Normal vibration signal of rotating machinery is low-frequency signal, yet high-frequency component would exist when having fault, especially when the fault is caused by mechanical shock. But the shock pulse is very weak compared to the low-frequency component. Demodulated resonance technique can filter the low-frequency component and keep the high-frequency component. Then generate resonance wave in which the shock pulse is amplified. The signals from the gearbox are coupled to each other which make the fault diagnosis difficult, with multiple sources decoupling technique can detect independent signal of each measuring point. First, construct relative influence coefficient of the system, and then calculate the gain matrix. Adjust the parameters of the matrix repeatedly to eliminate the impact from other points. Finally, the amplified independent shock signal is detected through which the fault can be identified. The application approves that this method is effectively in fault diagnosis and fault location.
[1]
HuangTao,et al.
Fault diagnosis of rolling bearing based on the energy operator demodulation approach
,
2010
.
[2]
S.J. DiMaggio,et al.
Basic system identification for condition monitoring of turbopumps [rocket engines]
,
2001,
2001 IEEE Aerospace Conference Proceedings (Cat. No.01TH8542).
[3]
Umberto Meneghetti,et al.
Application of the envelope and wavelet transform analyses for the diagnosis of incipient faults in ball bearings
,
2001
.
[4]
Y.-T. Su,et al.
On initial fault detection of a tapered roller bearing: Frequency domain analysis
,
1992
.
[5]
Jiancheng Song,et al.
Characteristic analysis of asynchronous motor fault signal based on wavelet packet decomposition
,
2011,
2011 International Conference on Electrical Machines and Systems.