Tool wear is inevitable due to the thermal-mechanical coupling in micro-milling, Micro-milling of difficult-to-cut material Inconel 718 leads to significant flank wear on the cutting tool. Tool wear influences the dimensional accuracy and surface quality of products, and also influences tool life. Tool wear during micro-milling process cannot be obtained by current commercial finite element software directly. This paper presents a tool wear prediction method for micro-milling cutter based on long short-term memory network. Firstly, experiments of micro-milling Inconel 718 are carried out, the cutting forces in three directions, reduction ratio of the tool diameter and the flank wear of the cutting tool are obtained. Then, through correlation analysis, it is found that Fx has the strongest correlation with reduction ratio of the tool diameter, the correlation coefficient is 0.9390, and Fy has the strongest correlation with the flank wear of the cutting tool, the correlation coefficient is 0.9453. Finally, based on the long short-term memory network, a prediction model of reduction ratio of the tool diameter is established with Fx as input. The root mean square error (RMSE), the mean absolute error (MAE) and the mean absolute percentage error (MAPE) are 0.4246, 0.3533 and 2.92 % respectively. And a prediction model of the flank wear of the cutting tool is established with Fy as input. The RMSE, MAE and MAPE are 0.4463, 0.3649 and 0.20 % respectively. The research achieves the prediction of tool wear in micro-milling Inconel 718, which lays a foundation for tool condition monitoring.
Read full abstract