A Study on Wire Electric Discharge Machining Process Parameters Prediction Model Using Deep Learning Neural Network

Abstract

Wire electric discharge machining (WEDM) is a nontraditional machining process which shows high capability for improvement in the surface properties of the machined product as high-quality machined products has high demand in the manufacturing industry, and WEDM process is capable for machining of all sorts of conductive materials of any hardness. The presented work focuses on the formation of prediction model using deep learning feed forward back propagation neural network (FFBP ANN) for the WEDM process parameters which will give the user a choice for selecting the best process parameters as per their requirement. The proposed FFBP ANN model has six neurons, ten neurons, and two neurons in the input layer, hidden layer, and output layer, respectively, which will remove all the limitation of the previously developed models and conventional techniques for finding the best machining process parameters. Then, proposed model’s predicted values are compared with the experimental values and regression prediction model values for checking the capability of the proposed FFBP ANN model, and it is found that accuracy of the proposed model is higher than the regression prediction model. The average percentage error between the FFBP ANN-predicted values and the experimental values for cutting speed and surface roughness is 2.99 and 1.74%, respectively. The correlation coefficient between the FFBP ANN-predicted values and the experimental values is found to be 0.9909. Then, statistical validation of the proposed model is done, and the results showed that the FFBP ANN model has a statistically acceptable goodness of fit.