Title: Modelling and optimisation of wire electrical discharge machining process on D2 steel using ANN and RMSE approach

Authors: U.K. Vates; Nirmal Kumar Singh; R.V. Singh

Addresses: MAE, ASET, Amity University Uttar Pradesh, Noida, UP, India ' Department of Mechanical Engineering, ISM Dhanbad, India ' Department of Mechanical Engineering, MRIU, Faridabad, India

Abstract: Attempt has been made to optimise surface roughness (Ra) of cold worked high carbon and high chromium hard die steel (AISI: D2) by wire electrical discharge machining (WEDM) process. In present study fractional factorial design of experiment has been used at two levels to conducted five different sets of experiment. Tan-sigmoid and pure line transfer function-based four layered back propagation artificial neural network (BPANN) approach have been applied to develop a suitable model with six WEDM process input parameters and two output parameters, i.e., Ra and material removal rate (MRR). The effect of input parameters has been analysed by training data for the best model using analysis of variance (ANOVA). The model S2 has been found satisfactory as training correlation coefficient (R2) is 99.2% and adjusted correlation coefficient (Radj)2 is 99.1%. The minimum surface roughness has been found using root mean square error (RMSE) approach.

Keywords: wire EDM; electrical discharge machining; WEDM; material removal rate; MRR; surface roughness; surface quality; D2 steel; back propagation ANNs; artificial neural networks; BPANN; analysis of variance; ANOVA; root mean square error; RMSE; modelling; optimisation; electro-discharge machining.

DOI: 10.1504/IJCMSSE.2016.10002548

International Journal of Computational Materials Science and Surface Engineering, 2016 Vol.6 No.3/4, pp.161 - 185

Received: 27 Jan 2015
Accepted: 11 Aug 2015

Published online: 20 Jan 2017 *

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