Automatic Identification of Axis Orbit Based on Normalized Pattern Spectrum

This paper reports the results of a study of normalized pattern spectrum and its application on automatic identification of axis orbit. The normalized pattern spectrum is defined by morphological openings and closings. The comparability of two kinds of pattern spectrum is measured by a similarity coefficient. Through computing similarity coefficient, the pattern identification is accomplished. Analysis results show that normalized pattern spectrum has distinct advantages: no sensitive to image locality, image reversal, image scale. Shape representation of axis orbit is one of the most important methods on fault diagnosis of rotating machine. It is verified that axis orbits, with certain morphological processing before computing the normalized pattern spectrum, can be recognized effectively.