Prediction of behavior of shape memory polymer and reverse engineering

Abstract

The most important feature of 4D printing is that the shape changes with time. The shape memory polymer is a material mainly used in 4D printing and has a property of returning to the original shape memorized in response to the temperature stimulus. When a Shape memory polymer is made of components of a specific shape, a mechanism of linear, bending, and twisting can be imparted. However, the behavior of shape memory polymer is only verified by experimental methods. Although there have been attempts to model the behavior of shape memory polymer using mathematical equations, there has been a limit to more experiments to obtain the parameter values of the modeling. In this study, we have implemented a model to predict the behavior of shape memory polymer using DNN (Deep neural network) technique, which is widely used for prediction. In addition, we have implemented a reverse engineering model that suggests how to design shape memory polymer when specific behavior types are given. For the dataset collection, we experimented the behavior of the shape memory polymer of various combinations, analyzed it using image analysis program, and constructed the database. The performance of the two models was evaluated by using the evaluation method such as the coefficient of determination, and it was shown that the behavior of the shape memory polymer can be predicted and reversed engineering using both models without direct experiment.