Machining quality monitoring (MQM) in laser-assisted micro-milling of glass using cutting force signals: an image-based deep transfer learning

This research proposes a method for machining quality monitoring (MQM) in laser-assisted micro-milling (LAMM) of glass. In tool-based mechanical processing including LAMM, the machining quality is generally affected by machining parameters and tool condition; therefore, previous studies have intensively focused on finding optimal machining parameters and monitoring tool condition to secure machining quality. However, prior work has not considered the degradation of machining quality over time. Furthermore, previous studies have manually designed features from sensory signals; these approaches are difficult to be applied without prior domain knowledge in LAMM of glass. In LAMM, MQM is more important than it is in metal cutting because glass materials are likely to have cracks from the mechanical contact between the workpiece and the tool. In this research, we employ a novel image-based deep transfer learning method for MQM in LAMM of glass. Our approach is based on a pre-trained model trained on a large-scale image dataset; this model is equipped to extract meaningful features from the images. To visually reflect the machining quality, we propose a multi-layer recurrence plot (MRP) that enables the cutting force signals to be transformed into two-dimensional images. From the experimental validation in this research, the proposed MQM method is found to have the best classification accuracy of machining quality, as compared to other existing methods. The proposed method is expected to predict the machining quality of the micro-milling of glass in advance with improved accuracy before the machining quality is degraded.