Development of a prediction method for the hyper-elastic material model coefficient through the indentation test and machine learning

In this paper, a hyper-elastic model coefficients prediction algorithm is developed to simplify the experiment to derive the hyper-elastic model coefficients needed for nonlinear finite element analysis (FEA). In the simulations, the correlation between the hyper-elastic model coefficients and the selected measurement data is analyzed through the replicate simulation. A predictive flow graph using TensorFlow is obtained using the acquired data and machine learning techniques. Using these predictive flow graphs, the random hyper-elastic model coefficients are predicted. In addition, the model coefficients of real hyper-elastic materials are predicted using the developed algorithm. Although the accuracy of the prediction is decreased, the model coefficient prediction techniques using manipulator and machine learning algorithms show great potential. An improvement to the pressure test will be attempted in the future to increase the probability of the measuring field.