Features of spectral vibration diagnostics of traction power transformers in high-speed motion

Abstract

This paper uses spectral analysis methods to develop a mathematical model and algorithms for evaluating the diagnosed signals corresponding to certain defects of windings and magnetic circuits of traction power transformers. A distinctive feature of the study is at certain frequencies of spectral dependences, taking into account the effects associated with defect formation that have a different cause of their manifestation, for example, wide-field fluctuations and changes in the shape and width of spectral peaks. The presence of random fluctuations caused the application of the normal Gaussian distribution law of the measured values of the spectra. To solve the problem of identifying the diagnosed spectra corresponding to certain defects of the power transformer, algorithms for identifying the vector of diagnostic features based on the method of statistical recognition theory for large amounts of information are proposed.