A Diagnosis for Signature Analysis by the Correlation Integral with Re-embedding Procedure

In this paper, a method to detect some kind of mechanical failures has been proposed. According to the fact that any defect happen on the mechanical system would be inherently defined es statistical differences, the correlation exponents based on the Grassberger-Procaccia algorithm and the modified version of the Grassberger-Procaccia algorithm with re-embedding procedure could be adopted as the diagnostic index. Chattering vibration during cutting machinery has been investigated as a typical example to detect the abnormal vibration. Then admitted difference has been resulted between the two stages of normal condition and chattering on the tendency of correlation exponents. Further, In order to reduce the noise, only the principal components are used for re-embedding space and the modified version of the Grassberger-Procaccia algorithm is applied on the re-embedding space. Then the correlation exponents indicate more clear distinction between normal and abnormal condition than that of a ordinary Grassberger-Procaccia algorithm.