Detection of cutting tool faults in milling using empirical mode decomposition

This thesis presents detection of local tool faults in milling using vibration-based Empirical Mode Decomposition (EMD) approach. Firstly, the statistical parameters of vibration which are used as the basis for fault detection in cutting tools (RMS, Crest factor, and kurtosis) are explained and the method of empirical mode decomposition are detailed. Vibrations resulting from healthy and faulty cutting tools are analysed using conventional approaches and proposed method (i.e. empirical mode decomposition). It has been found that conventional time and frequency domain approaches remain unsatisfactory for the detection of local fault, whereas fault symptoms are clearly detected on signal components decomposed by the empirical mode decomposition.