Vibration singularity analysis for milling tool condition monitoring

•A wavelet basis selection approach is proposed. The wavelet basis with 2 vanishing moments is efficient to analyze the singularities of vibration signals in milling.•Means of Holder Exponents and quantities of singular points are sensitive to the transition points of different tool wear states. This sensitivity is independent of values of flank wear, milling tools, material of workpieces, and processing parameters.•A tool condition monitoring approach is proposed which utilized a SVM model and a transition point identification approach (TPIA). The proposed approach uses outcomes of TPIA to modify the classification results of SVM model, which can improve the classification rate of the SVM model by 8.9%.