张德海,付亮,李艳芹,等.基于GA–BP神经网络的汽车内饰板虚拟制造研究[J].精密成形工程,2022,14(9):41-49.
ZHANG De-hai,FU Liang,LI Yan-qin,et al.Virtual Manufacturing of Automotive Interior Trim Panels Based on GA-BP Neural Network[J].Journal of Netshape Forming Engineering,2022,14(9):41-49. |
| 基于GA–BP神经网络的汽车内饰板虚拟制造研究 |
| Virtual Manufacturing of Automotive Interior Trim Panels Based on GA-BP Neural Network |
| |
| DOI：10.3969/j.issn.1674-6457.2022.09.006 |
| 中文关键词: DYNAFORM GA–BP神经网络 内饰板 |
| 英文关键词: DYNAFORM GA-BP neural network interior panel |
| 基金项目:江苏省盐城市“515”创新领军人才项目(盐委[2020]40号);河南省科技攻关项目(202102210087);郑州市科技局产学研项目(郑科函[2020]3号) |
|
| 摘要点击次数: 1148 |
| 全文下载次数: 346 |
| 中文摘要: |
| 目的 以某汽车内饰板为研究对象进行虚拟制造，以提前得到相对准确的工艺参数并减少成形缺陷的产生。方法 研究了工艺参数对产品拉延成形质量的影响，并确定了拉丁超立方抽样区间，在抽样区间内抽取60组样本数据，以最大减薄率为目标值，以前50组样本数据为测试集、后10组样本数据为预测集，使用基于GA–BP神经网络的遗传算法得到最优工艺参数，并将其代入有限元分析软件DYNAFORM中进行虚拟制造。结果 训练后GA–BP模型的预测值与期望值最大误差为0.299 7%，最大预测误差率为1.747 38%;遗传算法预测的最大减薄率为16.548%，虚拟制造得到的减薄率为16.167%，虚拟制造值与预测值的大小仅相差0.318%，仿真误差的误差率为2.36%。结论 虚拟制造结合先进算法的优化方法可以指导后续生产。 |
| 英文摘要: |
| The work aims to conduct virtual manufacturing with automotive interior panel as the research object to obtain relatively accurate process parameters in advance and reduce the occurrence of forming defects. Firstly, the effects of process parameters on the quality of product drawing was studied, and the Latin hypercube sampling interval was determined; secondly, 60 sets of sample data were drawn in the sampling interval, and the maximum thinning rate was used as the target value, the first 50 sets of sample data were used as the test set and the last 10 sets of sample data were taken as expected set. Genetic algorithm based on GA-BP neural network was used to obtain the optimal process parameters, which were substituted into DYNAFORM for virtual manufacturing. After training, the maximum error between the test value and the expected value of the GA-BP model was 0.299 7%, and the maximum prediction error rate was 1.747 38%; the maximum thinning rate predicted by the genetic algorithm was 16.548%, and the thinning rate obtained by virtual manufacturing was 16.167%. The difference between the virtual manufacturing value and the predicted value was only 0.318%, and the error rate of the simulation error was 2.36%. The optimization method of virtual manufacturing combined with advanced algorithms can guide subsequent production. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
渝公网安备 50010702501719号