|
摘要：
|
|
由于气门电热镦粗的工艺参数大多数是借助于经验来选择的,某些参数的选择不合理或者参数之间配合不好,都会造成工艺不稳定,从而使生产的成品率下降。并且在成形过程中,电加热和镦粗同时进行,很难建立合理且实用的数学模型。本文利用神经网络具有黑箱特性和非线形映射能力强的特点,提出了一种组合神经网络结构(ANN)来逐步确定气门电热镦粗的工艺参数。以实际生产中的数据作为ANN的学习训练样本,经过训练的网络不仅可以达到描述确定气门电镦成形控制参数的目的,而且还具有一定的预测功能,从而为气门电镦工艺提供了较合理的控制参数。
|
|
The parameters of electrical upsetting process for valve are mostly selected by experience.Therefore,some parameters are unreasonably selected or not matched very well.This may cause process unstable and cause the quality of products descends.During the process,electric heating goes along with forging,and it is really difficult to set up the rational and useful mathematics model.Artificial Neural Networks has the character of black box and strong no-linear mapping capacity.In this article,a kind of combined neural networks is utilized to confirm the parameters gradually.The training sample of ANN is the data from experience of production and the networks are trained in order to attain the object of describing,ascertaining and somehow previewing the parameters of valve electrical upsetting process.Therefore,this method provides rational control parameters for valve electrical upsetting techniques.
|
|
基金项目：
|
|
|
|
作者简介：
|
|
|
|
参考文献：
|
[1]刘天湖,孙友松,肖小亭,等.发动机气门电热镦粗工艺参数分析与计算[J].锻压技术,2002,(5):2 5.
[2]汪国顺,夏巨谌,胡国安,王新云,等.气门电热镦粗工艺的数值模拟[J].塑性工程学报,2004,(2):19 26.
[3]飞思科技产品研发中心.神经网络理论与MATLAB 7实现[M].北京:电子工业出版社,2005.
[4]丛爽.面向MATLAB工具箱的神经网络理论与应用[M].合肥:中国科学技术大学出版社,2003.
[5]Simon Haykin,叶世伟,史忠植.神经网络原理[M].北京:机械工业出版社,2004.
[6]Hansen L K,Salamon P.Neural network ensembles[J].IEEE Trans.Pattern Anal.Mach.Intell.12:993 1001.
[7]Roland Linder,Dawn Dew.The'subsequent artificial neuralnetwork'(SANN)approach might bring more classificatorypower to ANN\based DNA microarray analyses[J].Bioin-formatics Dec12,2004,20:70 74.
|
|
服务与反馈：
|
|
【文章下载】【加入收藏】
|
|
|
|
