372 / 2022-03-15 17:32:04

Condition assessment of transformer winding based on the granular complex network of vibration signals

condition assessment,power transformer,vibration signals,fuzzy information,winding deformation

Power transformers are one of the essential and expensive equipment in power system. Among all the failures for an in-service transformer, winding deformation always remains the highest failure rate, which is mainly caused by the fault of outlet short-circuit. Hence, it is essential to timely and correctly assess the winding condition after the short-circuit impacts to ensure the reliable operation of power transformer. In view of the close relations of vibration signals of transformer tank with the mechanical condition of transform winding, this paper presents a method to describe the fluctuation trend of vibration signals under sudden short-circuit impact based on the granular complex network. And modularity of the granular complex network is calculated to identify the winding condition of transformer for high accuracy.

The short-circuit impulse test of a real transformer with rated voltage of 110kV was made for different short-circuit currents. Several vibration acceleration sensors were placed on the transformer tank to collect the vibration signals with the sampling frequency of 50kHz. The short-circuit impedance (SCI) of transformer was also measured after each short-circuit impulse test. When the SCI of transformer exceeds the limited value, the test was stopped.

To better describe the fluctuation trend of transient vibration signals of transformer under short-circuit impact, the fuzzy information granulation (FIG) with fuzzy C-means clustering algorithm is first built with its time envelope. According to the mapping between the FIG and the node and its edges, the granular complex network is obtained for the vibration signals with different short-circuit currents with the community structure divided. It is seen that the community structure and its modularity of the granular complex network varies apparently with the different short-circuit currents and mechanical condition of transformer winding.

The patterns of transient vibration signals resulted from the winding vibrations under short-circuit impact are closely related to the variation short-circuit currents and winding condition. The granular complex network can effectively describe the fluctuation trend of transient vibration signals. Compared with the SCI, the modularity of community structure of the granular complex network of transient vibration signals can better assess the winding condition with high accuracy.