Signal periodic segment processing technique for diagnosing gear pairs faults

This thesis presented a technique for diagnosing gear pairs faults and introduced the procedure in detail. When there is a fault in the gear pairs, there will be some pulse (s) in the vibration signal which is picked with high sampling rate on the gearbox. Firstly, the vibration signal was processed by time domain averaging to weaken noise. Then the average signal was divided in equal segments with the same number of the gear teeth which was to be diagnosed in the gear pairs. The analysis of individual segment reveals more sensitive changes of the vibration signals in both time and frequency domain caused by gear faults. Further more, the location of the failed gear tooth can be identified in terms of the position of the segment which is different from normal segment. Both the numerical simulated signal and the experimental investigation verify the availability of the signal periodic segment processing technique.