Neural Network Predictive Model in Cutting Tool Condition Detection

The milling process, characterised by its complexity and the use of a multi-point cutting tool, requires close monitoring of machining responses such as tool wear, forces and vibrations. This vigilance is critical to maintaining operational efficiency and workpiece quality. This article discusses a study of tool condition monitoring in the milling process using a multi-sensor system to improve workpiece quality and production efficiency. Preliminary signal analysis from accelerometers and microphones reveals differences between cutting layers that indicate changes in tool condition. The research aims to develop a predictive model for tool condition monitoring using neural networks for material layer identification with a high accuracy of 0.9229. In addition, ROC curve and AUC value analysis confirm the effectiveness of the classifiers, supporting their potential to reduce raw material consumption and promote sustainable manufacturing practices.