Wire electrical discharge machining (WEDM) technology has been used wi
dely in production, aerospace/aircraft, medical and virtually all area
s of conductive material machining. Its complexity and stochastic natu
re have stimulated numerous attempts to model it accurately. This pape
r presents an attempt at modeling the process through Response Surface
Methodology and Artificial Neural Networks. A response surface model
based on a central composite rotatable experimental design, and a 4-16
-3 size back-propagation neural network have been developed. The pulse
-width, the time between two pulses. the wire mechanical tension and t
he injection set-point are selected as the factors (input parameters),
whilst the cutting speed, the surface roughness and the surface wavin
ess are the responses (output parameters). The two models are compared
for goodness of fit. Verification experiments have been carried out t
o check the validity of the developed models. It is concluded that bot
h models provide accurate results for the process. (C) 1997 Elsevier S
cience S.A.