Multivariate mean square error for the multiobjective optimization of AISI 52100 hardened steel turning with wiper ceramic inserts tool: a comparative study

Abstract

This work aims at comparing the optimization of an AISI-52100-steel turning process with a wiper tool attained through the use of the weighted multivariate mean square error (MMSE) method and that obtained through the use of the MMSE. Both of these methods combine principal component analysis and response surface methodology, with the difference that the weighted approach allows the assignment of different degrees of importance to each response. Three input factors were considered: cutting speed (Vc), feed rate (f) and depth of cut (d). Six highly correlated output characteristics (process responses) were considered: tool life (T), cutting time (Ct), total turning cycle time (Tt), processing cost per piece (Kp), arithmetic mean roughness (Ra) and maximum peak to valley roughness (Rt). It should be kept in mind that the material removal rate was a constraint of the problem and not a dependent factor, as a means to guarantee the existence of a minimum productivity. The multiobjective optimization results have been validated experimentally. All models developed in this study, both for turning outputs and for principal component scores, are then suitable for predicting and controlling turning processes similar to that studied here.