Predictive Modeling of Surface Roughness in the Machining of Inconel 625 Using Artificial Neural Network

Abstract

Inconel 625 belongs to the category of nickel-based alloy and it has wide applications in aerospace, defense, marine, nuclear, and oil and gas industries, due to its superior properties at elevated temperatures. However, machinability of Inconel 625 is very poor, because of high strength at higher temperature as well as high rate of tool wear and work hardening. In this context, this research work developed a predictive model using the artificial neural network, during the turning operation on nickel-based alloy Inconel 625, with chemical vapor deposition-coated inserts. Cutting speed, feed rate, and depth of cut were selected as input variables and surface roughness was selected as the response variable. Totally, twenty-seven experiments were performed, based on the design of experiments, using L27 orthogonal array and the measured result was analyzed using statistical software MATLAB. Mean absolute percentage error between predicted results and experimental values was 4.90% for surface roughness. The overall correlation coefficient between the artificial neural network-predicted result and experimental values was 0.97734, which showed good agreement between experimental values and predicted results and was nearly accurate.