Research On Monitoring Of Transformer Winding Condition Based On Vibration Method

In recent years, the fault diagnosis of power transformer windings has gradually become a hot research at home and abroad. The state of the running transformer windings can be reflected by the vibration signal on the surface of the transformer’s tank. But the signal is the blended by the signal of the core and winding, which can not accurately reflect the fault state of the transformer winding. The mixed vibration signal is analyzed by modern signal process technique. The fault diagnosis is real-time according to the effectively characteristic signal extracted from the running transformer. The transformer vibration method adopted the non-intrusive and non-electrical connection detection methods, so the collection method is simple and the data is accurate. It has wide application prospect.This paper is studied the vibration’s principle and propagation mode of the windings and iron core which is the main vibration source in the transformer. The windings are affected on the electromagnetic force due to the leakage magnetic field of the transformer windings. It is analyzed the value and source of electromagnetic force, and the formula is derived. According to the formula, it can be concluded that the current in the windings is direct proportion to the electromagnetic force and the vibration’s base frequency is twice the line frequency. The iron core’s vibration is because of the magnetostrictive effect. The vibration acceleration formula of the iron core can be inferred. It shows that vibration acceleration is direct proportion to the votage and the vibration’s base frequency is twice the line frequency. The vibration signal acquisition system is constructed including the piezoelectricity acceleration sensor and acquisition and processing of vibration signal based on the vibration characteristics of the surface of the oil tank. The vibration signal is collected on the 220 kV transformer in the no-load test, load test and online test. The result basically coincides with the theoretical analysis in the no-load and load test, which verifys the feasible of the vibration method in transformer’s state inspection. It is confirmed that the 1/2 position of the low voltage side is the final point by the analysis of vibration characteristics in the different points of the transformer tank on the basis of iron core and windings’ vibration signal characteristics online. According to the vibration signal of three different years of 220 kV running transformer, it is concluded that the vibration amplitude is larger and the higher harmonic gradually occupy the main frequency segment as with the year’s rising.The sub-band decomposition independent component analysis(SDICA) is adopted to verify the theoretical simulation signal on the online transformer tank. And then the signal of the actual running transformer is disassembled by SDICA gaining the windings signal and iron core one respectively. Finally, the energy entropy of the vibration signal is calculated by complementary ensemble empirical mode decomposition(CEEMD), which is the feature information used in the state inspection and fault diagnosis. The 35 kV transformer is experimented in the short circuit test and man-made winding failure getting the data before and after the tests. The computed entropy of the data shows that CEEMD energy entropy can extract the feature information reflected the state of the transformer form the winding vibration signal. The state of the transformer’s winding can be detected by SDICA which can be applied to the winding’s condition monitoring and fault diagnosis for the different transformer..