A sensitivity analysis on artificial neural networks fracture predictions in sheet metal forming operations

Rosa Di Lorenzo, Giuseppe Ingarao, Fabrizio Micari

Università di Palermo, Dipartimento di Tecnologia Meccanica, Produzione e Ingegneria Gestionale,,viale delle Scienze 90128, Palermo, Italy.

DOI:

https://doi.org/10.7494/cmms.2008.2.0188

Abstract:

In the last years the investigation of formability limits in sheet metal forming operations was one of the topic in the academic and industrial research due to the wide interest on fracture prevention in such processes. Many approaches were proposed mainly based on the development of fracture criteria or on the utilisation of Forming Limit Curves (FLCs). Actually, such approaches are not effective enough, in particular, when complex deformation path are concerned, namely when multi-step processes are taken into account. The authors have recently proposed a different approach to fracture prediction based on the utilisation of artificial intelligence tools. Such approach is based on the idea that a properly designed and trained artificial neural network is able to predict fracture occurrence for different deformation conditions i.e. for different processes. The early results of the application of such approach were very satisfactory but the robustness of the prediction has to be demonstrated. In this paper, the authors present the results of a sensitivity analysis performed on the neural network fracture predictions in order to assess the robustness of such predictive tool.

Cite as:

Lorenzo, R., Ingarao, G., Micari, F., (2008). A sensitivity analysis on artificial neural networks fracture predictions in sheet metal forming operations. Computer Methods in Materials Science, 8(2), 103 – 110. https://doi.org/10.7494/cmms.2008.2.0188

Article (PDF):