Parametric modelling on prediction of surface finish in turning of difficult-to-machine steels

Abstract

Parametric modelling based on multiple regression analysis (MRA), artificial neural networks (ANN) and case-based reasoning (CBR) is developed to predict surface finish during the turning process. Experiments are conducted on difficult-to-machine steels such as AISI 504, AISI D2 and AISI 52100 under different machining conditions with cutting tools viz., multicoated carbide, cermet and alumina inserts. The influence of each input (machining) parameter on surface finish obtained on the workpiece has been determined using analysis of variance (ANOVA) technique. 114 experimental data sets are used for developing the parametric models. 20 sets of validation experiments are conducted in order to evaluate the performance of the developed models. The models are compared based on certain quantitative (statistical measures) and qualitative aspects. It is concluded that CBR model outperformed the other two models in predicting surface finish for the machining conditions considered to a reasonable accuracy.