A neural network model to predict surface roughness during turning of hardened SS410 steel

Abstract

In manufacture sector, the surface finish quality has considerable importance that can affect the functioning of a component, and possibly its cost. The surface quality is an significant parameter to evaluate the productivity of machine tools as well as machined components. It is also used as the critical quality indicator for the machined surface. In recent years the prediction of surface roughness has become an area of interest for machining industry. Cutting force, cutting temperature, tool wear, and vibration signals are some of the factors that can be used individually to predict surface roughness, but when it is used collectively a more accurate prediction of surface roughness is possible since each of the above-mentioned factors have their own characteristics effects of surface roughness. In the present study, an attempt was made to fuse cutting force, tool wear and displacement of tool vibration along with the cutting speed, feed and depth of cut to predict the surface roughness of hardened SS 410 steel (45 HRC) using a multicoated hard metal insert with a sculptured rake face. Regression models and an artificial neural network model were developed to fuse the cutting force, cutting temperature, tool wear and displacement of tool vibration to predict the surface roughness. From the results it was observed that the prediction of surface roughness by the artificial neural network had a higher accuracy than the regression model.