SENSITIVITY ANALYSIS OF THE ARTIFICIAL NEURAL NETWORK INPUTS IN A SYSTEM FOR DURABILITY PREDICTION OF FORGING TOOLS

1 MRZYGŁÓD Barbara
Co-authors:
2 Hawryluk Marek 2 POLAK Sławomir
Institutions:
1 AGH University of Science and Technology, Faculty of Metals Engineering and Industrial Computer Science, al. A. Mickiewicza 30, 30-059 Cracow, Poland, EU, mrzyglod@agh.edu.pl
2 Wroclaw University of Science and Technology, Wyb. Wyspiańskiego 27, 0-370 Wroclaw, Poland, EU, marek.hawryluk@pwr.edu.pl
Conference:
26th International Conference on Metallurgy and Materials, Hotel Voronez I, Brno, Czech Republic, EU, May 24th - 26th 2017
Proceedings:
Proceedings 26th International Conference on Metallurgy and Materials
Pages:
484-489
ISBN:
978-80-87294-79-6
ISSN:
2694-9296
Published:
9th January 2018
Proceedings of the conference were published in Web of Science and Scopus.
Metrics:
29 views / 9 downloads
Abstract

The paper presents the results of neural network sensitivity analysis used in prediction system of tool durability in die forging processes. Data collected during many experiments, tabulated in the form of knowledge vectors, has been used as a source of training data for artificial neural networks. The sensitivity analysis makes it possible to differentiate between the important variables and those which do not make a significant contribution to the results of the network operation. The obtained results of global sensitivity analysis, conducted for the elaborated network in the context of predicting the life of forging tools from the expert viewpoint, indicate general correctness and validity of the adopted model (solution), ascribing the highest sensitivity to the nitritiding input variable (related to hardness), which is in reality the main factor determining the tool resistance to the destructive effect of failure mechanisms.

Keywords: artificial neural network, decision support system, durability of forging tools
Scroll to Top