Abstract:In order to predict the wear state of the milling cutter during the milling process so as to timely find and replace the milling cutter that would be blunt, and to ensure the quality of products, the sensors was used to collect vibration signal data of CNC milling machine and milling cutter in the processing process, and the big data method was used to study the analysis and prediction method of the wear state of the CNC milling cutter. In order to ensure the recognition accuracy, recognition stability and robustness of the analysis model, the wavelet packet decomposition theory was used to denoise the vibration signal data in x, y and z directions of the milling machine, the time domain features and energy features were extracted, and 34 features were screened out which were closely related to wear state.The XGBoost algorithm was used to establish the data analysis model of the wear state of the milling cutter, and the macro average value was used to evaluate the performance of the model. Combined with SMOTE technique, the feature vector was oversampled to equalize the wear state category samples.The proposed method was verified by the open ballend milling machining data set.The experimental results show that by using the XGBoost algorithmthe, wear state data of milling cutters can be analyzed correctly and the early warning stage of milling cutter can be identified. XGBoost algorithm has high prediction accuracy, good stability and strong generalization ability, which is easy to be applied in the field of industrial big data