利用深度神经元网络预测超高分子量聚乙烯纤维蠕变行为

打开文本图片集

中图分类号：TQ342 文献标志码：A DOI：10.13338/j.issn.1006-8341.2025.04.006

Abstract In order to enhance the creep resistance of ultra-high molecular weight polyethylene （UHMWPE） fibers，UHMWPE fibers with varying graphene contents were prepared，and their creep behavior was predicted using a deep neural network （DNN） and the Burgers model. The study systematically investigated the effects of the number of hidden layers in the DNN，the number of neurons in the hidden layers，the types of activation functions，and the algorithms on the prediction accuracy of the model. The results show that the DNN model exhibits excellent performance in predicting the creep behavior of UHMWPE fibers. When the number of hidden layers and neurons were 36 and 167，respectively，the DNN model with ReLU as the activation function and Adagrad as the optimization algorithm has highly consistent prediction results with experimental results， with a minimum mean square error （MSE） of O.17 and a correlation coefficient R2 of 0.972 between prediction and experimental results. Compared to the Burgers model，the DNN model demonstrates higher prediction accuracy during the initial stretching stage （ 0～25s ）and for high graphene contents （≥10% ），indicating that the DNN model is more suitable for predicting the creep behavior of UHMWPE fibers.

Keywordsultra-high molecular weight polyethylene fiber；creep behavior;deep neural network；Burgers model

0 引言

超高分子量聚乙烯（UHMWPE）纤维是一种具有超高比强度 （>35cN/dtex） 和比模量 Φ（>1200cN/ dtex）的高性能纤维材料[1-3]，具有卓越的力学性能、低密度 （0.97g/cm3 ）及优异的耐化学腐蚀性[4-6]，已广泛应用于航空航天、防弹装甲和生物医用领域[7-8]。（剩余13965字）