Vibration features extraction of power transformer using an time-scale-frequency analysis method based on WPT and HHT

The success of health monitoring and condition assessment for power transformer with vibration analysis lies on proper extraction of vibration features. The extraction of features in turn depends on appropriate signal processing methods. In this paper, a WPT and HHT based time-scale-frequency analysis is developed to extract the vibration features from power transformer tank. In this method, the WPT is adopted firstly to decompose the signal into a set of narrow band signals. Prior to HHT, the decomposed components are screened with a CTS technique. Finally, HHT is applied to these components to obtain both frequency and amplitude characteristics. By comparing with the single WPT and HHT methods, the results show that the time-scale-frequency analysis can overcome the intrinsic shortcomings of these two methods, which will lead to unreliable vibration features for ignoring frequency information in wavelet-based method, mode mixing and meaningless IMFs generation resulting from EMD in HHT. Therefore, the developed method can provide a better alternative to implement vibration feature extraction from vibration signals of power transformer.