Research on the Timing of Replacing Worn Milling Cutters by Using Wear Transition Percentage Constructed Based on Spindle Current Clutter Signals.

The use of worn cutters not only reduces the machining accuracy but also increases the surface roughness. Therefore, it is important for enterprises to establish replacement rules for worn cutters. However, traditional wear regression studies require frequent shutdowns to measure tool wear as training samples. This undoubtedly increases the complexity of operations, making it difficult to apply in practical production. To address this issue, a novel method based on the wear transition percentage has been proposed to determine the optimal timing of replacing worn tools. This method does not require measuring tool wear and is suitable for different machining parameters. Firstly, the Vold-Kalman filter is employed to remove the rotation frequency and its harmonic components from the spindle current, resulting in spindle current clutter signals (SCCS) with low correlation with cutting parameters. Then, using convolutional neural networks (CNN) to learn the SCCS data features of severe wear and normal wear stages, a binary classification CNN model is obtained. Finally, the model is used to identify the full life SCCS data with different cutting parameters. The proportion of samples identified as normal wear to all samples during a certain period of time is used to calculate the wear transition percentage. The effectiveness of this method is verified by comparing it with the measured flank wear.