Experimental Investigation and Prediction of Mild Steel Turning Performances Using Hybrid Deep Convolutional Neural Network-Based Manta-Ray Foraging Optimizer

In machining industries, sustainable production activities are reduced because of the tool wear effect. The machining input parameters are mainly responsible for the effect of flank wear attained in the tool and the workpiece's surface quality. This work aims to determine and predict the machining performances of EN2-BS970/Mild steel by varying the input parameters such as cutting speed (V), feed (F), nose radius (r), and depth of cut (d). The hybrid Deep Convolutional Neural Network-based Manta-Ray Optimization (DCNN-MRO) is used to predict the machining outcomes, and it is performed in Matlab software version 2020a. The input machining parameters are designed by response surface methodology of box behnken design performed in Design-Expert software version 11. The experimented different cutting forces are feed force (Fx), radial force (Fy), cutting force (Fz), and the machining performances are tool flank wear, surface roughness, and Tool chip thickness. In which, the machining input parameter, namely cutting speed effectively influences the turning outcomes. The effect of tool flank wear and surface roughness by varying the cutting forces are also analyzed. The observed optimal surface roughness is 3.105 \(\mu m\), tool wear rate is 0.139mm, and tool chip thickness is 0.11mm. The measured outcomes are closer to the predicted outcomes obtained from hybrid DCNN. The average RMSE obtained from the proposed DCNN-MRFO is 0.03, and the non-hybrid DCNN is 0.3.