Condition-based maintenance (CBM) has become a hot research topic in recent decades. For enhancing the CBM ability, many researchers pay attention to the incipient fault detection and remaining useful life (RUL) prediction. Detecting faults earlier can provide enough lead time to manage the maintenance actions. Besides of this, predicting RUL more accurate can avoid the over maintenance and utilize the components’ life sufficiently. As we know, good degradation features are very important for RUL prediction. However, extracting a monotonous degradation feature is easier for stationary condition than non-stationary condition. In order to address above two issues, this paper uses a new signal processing method named narrowband interference cancellation (NIC) to do degradation analysis. It can extract periodic impulse signals from very noise background which mainly contains narrowband signals. So, the incipient faults can be found easier. The energy ratio calculated based on NIC can be used as degradation feature under non-stationary condition. To verify the effectiveness of NIC based processing, two gearbox run-to-failure tests were conducted. One is under relative stationary condition; the other is under non-stationary condition. The results of these two experiments demonstrate the validity of proposed methods.
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|