Tool wear prediction in high-speed turning of a steel alloy using long short-term memory modelling

Abstract

Abstract Tool flank wear prediction is an important factor in the machining process to guarantee processing quality and machining efficiency. Therefore, monitoring tool wear is vital to obtain a better machined surface and the lowest manufacturing cost. This paper proposes a prediction model for tool flank wear during the machining of a steel alloy derived from long short-term memory (LSTM) modelling. The LSTM model was tested using spindle motor current signals collected during experiments performed on a turning machine. The root mean square error (RMSE) was calculated on test data set for the LSTM models having one, two, and three layers and 1–10 hidden units. The results show that the most accurate LSTM model contained two layers and eight hidden units and has a test RMSE value of 0.00475. The LSTM model demonstrates its capability to capture tool flank wear during the machining process.