Feature Recognition of Oil Immersed Transformer Winding Looseness Based on Chaos Theory

摘要

In order to identify the characteristics of oil immersed transformer winding looseness more accurately and effectively, the fault simulation experiments of 110kV transformer winding looseness in different degrees under rated voltage are carried out, and the chaos theory is applied to analyze and study the vibration signal of transformer oil tank. C-C algorithm is used to select the best delay time and embedding dimension of vibration signal, and to reconstruct the phase space of vibration signal. According to the phase track curve of winding looseness in different states, it can be found that the vibration signal of transformer is distributed in ellipsoid shape in high dimension phase space. When the state of the winding changes, the phase trajectory also changes, and the larger the degree of winding looseness is, the greater the degree of opening the phase trajectory along the space is. There is a positive correlation between the opening degree of phase locus and the degree of winding looseness. In the experiment, this feature has strong repeatability. Through this feature, the loose defect of oil immersed transformer winding can be identified.