Monitoring of drill runout using Least Square Support Vector Machine classifier.

• Realized a monitoring system for runout in drilling. • A novel vibration and force measurement approach is used. • Identified runout sensitive features of vibration and force signals. • A LS-SVM model is developed for runout prediction with greater accuracy. • Timely prediction of runout will improve dimensional accuracy of component. Runout is a critical problem in drilling processes, which affects tool life, geometrical tolerances and also results in increased machining cost. Runout is out of balance of the drill that causes higher vibrations at rotational frequencies and lower cutting forces in the axial direction. The objective of this study is to predict and classify the state of runout. Timely prediction of runout facilitates remedial action and improves the quality of components. A novel vibration-force based multisensory approach with LS-SVM classifier for runout monitoring of a Computer Numerical Control (CNC) drilling process is presented in this paper. The experimental study shows that the runout has a significant effect on the frequency components of vibration and force signals. A Fast Fourier Transform (FFT) analysis extracts features of vibration magnitude at 1x RPM and magnitude of axial force at 100 Hz that indicates the unbalance which is proportional to the severity of runout. An increase in amplitude of vibration signal accompanied by decrease in amplitude of force signal indicates the existence of runout. A Least Square Support Vector Machine (LS-SVM) classifier is used for modelling the runout and the model is able to predict the presence of runout with R2 of 0.99 and classifies the different states of runout with a prediction accuracy of 80%. [ABSTRACT FROM AUTHOR]