UĞUR EŞME, İREM ERSÖZ KAYA, ARİF ÖZBEK, Funda KAHRAMAN, Aysun SAĞBAŞ

A comparison between the back-propagation (BPN) and General-Regression Networks (GRNN) in the modeling of the spot welding process

Bu çalısmada direnç kaynağının modellenmesinde geri yayılmalı (BPN) ve genel regresyonlu (GRNN) yapay sinir ağları metodu kullanılmıstır. Çekme mukavemeti her iki modelleme tekniği ile modellenerek sonuçlar birbiri ile kıyaslanmıstır. Bu çalısma ile direnç kaynağının modellenmesinde her iki modelleme yönteminin kullanılabileceği kanıtlanmıstır. Elde edilen sonuçlar, genel regresyonlu sinir ağları metodunun bu yöntemde daha kesin ve düsük hata ile sonuç verdiğini göstermistir. Bu çalısma aynı zamanda, geri yayılmalı yapay sinir ağlarının direnç kaynağın da parametre genellestirmesinde geri yayılmalı yapay sinir ağlarının genel regresyonlu modellemeden daha etkin olduğu nu göstermistir.

Direnç kaynağının modellenmesinde geri yayılmalı ve genel regresyonlu sinir ağlarının karşılaştırılması

The use of neural networks techniques to model spot welding is explored in this paper. Both the back-propagation (BPN) and General Regression Neural Networks (GRNN) are used to associate the welding process parameters with the features of the tensile shear strength. A comparison between the back-propagation and general regression neural networks in the modeling of the spot welding process is given. It is shown that both the back-propagation and general regression neural networks can model the spot welding process with reasonable accuracy. However, general regression neural network has better learning ability for the spot welding process than the back-propagation network, although the back-propagation network has better generalization ability for the spot welding process than does the general regression neural network.

PDF

___

1. Juang,S.C., Tarng,Y.S. and Lii, H.R., “A comparison between the back-propagation counter-propagation networks in the modeling welding process”, Journal of Materials Processing Technology 75, 54–62, (1998).
2. Freeman,J.A. and Skapura,D.M., “Neural Networks: Algorithms, Application and Programming Techniques”, Addison–Wesley, New York, 1991.
3. Andersen,K., Cook,G.E., Karsai,G. and Ramaswamy,K., “Artificial neural networks applied to arc welding process modeling and control”, IEEE Trans. Ind. Appl. 26 (5) 824– 830, (1990).
4. Zeng,X.M., Lucas,J. and Fang,M.T.C., “Use of neural networks for parameter prediction and quality inspection in TIG welding”, Trans. Inst. Meas. Control 15 (2), 87– 95, (1993).
5. Jin,B., Soeda,M. and Oshima,K., “Control of weldpool width and cooling time in TIG welding using a neural network model”, Welding Int. 10 (8), 614–621, (1996).
6. McClelland,J. and Rumelhart,D., “Parallel Distributed Processing”, vol.1, MIT Press, Cambridge, MA, 1986.
7. Specht,D., “A General Regression Neural Network”, IEEE Trans. Neural Networks, Vol. 2, No 6, pp. 568-576, (1991).
8. Cacoullos,T., “Estimation of a multivariate density”, Annals of Inst. Stat. Math., vol. 18, pp. 178-189, (1996).
9. Yeung,D., Chow,C., “Parzen Window Network Intrusion Detectors” Proceedings of the Sixteenth International Conference on Pattern Recognition. Quebec City, Canada, Vol.4, pp.385-388, (2002).
10. Ersoz,I., Ibrikci,T., Cakmak,A., Ersoy,O.K., “Secondary Structure Prediction of Hemoglobin by Neural Networks”, ANNIE 2004 Artificial Neural Networks in Engineering, University of Missouri-Rolla, Missouri, USA, 2004.