| 引用本文: | 侯柏林,叶颖,孙建芳,袁熙,苏峰华.铝合金薄板冲压成形过程中的侧壁起皱及有限元模拟[J].材料科学与工艺,2024,32(3):87-95.DOI:10.11951/j.issn.1005-0299.20220428. |
| HOU Bolin,YE Ying,SUN Jianfang,YUAN Xi,SU Fenghua.Study on side wall wrinkling and finite element simulation in stamping process of aluminum alloy sheet[J].Materials Science and Technology,2024,32(3):87-95.DOI:10.11951/j.issn.1005-0299.20220428. |
|
| 摘要: |
| 针对铝合金薄板的侧壁起皱问题,本文通过有限元软件分析工艺参数对成形质量的影响,提出了一种基于数值模拟与智能算法相结合的优化方法。首先,利用最优拉丁超立方抽样进行实验设计,并依据数值模拟获取实验值；其次,基于BP神经网络拟合工艺参数与成形质量之间的关系,预测结果的平均相对误差为2.69%,建立了准确的预测模型；最后,用遗传算法极值寻优获取了一组最优的工艺参数组合,起皱幅值的预测值和仿真值相对误差仅为4.03%,实验结果与仿真分析结果相近,验证了该优化方法的合理性和有效性。研究表明:以料厚、摩擦系数和压边力作为优化变量,以最大起皱幅值最小化为优化目标,建立几何模型,并利用有限元软件Autoform进行仿真分析；依据起皱轮廓线径向位移的实验和数值模拟对比,验证了有限元模型的正确性,表明利用神经网络和遗传算法极值寻优可以有效解决铝合金侧壁起皱缺陷。 |
| 关键词: 铝合金 侧壁起皱 BP神经网络 遗传算法 数值模拟 拉丁超立方抽样 |
| DOI：10.11951/j.issn.1005-0299.20220428 |
| 分类号:TG386 |
| 文献标识码:A |
| 基金项目:广州市基础与应用基础研究项目(202102080422)；广东省自然科学基金资助项目(2021A1515012266);2019年度广东省普通高校特色创新类项目(2019GWTSCX101). |
|
| Study on side wall wrinkling and finite element simulation in stamping process of aluminum alloy sheet |
|
HOU Bolin1, YE Ying2,SUN Jianfang2,YUAN Xi3,SU Fenghua2
|
|
(1.City College of Huizhou, Huizhou 516025, China; 2.School of Mechanical and Automotive Engineering,South China University of Technology,Guangzhou 510640,China; 3. Foshan Nanhai Lei Te Automotive Parts Co.,Ltd., Foshan 528244,China)
|
| Abstract: |
| To resolve the side wall wrinkling problem of aluminum alloy sheet,this study through finite element software investigated the influence of process parameters on the forming quality, proposing an optimization method based on the combination of numerical simulation and intelligent algorithm. Firstly, the optimal Latin hypercube sampling is used for experimental design, and the experimental values are obtained by numerical simulation. Secondly, based on BP neural network, the relationship between process parameters and forming quality is fitted. The average relative error of the prediction results is 2.69%, and an accurate prediction model is established. Finally, a set of optimal combination of process parameters is obtained by using the extreme value optimization of genetic algorithm. The relative error between the predicted value and the simulated value of wrinkling amplitude is merely 4.03%. The experimental results prove to be similar to the simulation results, verifying the rationality and effectiveness of the optimization method. The study shows that the geometric model is established with the material thickness, friction coefficient and blank holder force as the optimization variables, and the minimum maximum wrinkling amplitude as the optimization objective. Worthy of note, the finite element software Autoform is used for simulation analysis, and the comparison between the experimental results and the numerical simulation in relation to the radial displacement of the wrinkling contour verifies the finite element model, suggesting that the neural network and genetic algorithm can effectively solve the wrinkling defect of the aluminum alloy sidewall. |
| Key words: aluminum alloy side wall wrinkling BP neural network genetic algorithm numerical simulation Latin hypercube sampling |
