铸造充型过程参数优选的两种方法比较  

Comparison Between Two Optimal Selections of Parameters in Casting Filling Process

作　　者：贵刚[1,2] 于军[1,2] 李富明[1,2] 聂义勇[1,2] 

机构地区：[1]中国科学院沈阳自动化所,辽宁沈阳110016 [2]中国科学院研究生院,北京100039

出　　处：《小型微型计算机系统》2004年第4期685-689,共5页Journal of Chinese Computer Systems

基　　金：辽宁省自然科学基金项目 ( 972 0 2 0 )资助

摘　　要：在实际铸造生产之前 ,利用计算机对整个铸造过程进行数值模拟 ,以及对浇注口位置和浇注速度等浇注过程的工艺参数进行优化计算 ,可以尽早发现可能产生的缺陷以及缺陷发生的位置 ,从而缩短设计周期 ,提高铸件质量 ,降低成本 .本文比较了两类参数优选方法 .方法一 ,利用铸造充型过程数值仿真软件 ,获得一些有关控制参数的数值实验结果 ;然后将数值实验结果作为样本数据 ,运用三次样条插值方法进行插值 ,获得输入参数 (工艺参数 )和输出值 (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 ;进而得到目标函数的最小值 ,即得到最合适的浇注过程中的工艺参数 .方法二 ,使用数值实验数据对神经网络进行训练 ,再采用遗传算法对神经网络建立起的函数关系进行寻优 ,亦可得到最合适的工艺参数 .使用两种方法得到的解基本一致 .Before the mould being manufactured, casting filling simulation and process parameters, such as the nozzle location and casting speed,optimization during filling stage, will predict the possible defects and their locations,therefore will shorten the design time, upgrade the product quality, and reduce the cost. Two optimal selections are compared in this paper. One method, some numerical experimental results are obtained by ProCAST firstly, and then these experimental data are provided for the interpolation of cubic spline. The function y=(x1,x2), between the input and output parameters, can be constructed. Then the minimum of y can be found easily, that is to say, the most proper cast parameters are gotten. The other method, neural network is used to construct the function between the input and output parameters, which trained by the simulation data, then genetic algorithm is used to search the optimal value of the function. And the results of these two ways are in agreement. Using the first matured method, this paper demonstrates the second method to explore.

关 键 词：铸造 工艺参数 三次样条 神经网络 遗传算法 

分 类 号：TG21[金属学及工艺—铸造]