One-dimensional Residual Neural Network-based for Tool Wear Condition Monitoring

Tool wear is an essential factor affecting the machining process of a machine tool. Precisely predicting tool wear condition can not only improve machining efficiency but also effectively reduce the risk of workpiece quality degradation and machine tool damage. In recent years, deep learning has been applied to tool wear condition monitoring, and improving the accuracy of model prediction is a complicated process. In this paper, the cutting force signal was used as an input to the one-dimensional residual neural network to predict the tool wear condition. A model pre-trained method was proposed to improve the prediction accuracy, and after pre-training the model, the prediction accuracy is increased from 86.76% to 91.16%.