Abstract: Aiming at the problem of degradation degree description performance degradation stage division for rolling bearings, the variation pattern of spectrum entropy in complexity evolution was studied by using the Logistic chaos mapping sequence. A division method of rolling bearing degradation stages based on root mean square, spectral entropy, “bending time parameter” and gathgeva（GG） fuzzy clustering was proposed. The example analysis was carried out and the life test data from the IMS bearing test center. The results show that reflect the complexity evolution tendency is able to be reflected by spectrum entropy which has a advantage of sensitive to variation and fast calculation speed. The continuity of the same state on the time scale is able to be described by introduced Curved Time parameter which is more according to performance degradation pattern for mechanical equipments. The degradation conditions of mechanical equipment such as bearings can be divided accurately by GG fuzzy clustering.
Key words: spectrum entropy（SE）; gathgeva（GG）fuzzy clustering; rolling bearing; feature extraction