论文部分内容阅读
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 .本文用较为成熟的方法一例证了有待探索的方法二
In the actual casting production, the use of computer simulation of the entire casting process, as well as pouring gate location and pouring speed casting process parameters such as optimization calculation, you can detect possible defects as soon as possible and the location of defects, thereby shortening the design Cycle to improve the quality of castings and reduce the cost.This paper compares two methods of parameter optimization.Methods1.The numerical simulation results of casting process are obtained by numerical simulation software of casting process.The numerical results are used as the sample data, Cubic spline interpolation method to obtain the input parameters (process parameters) and the output value (filling the end of the cavity when the maximum temperature and the minimum temperature difference) between; and then get the minimum value of the objective function, that is, The most suitable process parameters in the pouring process.Method two, the use of numerical experimental data to train the neural network, and then use genetic algorithm to neural network to establish the function relationship optimization, also can get the most suitable process parameters.Using two The solution obtained by the method is basically the same A mature approach exemplifies the second method to be explored