Abstract: Real-time sensing of tool conditions in cutting through artificial intelligence and industrial big data is an important technical way to realize performance-oriented manufacturing, which is also a key connotation of high-performance manufacturing. However, in the current tool condition monitoring (TCM) algorithm, the feature extraction still relies on manual experience, which limits the application of TCM in high-performance manufacturing (HPM). Therefore, a TCM method for HPM is proposed based on feature adaptive fusion and ensemble learning techniques to ensure autonomy and accuracy in the monitoring of HPM. The proposed fusion method automatically assigns weights to the extracted features based on their performance, thus achieving adaptive feature fusion. AdaBoost algorithm is also used to ensure monitoring accuracy while automatically fusing features. The milling experiments of thin-walled parts show that the maximum RMSE and MAE values between the monitoring results and the actual results are 10.44, and the minimum is 5.16. The proposed method monitors the tool wear in the machining of aerospace thin-walled parts autonomously and accurately, which solves the problem of manual experience dependence in the condition monitoring of HPM milling tools.

Key words: high-performance manufacturing, tool condition monitoring, feature fusion, thin-walled parts, milling