A neural network procedure to predict the cutting temperature of coated tool cemented carbide in face milling process

Abstract

The high temperatures generated during the cutting process can quickly become one of the main reasons of premature failure of the cutting tools. In milling process due to the characteristic of discontinuity of the cutting operation, it becomes much more problematic. The determination of a reliable procedure to estimate the cutting temperatures in machining processes could allow the optimization of the cutting parameters providing a tool life longer. By this way, the present work proposes a neural network based procedure aiming to establish an experimental relationship between the cutting temperature generated in coated tools cemented carbide and the main cutting parameters in the face milling process: cutting speed "vc�, feed per tooth "fz�, depth of cut "ap� and cutting width "ae�. The trained networks were used to predict the cutting temperatures generated not just in the entrance but also in the exit of the cutting tool of the work piece. The choice of the Neural Network procedure was motivated by the satisfactory resuts presented by this technique when predicting and modeling nonlinear systems with non-correlated variables. The results showed that the neural networks methodology used is a promising technique to estimate the cutting temperatures, besides being an important tool to aid in the choice of the fit parameters, having as a consequence an increasing of the tool life. Keywords: Neural Networks, Cutting Temperature, Milling.