锻压技术

基于BP神经网络的条带刚凸特征回弹预测

英文标题：Springback prediction on rigidity and convexity charateristices of strip based on BP neural network

作者：冯斌毛建中胡晖

单位：湖南大学

分类号：TG386.1

出版年,卷(期):页码：2020,45(3):20-26

摘要：

为了研究核燃料组件格架的条带刚凸特征的回弹量与压边力、冲压速度、凸凹模间隙、摩擦系数等冲压工艺参数之间的关系，首先，获取包含50个GA拉丁超立方抽样的数据点以及10个随机抽样的数据点的数据集，前者作为训练集、后者作为测试集。将前者输入到BP神经网络进行训练，后者验证训练模型的精度。最后，通过响应面图研究各因素之间的交互作用以及各因素的敏感程度。结果表明：BP神经网络能够有效预测刚凸回弹量与冲压工艺参数之间的关系，相对于其他因素，压边力对回弹量的影响特别明显，冲压速度对回弹量的影响不明显，但与凸凹模间隙和摩擦系数有明显的交互作用。

In order to study the relationship between the springback amount of the rigidity and convexity charateristices of strip for the nuclear fuel assembly grid and the stamping process parameters, such as blank holder force, stamping speed, clearance between punch and die and friction coefficient. Firstly, a data set containing fifty GA Latin hypercube sampled data points and ten randomly sampled data points were obtained with the former as the training set and the latter as the test set. Then, the former was inputted to the BP neural network for training, and the latter verified the accuracy of the training model. Finally, the interaction among various factors and the sensitivity of various factors were studied by response surface graphs. The results show that BP neural network effectively predicts the relationship between the springback amount of rigid and convexity and the stamping process parameters, and compared with other factors, the influence of the blank holder force on the springback amount is particularly obvious. However, the impact of the stamping speed on the springback amount is not significant, and it has obvious interaction with the clearance between punch and die and the friction coefficient.

基金项目：

国家科技重大专项子课题（761215007）

冯斌（1995-），男，硕士研究生 E-mail：fengb34567@163.com 通讯作者：胡晖（1969-），男，硕士，高级工程师 E-mail：huhui@hnu.edu.cn

参考文献：

［1］潘金勇. 锆合金薄板成形极限线的理论预测与数值模拟
［D］.长沙：湖南大学,2018.

Pan J Y. Theoretical Research and Numerical Simulation of Forming Limit Line for Zirconium Alloy Sheet
［D］.Changsha: Hunan University,2018.

［2］何廷一,田鑫萃,李胜男,等.基于蜂群算法改进的BP神经网络风电功率预测
［J］.电力科学与技术学报,2018,33(4):22-28.

He T Y, Tian X C, Li S N,et al. Improved BP neural network based on artificial bee colony algorithm for wind power prediction
［J］.Journal of Electric Power Science and Technology,2018,33(4):22-28.

［3］文怀兴,张斌,杨新妮,等.基于BP神经网络的单点渐进成形回弹预测
［J］.热加工工艺,2018,47(15):109-112.

Wen H X, Zhang B, Yang X N, et al. Springback prediction of single point incremental forming based on BP neural network
［J］.Hot Working Technology,2018,47(15):109-112.

［4］张涛,樊文欣,朱芹,等.基于BP神经网络的连杆衬套强力旋压回弹量预测
［J］.特种铸造及有色合金,2017,37(4):380-382.

Zhang T, Fan W X, Zhu Q, et al. Prediction of springback of conn-ecting rod bushing based on BP neural network
［J］.Special Casting & Nonferous Alloys,2017,37(4):380-382.

［5］王茁.新开轨道交通城市的客流预测与方法分析
［J］.上海工程技术大学学报,2018,32(4):346-351.

Wang Z. Passenger volume flow prediction and method analysis of new rail transit cities
［J］.Journal of Shanghai University of Engineering Science,2018,32(4):346-351.

［6］田娥,孙建东,刘自萍,等.基于BP神经网络的弯管机回弹量预测
［J］.现代制造工程,2016, (3):70-73.

Tian E, Sun J D, Liu Z P,et al. Bending machine springback prediction based on BP neural network
［J］.Modern Manufacturing Engineer,2016, (3):70-73.

［7］付泽. 典型汽车用板变形滞后回弹的试验研究及有限元分析
［D］.北京：北京理工大学,2016.

Fu Z. Experimental Study and Finite Element Analysis on Time-dependent Springbak of Typical Automotive Sheets under Deforming
［D］.Beijing: Beijng Institute of Technology,2016.

［8］王智. 基于灰色理论和神经网络的弯曲回弹预测研究
［D］.成都：西南交通大学,2013.

Wang Z. Research on the Prediction of The Bending Springback Based on Grey Theory and Neural Network Model
［D］.Chengdu：Southwest Jiaotong University,2013.

［9］王振,白杨,郝长利,等.基于BP神经网络的曲轴润滑特性全局优化
［J］.小型内燃机与车辆技术,2018,47(4):42-48.

Wang Z, Bai Y, Hao C L, et al. Global optimization of crankshaft l-ubrication characteristics based on BP neural network
［J］.Small Internal Combustion Engine and Vehicle Technique,2018,47(4):42-48.

［10］汪倩. 基于Dynaform软件的高强钢矩形管绕弯成形模拟研究
［D］.兰州：兰州交通大学,2018.

Wang Q. A Simulation Study of Rotary Draw Bending for Rectangular Section Tubes of High Strength Steel Based on Dynaform
［D］. Lanzhou：Lanzhou Jiaotong University,2018.

［11］何平. 基于有限元分析的特种条带冲压模具数字化设计研究
［D］.长沙：湖南大学,2018.

He P. Research on Digital Design of Stamping Die for Special Strip Based on Finite Element Analysis
［D］. Changsha: Hunan University,2018.

［12］王小川，史峰，郁磊, 等. Matlab神经网络43个案例分析
［M］.北京：北京航空航天大学出版，2013.

Wang X C, Shi F, Yu L, et al. 43 Case Analysis of Matlab Neural Network
［M］.Beijing:Behang University Press,2013.

［13］Zheng D, Qian Z, Liu Y, et al. Prediction and sensitivity analysis of long-term skid resistance of epoxy asphalt mixture based on GA-BP neural network
［J］. Construction and Building Materials, 2018, 158: 614-623.

服务与反馈：

【文章下载】【加入收藏】