Abstract When evaluating a machined product’s quality and characteristics, the surface finish is crucial. It is difficult to model a machined component’s surface roughness and hardness for a number of reasons. In this regard, maximizing surface properties like hardness and minimizing roughness during the machining of low-alloy steel, also known as low-carbon steel, is investigated. Experiments using carbide tools for face milling have been carried out under standard settings. Combining artificial neural networks with meta-heuristic optimization algorithms is a trending method to model/predict, and optimize the machining process. In this regard, the study investigated how well this combination models, predicts and optimizes the SI characteristics of alloy E350. Neural networks, such as RBF, have been used for modeling and prediction research; the output of this network has been used as the optimization algorithms’ objective function. RBFNN prediction was found to be precise with average MSE values of 0.0017 (SR) and 0.0010 (MH), and the R 2 values of 0.97 (SR) and 0.95 (MH) of experimental vs predicted values. The following algorithms have been tested: FOA, BA, and PSO. It has been discovered that BA has a higher precision than the other algorithms. BA outperformed PSO with an improvement of 6.6% and 0.3% in predicting the roughness and hardness, respectively. The algorithm predicted optimal conditions have been identified to be v = 274.1 m/min, f = 0.054 mm/rev, a = 0.560 mm for roughness, and v = 152.4 m/min, f = 0.140 mm/rev, a = 1.492 mm for hardness, which are recommended to machine alloy E350. Finally, the precision of bat algorithm prediction has been verified with the validation experiments and confirmed with a low margin of error of 2.79% and 2.19% in roughness and hardness, respectively.
Marakini et al. (Wed,) studied this question.