Through the modelling, simulating, experimental research of the power transformer winding, the coil vibration signals under different clamping forces are analysed; some important features which can help disclosing the transformer winding looseness are proposed, including two asymmetry features, the axial vibration amplitudes distribution feature and nonlinear feature which represents the disorder of the vibration signal. The MDOF model which is based on the classical spring-mass model and counts the hard nonlinearities of the pressboard is established firstly, then the multiple dimensional MODF model was introduced to derive the global asymmetry feature. Combined with the winding inner properties, the MDOF model which is only counting the static nonlinearity of the winding is simulated to derive the top/bottom asymmetry feature and axial vibration distribution features. Further, by using the simplified SDOF nonlinear model, the nonlinear feature including the 200Hz frequency components and the entropy of the recurrence histogram are derived. At last, the experiment vibration signals sampled from two similar 35KV power transformers are analysed in detail to examine the proposed features; the results show a consistency between the simulations and the experiments and verify the proposed features. The proposed features can be a good complement to the fault diagnosis of the winding by vibration methods.
展开▼