| Popis: |
This paper describes a vibration spectral analysis technique -Coherence Phase Line Enhancer (CPLE)for rotating machinery health monitoring and diagnosis. Unlike the conventional method for Power Spectral Density (PSD) function estimation, the CPLE technique incorporates phase information into the spectral estimation process. This is achieved through detection of a unique coherent phase relationship associated with all speed-related signal components in a "pseudo" wave-number domain. Accordingly, the CPLE spectrum can enhance and detect well-hidden speed-related signatures (bearing, gear, rotor instability, etc.) that are smeared in its PSD. As a result, the CPLE spectrum provides a unique ability to separate vibration energy that is generated by speed-related mechanisms from all other independent vibration sources. Linking CPLE spectra as the input feature vector to a neural network-based pattern classifier provides a superior detection/classification capability to the PSD. Compared to other filtering techniques for signal enhancement (e.g. Adaptive Line Enhancer), the CPLE method offers superior detection capability without suffering from any filter-instability problem. The signal processing of CPLE is simple, robust, and highly suitable for real-time implementation. The significance of this new method is attributed to its enhanced capability to cope with severe operational environments where health-monitoring measurements are heavily corrupted by background noise, thereby improving incipient fault detection capability. |
