Novel sensor-based tool wear monitoring approach for seamless implementation in high speed milling applications

Abstract A sensor-based hybrid processing approach for tool wear monitoring is presented to overcome the practical limitations of implementing state-of-the-art tool condition monitoring systems in milling processes. It extracts features from vibration signals that are insensitive to the variations in cutting conditions, tool path and interfering noises. A machine learning model was developed to accentuate features separation based on tool condition. Extensive experimental validation tests in high speed and conventional milling applications demonstrated the approach capability to achieve 98% accuracy and reduce system training by up to 97%. Such performance, practicality and accuracy have never been reached before in this application.