The prediction of surface roughness and tool vibration by using metaheuristic-based ANFIS during dry turning of Al alloy (AA6013)

Abstract

In this article, the adaptive neuro-based fuzzy inference system (ANFIS) model is developed to estimate the surface roughness (Ra) and tool vibrations (Acc) of AA6013 aluminum alloy during dry turning. Turning experiments were carried out with seven different cutting speeds, five different feed rates and seven different depth of cuts. These three different cutting parameters were tested with each other in different variations. ANFIS model is optimized using the genetic algorithm (GA), particle swarm optimization (PSO) and ant colony optimization. Performance of the developed model is compared with that of multi-linear regression model, which is one of the conventional prediction approaches. At the end of the study, it is revealed that the GA-ANFIS with an R-value of 0.946 is seen as the best model among the proposed approaches in the estimation of Acc. The PSO-ANFIS with an R-value of 0.916 is seen as the best model among the proposed approaches in the estimation of Ra. GA-ANFIS model for Acc prediction and PSO-ANFIS model for Ra prediction are the best approaches among the models discussed in the study. Moreover, the relationship between Acc and Ra values was examined and an empirical model was proposed.