Research On Fault Diagnosis Of Transformer Winding And Analysis Method Based On Vibration Signal

Vibration signal of transformer can reflect its mechanical properties. Extracting usefulcharacteristic quantity with the information processing technology from the vibration signal ofthe mechanical failure is the hot point of the research of fault diagnosis. Because its detectionmode is non-invasive and without electrical connection, its application prospect is very widely.In this paper, based on the vibration signal of transformer, ensemble empirical modedecomposition(EEMD) and principle of energy entropy applied to machinery fault diagnosisof the transformer winding is studied.Mechanism and transmission routes of vibration is studied and the equivalentmathematical model of winding vibration is established. The relationship between vibrationand the pressing situation of winding is obtained by stress analysis. That the clamping forcebecomes smaller caused by the winding displacement and deformation results in the windingvibrating violently. Magnetostrictive phenomenon that causes the core vibration and itsinfluencing factors are researched, too.In this paper, S11-M-500/35type power transformer is taken for the study object, no-loadtest, short-circuit test and impulse test are done, analyze vibration data and compare with theresults of the test and hanging core inspection. Relative to the traditional method for electricaltesting, mechanical vibration method has higher measurement sensitivity and can acutelyreflects changes in the status of the transformer windings. In addition, in order to get richertransformer fault samples, a variety of man-made fault windings are made and its signalacquisition platform is built, and analyze the signal samples for fault diagnosis.As a core part of this paper, it focuses on extracting characteristic information about thevibration signal of transformer winding and diagnose the fault with ensemble empirical modedecomposition. The effectiveness of the method is verified，also.