Acoustic characterization and prediction of surface roughness

The present paper deals with a new methodology for predicting surface roughness of engineering surfaces. A contactor has been rubbed against a surface, and the friction noise thus generated is related with the roughness of the surface. The amplitude of friction noise and the magnitude of contact force have been utilized to train a back propagation neural network for future prediction. It has been observed that the neural network can learn the correlation between the friction noise, contact force and the surface roughness and could effectively be used in practical applications to predict surface roughness.