The Audio Feature’s Characterization Of The Operating State Of The Power Transformer And Research On Extraction Method

With the development of electric power industry and the expansion of capacity, we put forward higher requirements on the security and stability of power system. Power transformer is an important part of power system operation, and it is a vital equipment to establish a link between the different voltage levels. The stability and reliability of the transformer operation which has great influence on the security and stability of power system, is of great significance for the entire power grid. In the process of power transformer in the long run, winding fault and core failures are the most common types of transformer failures. In order to carry out on-line monitoring of transformer and fault diagnosis of anticipation, to abandon the risky transformers,it is necessary for us to carry out a method of a timely response to the operating state of power transformer.This thesis compared several common methods of on-line monitoring at home and abroad. Vibration analysis has advantages with fast, accurate, low-cost of running the reaction transformer. This thesis is based on vibration signal, discussing how to create the signatures model about various types of state and construct a new method to extract audio features.From the structural principle of the transformer, this thesis discussed the sources of tank wall vibration and the transmission of vibration. We established a winding mechanical model and calculate changes in the winding clamping force in order to explore the relationship between the state and the pressing force of winding; And from the field concept,we studied the electromagnetic core and influencing factors. Studies have shown that vibration signal of iron core and winding in different operation condition has an obvious spectral characteristics.In order to observe the spectral characteristics clearly,the thesis completed fast fourier transformation(FFT) of the data which collected,and completed the spectrum analysis,using the digital filter to remove invalid harmonic components,and established envelope lines of the harmonic wave which could reflect the core and winding fault,and we extracted audio features,and construct signatures model under different fault types. By studying laws of audio feature under different operating conditions,we could find the theory to judge the fault type and develop overhaul programs.On this basis,we designed an audio collection instrument which contained vibration sensors and MPS-140801-IEPE signal acquisition card and compiled a human-computer interaction program which based on LabVIEW.We realized the real-time audio signal acquisition and later signal analysis of transformer.This article relies on innovation of science and technology project of Jilin province electric power maintenance company.We have completed three live working in Hexin500kV substation,conducted signal acquisition experiments on1#Main C phase which on load operation,and collected core and winding audio signal of transformer under different conditions.By using the method of feature extraction which mentioned previously,we got the audio features of law of various fault types, and verified the feasibility and effectiveness of the method.