Kohonen Maps for Chip Form Classification in Turning

Chip form classification during longitudinal turning of carbon steel with coated carbide inserts, yielding different chip forms, was performed using cutting force sensor signals. Advanced signal analysis was carried out by spectral estimation through a parametric method. In this approach, the signal spectrum is assumed to take on a specific functional form, the parameters of which are unknown. The spectral estimation problem becomes one of estimating the unknown parameters of the spectrum model, rather than the spectrum itself. From the cutting force signal, a set of features, corresponding to the characteristic parameters of the spectrum model, were obtained by linear predictive analysis. Decision making on chip form through the analysis of sensor signal features was performed using an unsupervised neural network methodology based on Kohonen maps. The selection of the parameters and variants of the map during the training phase are studied to improve the quality of the Kohonen maps.