Title: Modelling manufacturing processes: a comparison between multiple regression analysis and the neural networks approach

Authors: Silvia Miguez, Guillem Vallicrosa, Joaquim Ciurana

Addresses: Mechanical Engineering and Industrial Construction Department, University of Girona, Av. Lluis Santalo s/n, 17071 Girona, Spain. ' Mechanical Engineering and Industrial Construction Department, University of Girona, Av. Lluis Santalo s/n, 17071 Girona, Spain. ' Mechanical Engineering and Industrial Construction Department, University of Girona, Av. Lluis Santalo s/n, 17071 Girona, Spain

Abstract: This paper focuses on modelling process parameters in machining. Grinding and turning are much influenced by parameters used in operations. Poor selection of these process parameters can cause thermal damage in the material, poor accuracy in the final part and premature tool wear. It is important to predict features of final part produced based on process parameters then selection of process operational parameters is highly critical for successful grinding and turning. A set of designed experiments are carried in grinding and turning. Relation between process parameters and quality characteristics has been modelled with linear regression and artificial neural networks (ANN). The paper examines the performance of the models, and compares the models| outputs to determine which model offers the best results. The correlation obtained by the ANN models is better than the one obtained by linear regression. This result shows that proposed models are suitable to identify optimum process settings.

Keywords: manufacturing processes; linear regression; artificial neural networks; ANNs; grinding; turning; process modelling.

DOI: 10.1504/IJMMS.2010.036066

International Journal of Mechatronics and Manufacturing Systems, 2010 Vol.3 No.5/6, pp.405 - 424

Published online: 17 Oct 2010 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article