Tool Wear Prediction Using Multiple Regression in Turning


  • Lee Woon Kiow


feature extraction, tool wear, regression, vibration, wavelet decomposition


Cutting tools are subjected to an extremely severe rubbing process. They are in metal-to-metal contact between the chip and work piece, under high stress and temperature. This situation cause the inconsistencies and unwanted effects on the workpiece and cutting tools such as flank wear. Flank wear may lead to the decrease of the accuracy of produced parts, finishing surface, and economics of cutting operation. The objective of the study is to extract feature from the vibration sensor signal based on different cutting conditions, then evaluate the accuracy of the tool wear monitoring method from the supervised learning technique after wavelet analysis. Experiments have been conducted for measuring tool wear based on the factorial design technique in a turning of AISI 1045 Steel using carbide insert. Then, the cutting conditions and statistical features from vibration signal in time domain and generated from the wavelet decomposition were used as the input of regression model corresponding to the output, flank wear for the tool wear prediction by using MATLAB. The tool wear prediction of Model V has 79.13% of accuracy in the correlation between cutting condition, vibration signal and flank wear which show that the vibration signal generated from the wavelet at higher frequency level is more sensitive to predict flank wear as the accuracy is increased compared to the vibration signal in time series.




How to Cite

Kiow, L. W. (2023). Tool Wear Prediction Using Multiple Regression in Turning . Research Progress in Mechanical and Manufacturing Engineering, 4(1), 19–29. Retrieved from