Multi-domain and multi-scale analysis of milling signals from discrete-edge end mills and their processing methods

Abstract With the rapid development of Industry 4.0 and the increasing digitalization of manufacturing processes, tool condition monitoring has become a critical technology for ensuring machining quality and efficiency. Compared with traditional Conventional end mills, discrete-edge end mills exhibit significant advantages in chip breaking, vibration reduction, and heat dissipation due to their unique intermittent cutting mechanism. However, this structural characteristic also causes the milling signals generated during the cutting process to exhibit stronger non-stationarity, nonlinearity, and complex transient impact characteristics. To address the above issues, this study proposes a novel complexity analysis method for vibration signals of discrete-edge end mills by integrating the proposed CRSN architecture with multiscale fuzzy entropy (MFE), aiming to achieve a comprehensive and accurate representation of the latent state information embedded in the signals. Furthermore, a dedicated signal processing approach, termed VI-CNN, is developed for discrete-edge end mills milling signals, providing a reliable basis for high-precision tool condition monitoring. Experimental results demonstrate that the proposed CRSN network can effectively extract representative features from discrete-edge end mills vibration signals, thereby validating the intrinsic complexity of the milling process. In addition, the proposed VI-CNN signal processing method is capable of reducing signal complexity while preserving key structural and physically meaningful features.