On-load Tap-changer Fault Discrimination Based on Full Ensemble Empirical Modal Decomposition and Volterra Modeling

The on-load tap-changer is the only key component in a transformer that operates frequently and is prone to mechanical faults, which can lead to severe safety incidents. To accurately diagnose typical mechanical faults of On-Load Tap Changers (OLTC), this paper proposes a fault identification method based on Complete Ensemble Empirical Mode Decomposition (CEEMDAN) and the Volterra model. First, CEEMDAN is applied to the vibration signals of OLTC to extract vibration features. Subsequently, a Volterra model is established for each Intrinsic Mode Function (IMF) component derived from the decomposition to generate coefficient matrices. The singular values of these matrices are calculated as fault feature indicators of the OLTC. A PCA-SVM model is then constructed based on the feature vectors to classify and identify typical OLTC mechanical faults. Finally, diagnostic results show that the PCA-SVM model achieves a fault diagnosis accuracy of 97.5%, effectively identifying typical mechanical faults in OLTC.