Neural networks in surface roughness estimation

In-process inspection systems are necessary for the automatic control of surface quality in a sensor- based manufacturing environment. The machined surface digital image has been applied for the estimation of surface roughness. The measuring system has been built for machined surface digital image acquisition. This image has been analyzed and optimized 4-element feature versus has been prepared. The optimized neural network has been applied for machined surface parameters estimation. The estimation results proved that it is possible to apply the machined surface digital image for the surface roughness on-line monitoring.