Authors: T. Tawakoli, M. Rabiey, M. Lewis
Addresses: Competence Center for Grinding Technology and Superfinishing, Furtwangen University, Jakob-Kienzle-Str, 17, 78054 Villingen-Schwenningen, Germany. ' Competence Center for Grinding Technology and Superfinishing, Furtwangen University, Jakob-Kienzle-Str, 17, 78054 Villingen-Schwenningen, Germany. ' UNB Mechanical Engineering, University of New Brunswick, P.O. Box 4400, 15 Dineen Drive, Fredericton NB, E3B 2E3, Canada
Abstract: Grinding in general, as compared with other machining processes, offers advantages such as superior surface finish and low residual stresses. However, it is rather difficult to achieve these goals in particular with dry grinding due to the complicated nature of the process. Any change in grinding parameters or grinding wheel characteristics can have a great influence on process results. A specially-conditioned resin-bonded CBN grinding wheel was proposed earlier to reduce harmful heat generation in dry grinding and to optimise the process. A subsequent study was conducted using Neural Network method to further the advancement of this research. In this work a comparison of some of the very good results of Neural Network prediction is presented, which was carried out to characterise the complex interaction of parameters in dry grinding by special conditioning, based on the innovative concept. The results showed an excellent prediction of the grinding forces, roughness and workpiece surface burning by Neural Network analysis.
Keywords: dry grinding; resin bond; conditioning; CBN grinding wheels; neural networks; process prediction; Levenberg-Marquardt algorithm; radial basis network; RBN; grinding forces; surface roughness; workpiece surface burning.
International Journal of Materials and Product Technology, 2009 Vol.35 No.1/2, pp.118 - 133
Available online: 16 May 2009 *Full-text access for editors Access for subscribers Purchase this article Comment on this article