An Experimental Study On Tool Wear Condition Monitoring Of CNC Milling Using Machine Learning Algorithms

Abstract

Due to the nonlinear nature of the cutting process, conventional regression models cannot provide reliable forecasts. In order to predict certain cutting parameters of CNC milling tool wear, this research presents a hybrid method that makes use of deep neural networks (DNN). We ran orthogonal cutting experiments and two-dimensional finite element model chip formation simulations over a range of cutting settings, tool geometries, and wear conditions to collect data for hybrid training. We used a DNN in conjunction with the tried-and-true linear regression technique to build our predictive models. The accuracy of the hybrid model that included machine learning methods was higher than that of traditional linear regression.