Vibration Monitoring of Converter Transformer by Simplified permutation Entropy

Converter transformer is one of important equipment in the HVDC transmission system. Compared with the ordinary power transformer, the vibration signals of converter transformer always exhibit relatively stronger nonstationary and nonlinear features due to the complicated mechanical structure and current harmonic components. To better analyze the vibration signals obtained from a converter transformer for the condition monitoring work, a simplified permutation entropy (SPE) algorithm is proposed in this paper. The phase space reconstruction of the vibration signal of converter transformer is first built. And then the symbolized sequence of each vector in phase space is recorded. Finally, the normalized information entropy considering sequence weight as SPE value is calculated. The results of vibration signals for a $\pm 500\mathrm{kV}$ converter transformer have shown that, SPE algorithm can effectively extract the signal features, waveform characteristics of the four measurement points are well distinguished. SPE has the advantages of stable inter-class distinction ability and high flexibility, and it also reaches a much faster calculation speed compared with traditional permutation entropy (PE) algorithm, thus can provide a new reference for vibration monitoring of converter transformer.