Tool Wear Status Identification Method Based on Tdistributed Stochastic Neighbor Embedding and Gradient Boosting Decision Tree Integrated Model

Online identification of tool wear status is an essential function in the process of automated machining, and accurate identification of tool wear status is the key to improving machining quality. A novel method of tool wear status recognition relying on t-distributed stochastic neighbor embedding (t-SNE) downscaling and GS-GBDT integrated model was proposed. First, tool wear signals were collected by sensors and combined with data processing to clean the data. Then, t-SNE was used to downscale the features, and combined with exponential smoothing and order-preserving regression to optimize the downscaled features. Finally, grid search (GS) was used to optimize the parameters of the gradient boosting decision tree (GBDT) algorithm, and the proposed model was evaluated by combining various assessment indexes. The results indicate that the proposed approach has high accuracy in prediction and can serve as a technical basis for monitoring the tool wear status in machining processes.