基于BP神经网络的机械扩径工艺参数预测方法  

The prediction of processing parameters in mechanical expanding by BP neural network

作　　者：杨艳子[1] 郭宝峰[1] 金淼[1] 

机构地区：[1]燕山大学机械工程学院,秦皇岛066004

出　　处：《塑性工程学报》2008年第3期147-151,共5页Journal of Plasticity Engineering

基　　金：国家自然科学基金资助项目(50475080);河北省自然科学基金资助项目(E2006000246)

摘　　要：文章给出了一种基于BP神经网络建立管筒形零件机械扩径工艺参数与成形精度控制参数间的映射关系,并将其嵌入遗传算法以实现工艺参数优化的机械扩径工艺参数预测方法。所涉及的工艺参数包括扩径率、管坯横断面圆度和模具外径与制品内径之比;成形精度控制参数包括制品外径及其横断面圆度误差。该方法能够很好地预测材质为X52、规格为(406mm～720mm)×9mm的管线钢管机械扩径的工艺参数,并给出一个满足其成形精度要求的最佳工艺参数组合。This paper presented a forecasting method for mechanical expanding processing parameters which optimized by genetic algorithm with BP neural network which mapped the relationship between the mechanical expanding processing parameters of tube and the control precision parameters of forming. The involved processing parameters were ratio of expanding, roundness of steel tube and radius of die, and the forming precision of product including dimensional precision and shape precision. The optimization parameters of products with material X52, specifications （406mm～720mm） × 9mm were forecasted through this method.

关 键 词：机械扩径 有限元模拟 BP神经网络 遗传算法 正交实验法 

分 类 号：TG306[金属学及工艺—金属压力加工]