Hard turning behavior improvement using NSGA-II and PSO-NN hybrid model

The present work concerns an investigation and optimization of hard turning behavior of AISI 52100 bearing steel with CBN tool. The combined effects of the process parameters (cutting speed, feed rate, depth of cut, cutting time, and workpiece hardness) on performance characteristics (tool wear, surface roughness, and cutting forces) are investigated through the analysis of variance (ANOVA). Also, the relationship between process parameters and performance characteristics through the artificial neural networks and response surface methodologies are modeled. Moreover, an attempt has been made to find the best process parameters combination that optimizes simultaneously the performance characteristics. For this purpose, the particle swarm optimization-based neural network (PSO-NN) hybrid model and the non-dominated sorting genetic algorithm (NSGA-II)-based RSM models are proposed and compared for their performances. The results show that the workpiece hardness strongly influences the cutting forces, and it plays a dominant role in the tool wear and the surface roughness variations as compared to other process parameters. The depth of cut affects strongly the cutting forces. The cutting time has a considerable effect on all performance characteristics. They show, also, that the artificial neural network (ANN) model estimates the performance characteristics with high accuracy as compared to the RSM models. Moreover, it was found that NSGA II and PSO-NN approaches performed efficiently and they predicted near similar results. Furthermore, the NSGA-II exhibits better performance when compared to the PSO-NN methodology. However, the PSO-NN has been shown to outperform the NSGA-II methodology in computational time.