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。
For the problem of uneven thickness easily caused in the incremental forming of metal sheet, based on the multi-point incre- mental forming process, reasonable process parameters were elected, finite element model was established, orthogonal experiment program was designed, and the incremental forming process of pyramid part was simulated numerically by ANSYS/LS-DYNA. Then, range analysis, variance analysis and BP neural network optimization of orthogonal experiment results were carried out. The results show that the feed has the greatest effect on the uniformity of wall thickness of targeted forming area in the multi-point incremental forming of sheet, and the second greatest effect is on the tool-head radius, while the impact of feed speed is little. Comparing with the results of the orthogonal experiment, the error of prediction results by BP neural network is less than 5%. Therefore, a targeted part with even wall thickness can be obtained with the tool-head radius of 6 mm, feed of 0. 25 mm and feed speed of 30 mm ·s-1 in the multi-point incremental forming of 1060 Al alloy sheet.
出处
《锻压技术》
CAS
CSCD
北大核心
2015年第5期33-37,共5页
Forging & Stamping Technology
基金
国家自然科学基金资助项目(50975131)
关键词
BP神经网络
渐进成形
正交试验
数值模拟
BP neural network
incremental forming
orthogonal examination
numerical simulation