基于人工神经网络的Cu与Mg低质量比Al-Cu-Mg合金时效强化预测模型

来源期刊:中南大学学报(自然科学版)2009年第3期

论文作者:侯延辉 刘志义 邓才智 刘延斌 马飞跃

文章页码:657 - 662

关键词:时效;Levenberg-Marquardt算法;神经网络;析出相;模型

Key words:ageing; Levenberg-Marquardt algorithm; neural network; precipitation; modeling

摘    要:通过硬度检测和透射电镜(TEM)观察,研究低Cu/Mg质量比Al-Cu-Mg合金时效强化机理,建立神经网络预测模型,使其在实验条件范围内对时效力学性能进行有效预测。在实验基础上,采用Levenberg-Marquardt算法训练神经网络,建立以时效温度与时间为输入参数和硬度为目标函数的函数关系。结果表明:预测值与实验结果吻合较好,并证明了网络的可靠性与泛化能力;当时效温度越高时,达到峰值时效的时间越短,峰值时效的硬度也越大;在160~190 ℃时效温度范围内,合金峰值硬度随时效温度的升高而下降,对应硬度峰值的时效时间缩短。

Abstract: On the basis of experiment, neural network model was established, which could predict the mechanical performance of ageing Al-Cu, -Mg alloy with low mass ratio of Cu to Mg effectively. In order to obtain the relationship between the parameters and the mechanical performance, experiments were carried out in which the ageing temperature and time were input parameters and hardness was output parameter. The function between the input and output parameters were constructed by neural network trained by Levenberg-Marquardt algorithm. The result shows that the model has high precision and good performance, which provides theoretical foundation for further study of the effect rule of ageing parameters on mechanical property and for optimal design of the ageing process.

基金信息:国家“973”重点基础研究发展项目
中南大学国内博士生访学研究项目

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号