Title: ANN modelling for surface roughness in electrical discharge machining: a comparative study

Authors: Raja Das; M.K. Pradhan

Addresses: School of Advanced Sciences, VIT University, Vellore, Tamil Nadu-632014, India ' Department of Mechanical Engineering, Maulana Azad National Institute of Technology, Bhopal, 462051, India

Abstract: This is an attempt to present three different classes of artificial neural network (ANN) models, namely back-propagation network (BPN), radial basis function network (RBFN) and recurrent neural network (RNN) for the prediction of surface roughness (Ra) in electrical discharge machining (EDM). Surface roughness is an important issue in the manufacturing. The input variable chosen was the pulse current (Ip), the pulse duration (Ton) and duty cycle (τ). A series of experiments was conducted AISI D2 to acquire the data for training and testing, and it was found that the ANN models could predict Ra with reasonable accuracy, under varying machining conditions. A close correlation between the model prediction and the experimental results was witnessed. Moreover, it was noticed that all three models are offering quite an agreeable prediction. The RBFN model is quite analogous with other models but demonstrated a slightly better performance than others.

Keywords: artificial neural networks; ANNs; back propagation neural networks; electrical discharge machining; EDM; radial basis function neural networks; recurrent neural networks; RNN; surface roughness; surface quality; modelling; electro-discharge machining; pulse current; pulse duration; duty cycle.

DOI: 10.1504/IJSCOM.2013.058674

International Journal of Service and Computing Oriented Manufacturing, 2013 Vol.1 No.2, pp.124 - 140

Received: 27 Sep 2012
Accepted: 08 Dec 2012
Published online: 02 Jul 2014 *

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