ANN方法分析预测聚丙烯材料的力学性能

利用Ｂ－Ｐ人工神经网络（ＡＪＮＮ）对聚丙烯（ＰＰ）的力学性能进行了分析和预测。首先将ＰＰ材料接纯ＰＰ、共混和增韧及填充和增强ＰＰ等进行分类，并根据这些数据的特点建立Ｂ－Ｐ网络，然后用各类ＰＰ材料的组成和力学性能数据对网络进行学习训练，最后用“未知样品”的数据对网络进行验证。结果表明，所建立的网络能反映ＰＰ的力学性能特性，预测有一定的准确性，但不同类别的材料预测准确性不同。

Analysis and Prediction of Mechanical Properties of Polypropylene by Artifical Neural Networks

The mechanical properties of polypropylene (PP) are analyzed and predicted by artifical neural networks (ANN). PPs are classfied as pure, blended, toughened and filled and reinforced PP, and the B_P networks are set up based on the features of different PPs, and then the networks are trained by the composition and mechanical properties of various PPs. At last unknown samples are used to verify the network. The results show that the networks can predict the mechanical properties of PP, but different prediction accuracy is obtained for different PPs.