基于正交试验和BP神经网络的板材多点渐进成形工艺优化

发布时间：2018-02-11 16:36

本文关键词： BP神经网络渐进成形正交试验数值模拟　出处：《锻压技术》2015年05期 　论文类型：期刊论文

【摘要】：针对金属板材渐进成形过程中易出现壁厚不均的问题,在多点渐进成形工艺的基础上,选定合理的工艺参数,建立有限元模型,设计正交试验方案,利用ANSYS/LS-DYNA对方锥台制件渐进成形过程进行数值模拟,并对正交试验结果进行极差分析、方差分析和BP神经网络优化。结果表明:在板材多点渐进成形中,进给量对目标制件成形区壁厚均匀性影响最大,其次是工具头半径,进给速度影响不明显;BP神经网络模型的预测结果与正交试验结果相比误差小于5%;1060铝合金板材在多点渐进成形过程中,当工具头半径为6 mm、进给量为0.25 mm、进给速度为30 mm·s-1时,可获得壁厚较均匀的目标制件。
[Abstract]:In view of the problem of uneven wall thickness in the process of sheet metal progressive forming, on the basis of multi-point progressive forming process, the reasonable process parameters are selected, the finite element model is established, and the orthogonal test scheme is designed. Numerical simulation was carried out by using ANSYS/LS-DYNA to simulate the progressive forming process of the tapered parts, and the results of orthogonal test were analyzed by range analysis, variance analysis and BP neural network optimization. The results showed that: in multipoint progressive forming of sheet metal, The feed rate has the greatest influence on the uniformity of the wall thickness in the forming area of the target parts, followed by the radius of the tool head. The prediction result of BP neural network model is less than that of orthogonal test. When the radius of tool head is 6 mm, feed rate is 0.25 mm, feed speed is 30 mm 路s ~ (-1), the error is less than 5 ~ (10) mm in multi-point progressive forming of aluminum alloy sheet, and the feed speed is 30 mm 路s ~ (-1), when the tool head radius is 6 mm, feed rate is 0.25 mm. The target parts with uniform wall thickness can be obtained.
【作者单位】：江西理工大学机电工程学院;
【基金】：国家自然科学基金资助项目(50975131)
【分类号】：TG306

【参考文献】