Condition Assessment of Transformer Winding Through the Granular Complex Network of Vibration Signals

Vibration signals of transformer tank under the outlet short-circuit show the features of nonstationary and nonlinear, and contain abundant information of the mechanical condition of transformer winding. To investigate the fluctuation trends of transient vibration signals of transformer, the granular complex network (GCN) is built based on the envelope of time domain of transient vibration signals and the Fuzzy C-means algorithm. Then the degree distribution of the GCN are calculated to recognize the mechanical condition of transformer winding. The short-circuit impulse test of a real transformer with rated voltage of 110kV was made for different short-circuit currents to obtain the transient vibration signals. The calculated results have shown that the GCN is capable of describing the key and hidden information of the transient vibration signals. The degree distribution can clearly illustrate the deterioration process of mechanical condition of transformer winding.