基于神经网络的法兰轴冷挤压成型工艺优化

Process Optimization of Cold Extrusion Process of Flange Shaft Based on Neural Network

为研究冷挤压成型工艺参数对法兰轴成型质量的影响,解决法兰轴成型时法兰盘填充不满和阶梯轴肩的折叠缺陷问题,课题组设计3因素3水平正交实验进行研究。采用3层拓扑结构,以凸模挤压速度、模具与坯料之间的摩擦因数、阶梯轴处圆角半径为输入层神经元,以折叠角和成型载荷为输出层神经元构建法兰轴冷挤压成型工艺优化神经网络模型。研究结果表明该模型的预测性能较好,精度较高。通过生产验证出优化后的工艺方案可有效解决法兰轴充填不满和折叠缺陷,为解决多变量多响应复杂的多元非线性工程问题提供了参考。

In order to study the influence of cold extrusion process parameters on the forming quality of flange shaft and solve the problems of unsatisfactory filling of flange and folding defects of stepped shaft during the forming of flange shaft,a three-factors and three-levels orthogonal experiment was designed and a three-layers topology was used to construct.It was a bible network model for the optimization of cold extrusion process of flange shaft by taking the extrusion speed of punch,friction coefficient between mold and blank and the radius of rounded corner at the step axis as input layer neurons and folding angle and the forming load as output layer neurons.The model was trained by"normalization method",and the model was tested by prediction.The results show that the model has better prediction performance and higher accuracy.The production verification proves that the optimized process can effectively solve the problems of unsatisfactory filling of flange and folding defects,and provides a reference for solving the multivariate and multi-response complex nonlinear engineering problems.