Feature Analysis and Diagnosis of Planetary Gear and Bearing Fault in Generator Sets
In response to the problem of multiple types of faults and difficulty in identification in generator sets,the vibration acceleration signal characteristics of generator sets were analyzed to obtain the vibration signal characteristic laws caused by gear faults and bearing faults.Based on this,a multi fault diagnosis method combining feature extraction technology and deep neural networks was proposed.The results indicate that these two types of faults significantly increase the time-domain and frequency-domain amplitudes of the signal,and the first dominant frequency of the vibration signal increases from 6.0 Hz to 73.0 Hz.Gear faults mainly affect acceleration levels within the range of 6.3～16 Hz,while bearing faults significantly affect acceleration levels within the range of 630～10 000 Hz.In addition,feature extraction can improve accuracy by 7.33%,and the fault diagnosis method achieves an accuracy of 98.33%in the test set.The research results are of great significance for fault diagnosis and maintenance of generator sets.
NETL
NSTL
万方数据
