基于神经网络的GH4169螺栓圆角滚压强化性能预测  

Prediction of Fillet Rolling Strengthening Performance of GH4169 Bolt Based on Neural Network

作　　者：石大鹏 马震宇 何坤 胡庆宽 付建建 巴豪强 徐锋[3] Shi Dapeng;Ma Zhenyu;He Kun;Hu Qingkuan;Fu Jianjian;Ba Haoqiang;Xu Feng(Henan Aerospace Precision Products Co.,Ltd.,Xinyang,Henan 464000,China;不详)

机构地区：[1]河南航天精工制造有限公司,河南省信阳市464000 [2]河南省紧固连接技术重点实验室 [3]南京航空航天大学

出　　处：《工具技术》2025年第2期114-119,共6页Tool Engineering

摘　　要：为快速预测GH4169螺栓圆角滚压强化后的性能,本文提出一种基于神经网络的螺栓圆角滚压性能预测方法。采用ABAQUS结合Fe-safe软件进行螺栓滚压性能仿真计算,并通过试验进行模型有效性验证。将50组试验数据和50组仿真数据作为数据集进行神经网络预测模型训练,获取预测精度高和泛化能力强的神经网络模型。预测模型的输入参数为滚压力、滚压转速和滚压时间,输出参数为螺栓圆角处的3个变形量、残余应力和疲劳寿命,神经网络结构为3-25-25-25-25-5。研究表明,预测模型具有很高的预测精度和很强的泛化能力,对5个输出参数预测的平均绝对百分比误差均小于3%,且对重要输出(变形量B、残余应力和疲劳寿命)预测的决定系数R^(2)均高于0.99。To quickly predict the performance of GH4169 bolt fillet after rolling strengthening,a prediction method of bolt fillet rolling performance based on neural network is proposed in this paper.In this paper,ABAQUS combined with Fe-safe software is used to simulate the rolling performance of bolts,and the validity of the model is verified by experiments.Then,50 sets of test data and 50 sets of simulation data are used as data sets to train the neural network prediction model,and the neural network model with high prediction accuracy and strong generalization ability is obtained.The input parameters of the prediction model are rolling pressure,rolling speed and rolling time.The output parameters are deformation A,deformation B,deformation C,residual stress and fatigue life.The neural network structure is 3-25-25-25-25-5.The research shows that the prediction model has high prediction accuracy and strong generalization ability.The average absolute percentage error of the five output parameters is less than 3%,and the determination coefficient R^(2)of the important output(deformation B,residual stress and fatigue life)is higher than 0.99.

关 键 词：GH4169螺栓 圆角滚压 仿真分析 神经网络 性能预测 

分 类 号：TG306[金属学及工艺—金属压力加工] TH131[机械工程—机械制造及自动化]