摘要
芯棒拉拔力对于研究耳片压合衬套冷挤压过程的强化效果具有重要意义,而预测不同加工参数下的芯棒拉拔力有助于寻找最佳的冷挤压强化参数,在提高冷挤压强化效果的同时避免断棒现象的发生。首先,建立耳片压合衬套冷挤压强化过程的有限元仿真模型,获得压合衬套冷挤压过程中芯棒拉拔力曲线,并对芯棒拉拔力最大值进行实验验证。其次,通过Tent混沌初始化、收敛因子改进和自适应动态权重对GWO-BP神经网络模型进行改进,对不同挤压量、衬套壁厚、芯棒工作段长度及芯棒过渡圆弧半径下的芯棒拉拔力最大值进行预测。结果表明,相较于SVM、BP和GWO-BP模型,改进后的IGWO-BP神经网络模型的预测精度更高,最大误差不超过3.05%,可为企业实际生产提供理论依据。
The mandrel drawing force is of great significance for studying the strengthening effect of the cold extrusion process for lug press-fit bushings,and predicting the mandrel drawing force under different processing parameters can help to identify the optimal cold extrusion strengthening parameters,improve the strengthening effect and avoid the occurrence of mandrel breakage.Therefore,firstly,a finite element simulation model for the cold extrusion strengthening process of lug press-fit bushings was established,the mandrel drawing force curve during the process was obtained,and the maximum mandrel drawing force was experimentally verified.Secondly,the GWOBP neural network model was improved by Tent chaotic initialization,convergence factor modification and adaptive dynamic weighting,and the maximum mandrel drawing forces under different extrusion amounts,bushing wall thicknesses,mandrel working section lengths and mandrel transition arc radii were predicted.The results show that compared with the SVM,BP,and GWO-BP models,the improved IGWO-BP neural network model has higher prediction accuracy,with the maximum error of no more than 3.05%,providing a theoretical basis for the actual production in enterprise.
作者
易志东
黎向锋
郑泽庭
唐伟
李文生
刘烨欣
Yi Zhidong;Li Xiangfeng;Zheng Zeting;Tang Wei;Li Wensheng;Liu Yexin(College of Mechanical&Electrical Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China;Aerospace Precision Products lnc.,Ltd.,Tianjin 300300,China)
出处
《锻压技术》
北大核心
2025年第8期123-130,138,共9页
Forging & Stamping Technology
基金
国家自然科学基金联合基金项目(U20A20293)
天津市紧固连接技术企业重点实验室开放课题(HTJG-TJ-YJ-2024-2002)
苏州市科技成果转化专项(No.SZC202317)。
关键词
冷挤压强化
芯棒拉拔力
灰狼算法
神经网络模型
预测精度
cold extrusion strengthening
mandrel drawing force
grey wolf optimizer
neural network model
prediction accuracy
