APPLICATION OF ARTIFICIAL NEURAL NETWORKS FOR EVALUATING THE PERFORMANCE OF CUTTING TOOLS MADE OF PCBN

Abstract

The methodology and results of the study of the components of cutting forces, acoustic emission and vibrations during the finish turning of hardened steel ShKh15 (HRC 50…60) are presented. To find the relationship between the components of cutting forces and vibrations, an approach based on the creation of models that take into account various uncertainties was used. For this purpose, the developed artificial neural network (ANN) was used. Turning cutters with cutting inserts made of 5 types of polycrystalline materials based on cubic boron nitride were studied. The developed neural network initially had a complex architecture that included six hidden layers. The number of neurons in these layers varied from 32 to 256. The average relative error when using ANN was no more than 12%, which allows using ANN in real time in studies of the performance of cutting tools with PCNB. The results of the study open new prospects for the further development of adaptive machine tool control systems based on vibration control in the machining process.