Soft Modeling of WEDM Process in Prediction of Surface Roughness Using Artificial Neural Networks

Abstract

AbstractWire electrical discharge machining (WEDM) is a very popular unconventional machining technique which is widely used in the aerospace, nuclear and precision industries. This process makes it possible to accurately machine parts with complex shapes and varying hardness which is why it is preferred over the conventional machining methods. The important material properties like wear resistance, fatigue strength and strength are greatly affected by the surface roughness of the material. Conventional approaches of estimation of surface roughness are a very expensive and time-consuming process. In our study, surface roughness of a machined workpiece is predicted through artificial neural network (ANN). The experimental dataset was designed on the basis of Taguchi’s L16 orthogonal array which uses pulse on, current, pulse off and bed speed as the process parameters. Neural network toolbox in MATLAB is used for developing the ANN model which predicts the surface roughness. The efficacy of the ANN model is verified by the regression values.KeywordsWEDMSurface roughnessANNTaguchiRegressionSparkDielectric