Gear Faults Diagnosis Based on Wavelet Packet and Fuzzy Pattern Recognition

Abstract

The combined wavelet packet and fuzzy pattern recognition method was used to diagnose the gear faults. Firstly, we use wavelet packet db3 (Daubechies3) by level 4 to decompose and analyze the measured vibration signals from a test gearbox. Secondly, the energy of wavelet packet coefficients was calculated as the characteristic values, by means of which the target fuzzy samples and testing fuzzy samples was set up. The fault condition of the gear was classified and identified with the testing fuzzy samples. A compare between being done and without being done wavelet packet analysis before the method of the fuzzy pattern recognition was also carried on. Results show that the combined using of wavelet packet analysis and fuzzy pattern recognition on faults diagnosis with vibration signals can gain good effects to search and read papers conveniently.