基于BP神经网络的压铸成型工艺参数的模拟与优化  

Optimization and Application of Process Parameters for Die Casting Process Based on Neural Network of BP

作　　者：张庆[1] 李晓棠[1] 于峰[1] 陈泽[1] 

机构地区：[1]贵州大学机械工程学院,贵州贵阳550003

出　　处：《铸造设备与工艺》2011年第5期35-37,47,共4页Foundry Equipment & Technology

基　　金：贵州省科学技术基金资助项目(黔科合J字[2006]2108号)

摘　　要：利用数值模拟、BP神经网络和正交试验相结合的方法对压铸成型的工艺参数进行模拟和优化。并且通过一个简单的实例对该方法的可行性进行了验证。基于BP神经网络对压铸工艺参数及其相对应的铸件最低温度点样本进行训练,得到了工艺参数到铸件温度映射关系的神经网络模型,并验证该模型的准确性。结果表明,神经网络结合正交实验设计方法的优化算法,可以确定出最优的工艺参数组合,缩短了优化工艺参数的时间。By using numerical simulation, the BP neural network and orthogonal test method, the technological parameters of die casting was simulated and optimized,then was proved through a simple example.Based on BP neural network,the die casting technology parameters and corresponding low temperature samples were trained, the neural network model of the mapping relation between the process parameters and the casting temperature was obtained and the accuracy was verified.The results show that neural network combined with the orthogonal experiment design methods of optimization algorithm,can determine the optimal process parameter and shorten the optimization time of process parameters.

关 键 词：压铸成型 数值模拟 正交试验 神经网络 工艺参数 

分 类 号：TG249.2[金属学及工艺—铸造]