LAUSR

The fault diagnosis of motorized spindle contributes to the improvement of the reliability of computer numerical control machine tools. Presently, numerous mechanical fault diagnosis technologies suffer from the drawbacks of mode mixing, non-adaptive analysis, and low efficiency. Therefore, adopting an effective signal processing method for fault diagnosis of motorized spindle is essential. A method based on modified empirical wavelet transform (EWT) and kernel principal component analysis (Kernel PCA) is proposed. A new method, which determines the proper number of the Fourier spectrum segments, is applied when using EWT. To improve computational efficiency, Kernel PCA is adopted to reduce dimension. The support vector machine optimized by genetic algorithm is introduced to accomplish fault identification. The performance of the proposed method is validated through single and compound fault experiments. Results show that the recognition rate using the proposed method reached 98.8095 % and 98.4375 % in terms of single and compound fault diagnoses, respectively. Moreover, compared with empirical mode decomposition (EMD), ensemble empirical mode decomposition (EEMD), local mean decomposition (LMD) and EWT, the proposed method can save much computing time. The proposed method can be generalized to other mechanical fault diagnoses as well.