Application of morphological filter and correlation dimension in the fault diagnosis of rotor system

To analysis the effect of noise in signal to correlation dimension, the correlation dimension of the noised Lorenz signal was calculated. Morphological filter is introduced for de-noising the vibration signal. The correlation dimensions on the noised and de-noised vibration signals of rotor system at various states were analyzed. The result showed that the correlation dimension of the noised signal deviate from their typical value and cannot reflect the true status of rotor system. The favorable de-noising effect by morphological filter is also showed. The correlation dimension of de-noised signal can reflect and distinguish the true status of rotor system, while it can be used as characteristic quantity for identification of fault and estimation of status.