An investigation on grinding process of natural stones using artificial neural networks

Abstract

Predictions on the tool condition and the surface finish of workpiece in grinding process of metal materials have been studied in the past years using physical and empirical models. In this paper, the feasibility of using neural networks, based on signals detected by multi-sensorial system to monitor tool and workpiece surface conditions in grinding operation of natural stones, has been investigated. Grinding wheel wear evaluation has been carried out measuring flat area percentage on the active surface of the tool by means of a vision system. Workpiece surface roughness has been assessed by means of a mechanical profilometer. Neural network models have allowed to predict grinding wheel cutting ability and workpiece surface finish by measuring on-line the grinding forces and the surface wheel temperature variation.