Artificial-Neural-Network-Based Uncertain Material Removal Rate by Turning

Abstract

The present paper deals with the uncertain material removal rate (UMRR) of cylindrical turning of AISI 52100 steel (as workpiece) being turned by cubic boron nitride (CBN) (as single-point cutting tool insert). During machining operations such as cylindrical turning, assessment of material removal rate is often inharmonious, non-uniform and unpredictable due to variabilities in rotational speed of workpiece, feed and depth of cut provided by the cutting tool. It occurs due to unforeseen operational and manufacturing uncertainties. The present work is aimed to develop a computational model in conjunction with artificial neural network (ANN) approach. The constructed computational model is validated with the previous published experimental results. The traditional Monte Carlo simulation (MCS) is employed to compare the efficacy and accuracy of the constructed artificial neural network (ANN)-based surrogate model. The effect of both individual and combined variations of input parameters such as cutting speed, feed and depth of cut on the material removal rate is portrayed. The surrogate model is validated with the original Monte Carlo simulation (MCS), and the intensity of variation of output quantity of interest (QoI) is presented by the probability density function plots. The statistical analyses are carried out based on parametric studies, and the subsequent results are illustrated. The effect of depth of cut is observed to be maximum sensitive to influence the uncertain material removal rate, followed by feed and cutting speed.