Abstract: From the angle of vibration signal complex for rotation machine,the construction of characteristic entropy is studied and a set of multi-scale fuzzy entropy is put forward to identify the vibration signal with different faulty.Accounting for theory defect of traditional faulty identification which depends on prior knowledge,the identification method based on variable predetermination model is instructed into faulty diagnose for rotation machine.The new method is applied in a living example,which show the method is effective.

Key words: multi-scale fuzzy entropy, rotation machine, Variable Predictive Model Based on Class Discriminate(VPMBCD), modal identification