A comparison of different neural network training algorithms for hydromechanical deep drawing. (1st September 2004)

• Record Type: Journal Article
• Authors: Singh, Swadesh Kumar; Kumar, D. Ravi
• Is Part Of: International journal of material & product technology. Volume 21:Number 1-3(2004)
• Page Start: 186
• Page End: 199
• Publication Date: 2004-09-01
• Subjects: artificial neural networks -- back propagation -- counter pressure -- deep drawing -- EDD -- training -- hydromechanical deep drawing; Manufacturing processes -- Periodicals; Materials -- Testing -- Periodicals; 620.11
• Languages: English
• ISSNs: 0268-1900

Abstract

In the hydromechanical deep drawing process, a pressure chamber is attached to the drawing die and the cup is drawn into the chamber against the fluid pressure. This process offers several advantages over conventional deep drawing, such as higher drawability, more uniform thickness distribution, better surface finish etc. Optimisation of this process is more difficult because of the large number of variables which interact in a complex way. Artificial Neural Networks (ANN) are being applied to an increasing number of real-world problems of considerable complexity. They offer reliable solutions to a variety of problems (viz. prediction and modelling), where the physical processes are not understood or are highly complex. This paper compares several training algorithms in an attempt to find an ideal artificial neural network-training algorithm to model hydromechanical deep drawing. A comparison was made between ANN trained and experimental results of hydromechanical deep drawing using low carbon extra deep drawing (EDD) grade steel sheets of 0.96mm thickness.