BACK PROPAGATION NEURAL NETWORK-BASED CUTTING FORCES PREDICTION IN DRY TURNING SKD11 STEEL OF HEAT TREATED WITH CBN INSERTS

Abstract

Cutting force (CF) is one of the most important factors in improving machining efficiency. It directly affects the
cutting tool life and the quality of the product. This paper presents the results of building a model to predict the
value of the cutting force components using a back-propagation (BP) neural network when dry and hard turning
SKD11 steel after heat treatment. The artificial neural network (ANN) training dataset is collected from 27 full dry
turning experiments with CBN coated hard alloy cutting inserts and various cutting parameters including depth of
cut, feed rate and cutting speed. The value of the cutting force components is measured by a specialized and
modern force measuring device. The back-propagation neural network is established with many different
structures to evaluate and select the most suitable structure. Indicators such as Coefficient of Regression (R2), Root
Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) are used to assesment the quality of neural
networks. The research results show that the generated neural network can effectively predict the value of the
shear force components within the trained range. This result is the foundation for further research on vibration and
tool wear as well as building a cutting force monitoring model during machining.