摘要
在通过实验获得大量样本数据的基础上,利用Matlab语言与VB语言混合编程,开发了聚合物挤出胀大比预测软件。并利用BP网络建立不同口模挤出条件下挤出胀大比与剪切应力之间的关系来实现人工神经网络的训练。经过反复训练满意后,即可输入一系列剪切应力值来预测挤出胀大比。结果表明:网络的训练精度可控制在10-2以下,预测点与实测点吻合得较好,实现了由理论成果向应用技术的转化。
Based on the data acquired by the experiment, prediction software of polymer die swell ratio was developed by mixing programming technique with Matlab language and VB language. And BP network was used to establish the relationship of die swell ratio to shearing stress under different die extrusion conditions to achieve training of artificial neural network. After repeated training reached the satisfied result, the die swell ratio was predicted by inputting a series of shearing stress value. The results showed that training accuracy of network could be controlled within 10-2. Forecasts point with the measured agree well with those points. The transformation of theoretical resuhs into application technology was realized.
出处
《塑料》
CAS
CSCD
北大核心
2008年第2期100-102,28,共4页
Plastics
关键词
人工神经网络
挤出胀大比
预测
artificial neural network
die swell ratio
prediction
