Application of machine learning for tool condition monitoring using sensor integrated tooling

In manufacturing, effective Tool Condition Monitoring (TCM) systems are essential for maximising tool life and reducing production costs. Traditional methods, such as manual inspection of cutting edges, are effective but lead to significant machine downtime and hence, are rarely used. Alternatively, tool wear can be indirectly estimated through parameters such as cutting forces and vibrations experienced during machining. This paper presents a deep learning model designed to predict tool conditions from data collected using a special sensor integrated tool holder. To achieve this, the data was separated into uniform windows and transformed into the time-frequency domain to generate magnitude scalograms. These scalograms were used to train a Convolutional Neural Network (CNN) to classify the tool's condition into “New” or “Worn”. The results demonstrate the effectiveness of this approach for accurate and reliable TCM, highlighting its potential for automated tool condition monitoring.