Transmission gear fault feature extraction by using fuzzy entropy

Aiming at the abrupt problem of sample entropy similarity measure function, a gear in transmission fault feature extraction method based on fuzzy entropy was proposed. Fuzzy entropy replaced the hard threshold criterion in sample entropy by introducing fuzzy membership function, thus reducing the sensitivity and dependence of fuzzy entropy on parameters. Fuzzy entropy was extracted as the characteristic value of transmission gear failure, the tested signal types were normal, slight-worn, medium-worn and broken-teeth. Due to the different fault type correspondence with different fuzzy entropy, it was used as fault feature to evaluate the different fault condition. Practical results proved that fuzzy entropy had better classification ability than sample entropy.