Vibration singularity analysis for milling tool condition monitoring

Tool condition monitoring (TCM) is a very effective way to enhance productivity and ensure work-piece quality. This paper introduces a cutting condition independent TCM approach for milling with vibration singularity analysis. Holder Exponents (HE) are chosen as index to estimate the singularity of vibration signals. Wavelet transform modulus maxima (WTMM) are employed to estimate HEs. A wavelet basis selection method is established to select the optimal wavelet bases for the estimation of HEs. Means of HE values and Number of singular points of feed direction vibration components are found to be the most correlated with tool conditions based on experimental study. The sensitivity between these HE features and the transition points of different tool wear states is discovered. Verified by a public database, this sensitivity is found independent of the cutting conditions. Then a TCM approach is proposed which utilized a Support Vector Machine (SVM) model and a transition point identification method (TPIM). Experiment results indicate that this approach is efficient and the TPIM helps to reach more precise classification results.