基于人工神经网络的HDPE流变性能预测与研究

Prediction and Research of HDPE Rheological Properties Based on Artificial Neural Networks

针对影响高密度聚乙烯(HDPE)流变性能各因素之间的非线性关系建立了一个优化后的BP网络,然后通过实验取得样本数据,进而对网络进行训练,用训练合格后的人工神经网络对高密度聚乙烯在不同温度或不同剪切速率下的剪切应力进行预测,并绘制出预测的流动、流变曲线,最后对预测和实际测得的流动、流变曲线进行了比较分析。

In accordance with the non-linear relations among the factors which affect HDPE (rheological) properties,an optimized Back-Propagation network was established.Then sample data were obtained by experiments.Furthermore,the qualified artifical neural net-works was trained to predict the shearing strength of HDPE under various temperatures or shear rates,and the predicted flowing and rheological curves were plotted.Eventually,comparison and analysis were made on the flowing and rheological curves between the predicted and the actual results.