International Journal of Advanced Mechanical Engineering.
ISSN 2250-3234 Volume 4, Number 6 (2014), pp. 667-674
© Research India Publications
http://www.ripublication.com
Prediction of Surface Roughness in WEDM Process Using
Feed Forward Back Propagation Neural Network
Piyush Pant
1
, Navneet k Pandey
2
, S.Rajesha
3
and Gaurav Jain
4
1
M.Tech Research Scholar, JSSATE Noida, Uttar Pradesh
1, 2, 3
ME Deptt, JSSATE, Uttar Pradesh
ABSTRACT
The surface roughness quality significantly influences the machined parts
during their useful life. It is very hard to develop a comprehensive model
involving all the input parameters because of the complexity of the machining
process. In the present study, surface roughness is measured during wire cut
electrical discharge machining (WEDM) process for different values of pulse-
on time, gap voltage and wire feed rate using L27 OA. Artificial neural
network is used to model the surface roughness which is trained and validated
using feed forward back propagation method, Levenberg-Marquardt (LM)
training algorithm, tansig transfer function using one hidden layer, four
neurons and 1000 epochs were carried out. The LM model has produced
absolute fraction of variance (R
2
) values of 0.98069 for the training data,
0.99906 for validation and overall to be 0.96721. The mean square error
decreased from 0.005 to 0.000005 during ANN training. The predicted ANN
data was well within the limits and the simulation also showed good accuracy.
The results indicated that well trained neural network model is quite effective
for prediction of surface roughness within and beyond the experimental
domain.
Keywords: Surface roughness, Wire electrical discharge machining, Artificial
neural network.
INTRODUCTION
WEDM is a non-conventional precision machining process that is widely used in tool
and dies making industry. The machining principle is based on erosion of the work-
piece material using a successive discrete discharges occurring between the wire
electrode and workpiece. Die-making industry is very important to down-stream
industries and any technological changes in the die-making industry surely affect