LSTM Neural Network based Tensile Stress Prediction of Rubber Streching

Abstract

To explore the effective information contained in mass data and improve the accuracy of stress prediction under low strain rate, a stress prediction method based on a hybrid model of convolutional neural network (CNN) and long short-term memory (LSTM) network is proposed for the temporal characteristics and non-linearity of stress data. Massive historical stress data and strain data are constructed as continuous features according to the time sliding window as input. Firstly, feature vectors are extracted by CNN, constructed in the manner of sequence and used as input data of LSTM network. Then the LSTM network is employed to predict the stress. Stress data obtained in the process of rubber stretching are divided into two parts: training data and test data. The model is trained by training data and test data are used for validation of the proposed model. Experimental results show that the proposed prediction method has higher prediction accuracy than the standard LSTM network.