Current Signal Based Continuous On-line Tool Condition Estimation in Face Milling

In this paper, an online method for estimation of flank wear of cutting tool based on the measurement of spindle motor current is proposed for the face milling operation. Sensors for this signal are easier to install and maintain and are also significantly inexpensive compared to those for the cutting forces, the vibration and the acoustic emission signals. A novel combination of signal processing strategies, such as line frequency estimation, demodulation, segmentation and exponential smoothing is proposed for on-line computation of appropriate features from the measured signals. Multiple Linear Regression model, formulated in terms of the training features, is then used to estimate tool wear. The developed model is validated on testing feature set for its predicting abilities. From analysis of the experimental data, it is seen that the proposed method compares favorably with those using measurements of cutting forces and Artificial Neural Networks. Finally, probabilistic worst case prediction limits of tool wear are presented.