forest and Nearest Neighbor Algorithm from Machine Learning methods were used for contact length estimation. The estimation of contact lengths aims to obtain results in a short time and with a low error rate with the Machine Learning method. In addition, the effects of parameters such as distances between punches, loads, stiffness parameters, and punch radius on contact zones, initial separation loads and distances, normal stresses, stresses through depth, and contact stresses were investigated. The findings are presented in tables and graphics.