Wire electrical discharge machining (WEDM) technology has been widely
used in conductive material machining. The WEDM process, which is a co
mbination of electrodynamic, electromagnetic, thermaldynamic, and hydr
odynamic actions, exhibits a complex and stochastic nature. Its perfor
mance, in terms of surface finish and machining productivity, is affec
ted by many factors. This paper presents an attempt at optimization of
the process parametric combinations by modeling the process using art
ificial neural networks (ANN) and characterizes the WEDMed surface thr
ough time series techniques. A feed-forward back-propagation neural ne
twork based on a central composite rotatable experimental design is de
veloped to model the machining process. Optimal parametric combination
s are selected for the process. The periodic component of the surface
texture is identified, and an autoregressive AR(3) model is used to de
scribe its stochastic component. (C) Elsevier Science Inc., 1997.