铸造充型过程控制参数的智能选择  

Intelligent Selection of Parameters in Casting Filling Process

作　　者：贵刚[1] 苏丽杰[1] 聂义勇[1] 

机构地区：[1]中科院沈阳自动化所研究生院,沈阳110016

出　　处：《计算机工程与应用》2003年第28期46-48,63,共4页Computer Engineering and Applications

基　　金：辽宁省自然科学基金项目(编号:972020)

摘　　要：在压力铸造生产中,很多时候因为浇铸口、排气口位置及压铸力等控制参数选择不合理导致缩孔、冷隔或者气孔等缺陷。浇铸口和排气口位置等工艺参数经过优化计算后可以大大减少这些缺陷。文章利用铸造充型过程数值仿真软件,通过数值计算获得一些有关控制参数的实验结果;然后将数值实验结果作为样本数据,运用Levenberg-Marquardt算法训练神经网络,建立起输入参数(各个工艺参数)和输出值(充型时间和充型结束时型腔内最高温度与最低温度之差)之间的函数关系;再使用遗传算法对该问题寻优。从而得到最合适的浇铸参数。The cast parameters during casting filling stage,such as the nozzle and vent location,play important roles in die casting.For instance,improper nozzle location results in shrinkage,cold shut or air cavity.Optimized these parameters can reduce such defects.In this paper,some numerical experimental results are obtained by ProCAST firstly,then these experimental data are provided for the multi-layer feed-forward network training,which done by Levenberg-Marquardt algorithm in MATLAB application tool box.The function,between the input and output parameters,can be constructed by neural networks.Finally,the optimal problem may be solved by Genetic Algorithm.The most proper cast parameters are gotten.

关 键 词：压力铸造 浇铸参数 神经网络 遗传算法 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]