Rotor Crack Fault Diagnosis Based On Base And Multi-vibration Signal Fusion

The fault diagnosis method based on vibration analysis is the most common and effective method in the filed of rotor fault diagnosis. The rotor often be used directly as monitoring object in the process of rotor system fault diagnosis composed by the rotor and its supporting parts. The sensors are usually installed on the bearing seat near rotor parts or on the special sensor supporting frames to achieve sensor installation. But these methods are all very inconvenient and poor of generability in the practical application. And the rotor fault diagnosis uses the base as monitoring object is very convenient and versatility. However, current research in this area is rare because the base far away from the fault source and the vibration signals collected relatively weak. Multi-sensor information fusion can fully describe the diagnostic object because it extended spatial and temporal scales, allowing for more information. The cracked rotor fault diagnosis method based on the base and Multi-sensor information fusion is posed in paper, and studied mainly from the following aspects:1. The transient analysis of the cracked rotor tester based on sliding bearing is done by the finite element analysis software of ANSYS. The vibration characteristics of the base are analyzed. Amount the factors of it and the proposal of the sensors on the base.2. The test rig made by American Spectra Quest and the data acquisition system made by Austria Dewetron were used to experimental study the multi-sensor vibration test based on the base. Analysis the signal through the correlation function method, verify the multi-sensor layout on the base.3. The information fusion method of auto adapted weighted is studied. The weight values of each sensor on base are calculated under the premise of the minimum mean square error. Then, the confertus signals are got according to weight values. The fault characteristic value is picked up by multi-layer decomposition of wavelet analysis. The fault of cracked rotor is diagnosed effectively at last by combining the Bode diagram in the process of speed up.4. The multi-sensor information fusion method based on SVM is researched. The feature on different time domain, the signals of multi-sensors on base and the single sensor on the bearing seat, normal running and the cracked rotor with different depth are all studied by experiments, and the optimal fault diagnosis proposal is reached.In this paper, the restriction of structure and working conditions in rotor fault diagnosis were broken, the multi-sensor information fusion on base fault diagnosis of rotor crack was researched, a new reference method fault diagnosis of rotating machinery was provided.