Modeling of EDM Process Parameters in Machining of 17-4 PH Steel using Artificial Neural Network

Abstract

Objectives: The present work on the development of artificial neural network modeling and prediction of the machining quality for Electrical Discharge Machining of martensitic Precipitation Hardening (PH) Stainless steel and copper tungsten as tool electrode. Methods/Statistical analysis: The important process parameters in this study are peak current, pulse on time, pulse off time and tool lift time with machining qualities as material removal rate and surface roughness. To conduct the experiments L27 orthogonal array was used. Findings: Prediction of Material Removal Rate and Surface Roughness with regression analysis when compared with the experimental results shows variation due to nonlinear complex phenomena which influence the accuracy and precision of the product. In such circumstances, a Artificial neural network model is developed using MATLAB programming on the Levenberg-Marquardt back propagation technique with appropriate architecture of the logistic sigmoid activation function to predict the responses. The experimental data were segregated in three parts to train the network, to testing for convergence and finally to validate the model. The developed model has been verified experimentally for training and testing in considering the number of iterations and mean square error convergence criteria. Improvements/Applications The developed model results are to approximate the responses quite accurately. Results revealed that the proposed model can be successfully employed in the prediction of the complex EDM process.