In-process self-configuring approach to develop intelligent tool condition monitoring systems

Abstract

A self-configuring real-time tool condition monitoring (TCM) system for milling applications using vibration signals is introduced. A suite of signal processing and machine learning algorithms was developed to define a generalized correlation between distortion-resistant features of usable and worn tools. Using only a few seconds of learning data acquired at the early stage of tool life, the system synthesizes worn tool features in-process to define the decision-making boundaries, independent of the utilized cutting parameters, machines, and sensors. It provides high detection accuracy and reduces the lead time and cost needed for system development and calibration, introducing the plug-and-play concept to TCM.