信息扩散和径向基神经网络的注塑模型建立
详细信息 本馆镜像全文    |  推荐本文 | | 获取馆网全文
摘要
在使用径向基神经网络建立注塑工艺模型时,虽然能够得到较好的模型,但是建模时训练样本数量将会对模型的质量产生较大的影响。本研究对建模所需的原始样本数据首先进行信息扩散处理,然后再使用径向基神经网络建立注塑工艺参数与塑件沉降斑指数之间的模型。从结果上看,在注塑训练样本数量相同的情况下,运用该方法均可以得到优于仅使用普通径向基网络构建的模型。
Although RBF neural network was widely used to uncover the relationship between injection molding process parameters and part quality,which could generate high quality models,the number of training samples was a nonnegligible factor which seriously affected the accuracy of the injection model. In this paper,the information diffusion theory is applied to derive new patterns,which were used to generate sink model with RBF neural network,from original training samples. The results showed that this method could generate better injection model when the training samples were equal.
引文
[1]姜勇道,周华民,傅建,等.基于神经网络技术的注塑成型注射压力和熔体温度预测[J],塑料工业,2003,31(6):27~29.
    [2]陈建松,闵雁,伍乃骐.Taguchi与RBF神经网络在注塑翘曲参数优化中的应用[J].机电工程技术,2008,37(1):89~92.
    [3]哈姆,科斯塔尼克.神经计算原理[M].叶世伟,王海娟,译.北京:机械工业出版社,2007.
    [4]黄崇福.信息扩散原理与计算思维及其在地震工程中的应用[D].北京:北京师范大学,1993.
    [5]刘涛,邵东国,刘武艺.基于信息扩散和BP神经网络的水资源预测[J].武汉大学学报,2007,40(3):31~33,75.
    [6]C.F.Huang.Information diffusion techniques and small sample problem[J].International Journal of Information Technology and De-cision Making,2002,1(2):229~249.
    [7]何书,王家鼎,王欢,等.基于信息扩散和BP网络的黄土边坡稳定性分析[J].西北大学学报,2008,38(6):983~988.
    [8]Chongfu Huang,Claudio Moraga.A diffusion-neural-network for learning from small samples[J].International Journal of Approximate Reasoning,2004,35(2):137~161.

版权所有:© 2023 中国地质图书馆 中国地质调查局地学文献中心