In the deep drawing process, the initial blank has a simple shape but its perimeter shape becomes very complex after drawing. If the initial blank shape is designed in such a way that it is formed directly into the desired shape by the drawing process, this will lead to a reduction in the trimming process and a decrease in the drawing force and raw material. The present paper proposes a novel approach to the initial blank optimization in deep drawing by using an artificial neural network (ANN) to obtain the shape of the initial blank in one step. The finite element method (FEM) is employed for simulating the deep drawing process to provide training data for the ANN. The aim of the neural network is to predict the initial blank shape for the desired final shape. Results from sensitivity analysis and experimental tests were compared. The FEM results were verified through experiment.
KuT. W.LimH. J.ChoiH. H.HwangS. M.KangB. S., ‘Implementation of backward tracing scheme of the FEM to blank design in sheet metal forming’J. Mater. Process. Technol.111 (2001): 90–97.
2.
BiglariF. R.AgahiA.NikfarjamO.NikbinK., ‘Optimum blank design based on modified sensitivity approach’ In Proceedings of the Royal Aeronautical Society virtual testing conference, London, UK, 25–26 October 2006, pp. 158–163.
3.
KarmaM., ‘Blank development and tooling design for drawn parts using modified slip line field based approach’J. Engng Ind. Trans. ASME111 (1989): 345–350.
4.
SowerbyR.DuncanJ. L.ChuE., ‘The modeling of sheet metal stampings’Int. J. Mech. Sci.28 (1986): 415–430.
5.
BlountG. N.FischerB. V., ‘Computerized blank shape prediction for sheet metal components having doubly-curved surfaces’Int. J. Prod. Res.33 (1995): 993–1005.
6.
TohC. H.KobayashiS., ‘Deformation analysis and blank design in square cup drawing’Int. J. Mech. Sci.25 (1985): 15–32.
7.
KimN.KobayashiS., ‘Blank design in rectangular cup drawing by an approximate method’Int. J. Machine Tool Des. Res.26 (1986): 125–135.
8.
LeeC. H.HuhH., ‘Blank design and strain prediction of automobile stamping parts by an inverse finite element approach’J. Mater. Process. Technol.63 (1997): 645–650.
9.
KimS. D.ParkM. H.KimS. J.SeoD. G., ‘Blank design and formability for non-circular deep drawing processes by the finite element method’J. Mater. Process. Technol.75 (1998): 94–99.
10.
ParsaM. H.MatinP. H.MashhadiM. M., ‘Improvement of initial blank shape for intricate products using slip line field’J. Mater. Process. Technol.145 (2004): 21–26.
11.
LanaF.LinJ.ChenJ., ‘An integrated numerical technique in determining blank shape for net shape sheet metal forming’J. Mater. Process. Technol.177 (2006): 72–75.
12.
GuoY. Q.BatozJ. L.NaceurH.BouabdallahH. S.MercierF.BarletO., ‘Recent developments on the analysis and optimum design of sheet metal forming parts using a simplified inverse approach’Comput. Struct.78 (2000): 133–148.
13.
StelsonA., ‘Real time identification of work piece material characteristics from measurements during brake forming’J. Engng Ind., Trans. ASME105 (1983): 45–53.
14.
ChandraA., ‘Real-time identification and control of spring back in sheet metal forming’J. Engng Ind., Trans. ASME109 (1987): 265–273.
15.
ManabeK.YangM.YoshiharaS., ‘Artificial intelligence identification of process parameters and adaptive control system for deep-drawing process’J. Mater. Process. Technol.80–81 (1998): 421–426.
16.
ManabeK.YoshiharaS.YangM.NishimuraH., ‘Fuzzy controlled variable blank holding force technique for circular cup deep drawing of aluminum alloy sheet’Proc. NAMRC No. MF95-121 (1995): 41–46.
17.
ShimH. B.SonK. C., ‘Optimal blank design for the drawings of arbitrary shapes by the sensitivity method’J. ASME123 (2001): 468–475.
18.
CarpenterW. C.HoffmanM. E., ‘Selecting the architecture of a class of back propagation neural networks used as approximations’J. AIEDAM11 (1997): 33–44.
19.
SiJ.ZhouG.LiH., ‘Theory and application of the supervised learning method based on gradient algorithms, Part II—training mechanism’J. Tsinghua Univ.37 (7) (1997): 71–73.
20.
HagenM. T., ‘Training feed forward networks with the Levenberg—Marquardt algorithm’IEEE Trans. Neural Network5 (1994): 989–993.
21.
WeiliX.YuyingY.WangZ. R.LinZ. Q., ‘Stamping and stamping simulation with a blank holder gap’J. Mater. Process. Technol.120 (2002): 62–67.
22.
KobayashiS.AltanT., Metal forming and the finite element method (New York: Oxford University Press1989).
