A Comparative study on C45 carbon steel machining Experimental Values with an Artificial Neural Network

Abstract

This study examined the ways in which C45 medium carbon steel's machining characteristics were affected by a rise in trace elements such as
phosphorus and sulfur. In this work, two distinct samples with specific varying percentages of sulfur and phosphorus were compared using tool
inserts which are coated to study machining characteristics [variation/sample-1 S -0.006% and P-0.013% (Lesser percentage of phosphorous
and sulfur) and variation/sample-2 S-0.017% and P-0.025% (phosphorous and sulfur having higher percentage)]. Using turning machine the
Surface finish, cutting forces and material removal, tool tip temperature, and tool wear at flank were all examined and listed in the table for
ANN evaluation with two distinct phosphorus and sulfur percentages. The 3-[9]1-1 network model was trained on a dataset of experimental
values and then ANN findings are compared with the machining outcomes. This study uses 81 datasets for performance evaluation. The
Levenberg-Marquardt approach is used to train the network and the findings of these methods are compared. Iteration twelve yielded the bestfit validation result for the response parameter R = 0.95, indicating near-linear association and very good correlation between the values from
the experiments and the projected values. With the increase of trace minerals, like sulfur and phosphorus, better validation metrics of R=0.9848
were achieved by the sixth iteration.