Bestimmung des Formänderungsvermögens bei der Kaltmassivumformung

Up to now there is no quantitative prediction of the workpiece material deformability in cold forming processes possible. Using the current fracture criteria the moment of a crack initiation is not predictable. Such a numeric determination of the deformability would afford an improved economic interpretation of cold forming processes and thus represent a scientific innovation. Hence the aim of the thesis was the development, verification and validation of a new model for the numeric determination of the deformability of cold formed metal materials. In order to achieve this aim, a multi-level proceeding was converted. In the context of the model development a completely new damage model was developed, which separates from the past macro-mechanical and micro-mechanical damage models completely. This was the only way to take up the advantages of existing damage criteria and to neglect their disadvantages at the same time. The basis of the model is to generate a forming history, which describes the deformability of the material. Thus several experimental tests were developed and the deformability of the material 16MnCr5 was determined experimentally and afterwards described by several forming histories numerically. Thus a database was generated, which represents the deformability of the material 16MnCr5 on basis of several forming histories numerically. In a second step a method developed on basis of the pattern recognition, which is able to evaluate the database of forming histories with the aim of the damage prediction. In addition an artificial neural network (ANN) was selected. In a rough optimization with the aim of the network structure and a fine optimization for the selection of a suitable learning algorithm as well as learning parameters an operational ANN was developed. On the basis several experimental tests as well as industrial parts a serviceable ANN was finally verified and validated, too. With this doctor thesis it succeeded for the first time to obtain on basis of artificial intelligence a qualitatively and also quantitatively very good prediction of ductile workpiece damage in cold forming processes. On the basis several experimental tests it was shown that the developed method can predict damages near the surface and fractures on the inside precisely. Further on the deformability within workpiece ranges, which notice a change of the load direction during the forming operation, can be interpreted too. Thus the developed method represents a substantial improvement in relation to the current fracture criteria. The comparison of the new method using ANN with the damage criterion of Cockcroft and Latham showed that the error within the crack prediction could be reduced at five different forming processes from 600% to 6%.

Grantor dc:publisher

Shaker

Year dc:date

2007

Author dc:creator

• Breuer, Dirk

Contributors dc:contributor

• Klocke, Fritz

• info:eu-repo/classification/ddc/620
• Finite-Elemente-Methode
• Neuronales Netz
• Fertigung
• Mehrstufiges Umformen
• Umformen
• Ingenieurwissenschaften
• Kaltmassivumformen
• Metallischer Werkstoff
• Spannungs-Dehnungs-Beziehung
• Formänderungsvermögen
• Werkstoffschädigung
• Numerisches Modell
• cold forming
• fracture
• artificial neural network (ANN)
• fracture criterion
• UMFOTF100

Statement dc:rights

• info:eu-repo/semantics/openAccess

Language dc:language

ger

Identifier

https://publications.rwth-aachen.de/search?p=id:%22RWTH-CONV-123987%22

OAI identifier oai:identifier

oai:publications.rwth-aachen.de:62416