Surface roughness prediction during grinding: A Comparison of ANN and RBFNN models

Grinding is one of the most widely employed manufacturing processes when accurate finishing of workpieces is required. In order to investigate the effect of processing parameters to grinding performance, soft computing methods constitute a reliable and economical alternative to other simulation methods, such as the Finite Element Method (FEM). In this study, a comparison between classical Artificial Neural Network (ANN) models and Radial Basis Function Neural Network (RBFNN) models is conducted for a case of face grinding of various types of steel workpieces, cutting wheel types and depths of cut and their performance towards the prediction of surface roughness is evaluated. Results indicate that RBFNN can provide better results than classical ANN networks and adequately model the surface roughness during grinding processes. Key-Words: grinding, finishing operations, manufacturing, surface roughness, artificial neural networks, radial basis functions, soft computing