Tool wear monitoring by estimating milling torque and calculating the wear energy coefficient under variable depth of cut conditions.

Abstract

Tool wear monitoring (TWM) is important for machining accuracy and workpiece surface quality, the electric current signal method can't realize the visualization of TWM in variable cutting depth cutting, and the force signal method has difficulties in data acquisition. Combining the characteristics of both methods, the paper proposes a novel TWM method by estimating milling torque and constructing wear coefficient. Firstly, a time-shifted modal decomposition method (TSMD) is proposed to remove the high-frequency interference from the current data, and extract the milling torque component from the feed current. Then, a prediction model driven by current data is established, which estimates instantaneous torque using the residual-enhanced dual embedding channel attention model (RL-EDCA). Finally, the tool wear energy coefficient is proposed based on the characteristics of machine-tool power consumption, its overall change trend is consistent with the trend of tool wear amount as the experimental result, which realizes the visual monitoring of tool wear state in curved surface machining, and provides a new framework for TWM in small batch production.