摘要
建立了薄板V形自由折弯的有限元模型,利用拉丁超立方抽样法确定了影响折弯回弹的主要因素组合方案,利用有限元模拟的回弹角度作为样本数据,对用于预测折弯回弹角度的BP神经网络模型进行了训练,采用平均影响值法对影响回弹角度的因素进行了敏感性分析,并利用蒙特卡罗法对折弯过程中的不可控因素进行了模拟,研究了折弯回弹角度的分布规律与变化区间。结果表明,建立的BP神经网络模型精度较高,可以准确预测V形自由折弯回弹角度;受不可控因素影响,折弯回弹角度基本符合正态分布,在置信系数为99.73%时,角度的波动达19.87%,表明不可控因素的波动对折弯回弹角度具有显著影响。
The finite element model of V-shaped free bending of thin plate was established.The Latin hypercube sampling method was used to determine the combination scheme of the main factors affecting the bending springback.Taking the springback angle simulated by finite element as the sample data,the BP neural network model used to predict the bending springback angle was trained,and the sensitivity of the factors affecting the springback angle was analyzed by mean impact value(MIV)method.The uncontrollable factors in bending process were simulated by Monte Carlo method,and the distribution law and variation range of bending springback angle were studied.The results show that the accuracy of the established BP neural network model is high,it can effectively predict the springback angle of V-shaped free bending.Affected by uncontrollable factors,the bending springback angle basically conforms to the normal distribution.When the confidence coefficient is 99.73%,the angle fluctuation reaches 19.87%,indicating that the fluctuation of uncontrollable factors has significant impact on bending springback angle.
作者
周玉甲
胡峰松
袁佳健
ZHOU Yu-jia;HU Feng-song;YUAN Jia-jian(College of Mechanical and Electrical Engineering,Hunan Communication Polytechnic,Changsha 410132,China;Institute of Information Science and Technology,Hunan University,Changsha 410082,China)
出处
《塑性工程学报》
CAS
CSCD
北大核心
2023年第11期21-27,共7页
Journal of Plasticity Engineering
基金
湖南省教育厅科学研究项目(22C0951)。
关键词
自由折弯
回弹
神经网络
敏感性
不可控因素
free bending
springback
neural network
sensitivity
uncontrollable factors