A life prediction method of mechanical structures based on the phase field method and neural network

Abstract

In this paper, a new method which combines the back-propagation artificial neural network and phase field method is proposed to predict the residual useful life of mechanical structures. Firstly, the phase field model of brittle fracture is constructed by phase-field method, and the data set about the corresponding strain value and lifespan of the mechanical structure is obtained. Then, the data set is employed as the training samples by back-propagation artificial neural network to obtain the lifespan prediction mathematical model of mechanical structure. With the help of prediction model, only the corresponding structure strain value can be used to predict the remaining life of the structure. Finally, the prediction performance of the proposed method is evaluated by numerical examples. The results show that the proposed method combining the phase-field method and neural network technology for predicting the remaining life of the structure has high accuracy.