Method for extracting feature vectors of core vibration signals of distribution transformer

Abstract

The invention relates to a method for extracting the feature vectors of the core vibration signals of a distribution transformer. First, when the transformer is idle, a distribution transformer vibration signal collection device is utilized to collect the core vibration signals of the distribution transformer in different states. Bispectrum estimation is carried out on the collected core vibration signals by a direct method so as to obtain an initial feature matrix (as shown in the description). As elements in the obtained matrix (as shown in the description) are all plurals, the mods of all the elements of the matrix (as shown in the description) need to be obtained before singular value decomposition is carried out, that is, the amplitudes of all the elements are calculated, so that a real matrix (as shown in the description) of a bispectrum amplitude matrix is constructed. Finally, singular value decomposition is carried out on the elements (as shown in the description), and the two-dimensional bispectrum amplitude matrix (as shown in the description) is reduced to a one-dimensional one. The feature vectors of the core states are judged according to the construction of the former 20 obtained main singular values. The invention aims at accurately and effectively diagnosing the running state of the core in the running process of the distribution transformer, thereby more reasonably making a state detecting strategy for the distribution transformer and improving the running reliability and stability of the distribution transformer.

Description

technical field [0001] The present invention relates to the technical field of fault diagnosis of distribution network transformers, in particular to a method for extracting characteristic quantities of vibration signals of distribution transformer iron cores using bispectral singular value decomposition method, thereby realizing fault diagnosis of distribution transformer iron cores, especially A method for extracting feature quantities of distribution transformer core vibration signals. Background technique [0002] The maintenance methods of distribution transformers can be divided into three types: after-the-fact maintenance, planned maintenance and condition-based maintenance. Due to the low attention paid to the distribution network by domestic power companies in the early stage, the maintenance of distribution transformers adopts a management mode that combines post-event maintenance and planned maintenance. This maintenance management mode has a certain scientific r...

AI Technical Summary

Problems solved by technology

[0015] The two traditional methods of oil chromatographic analysis (DGA) and electrical measurement have a good effect on judging the multi-point grounding fault of the iron core, but they are helpless for the defects in the mechanical structure of the transformer that may be induced by the loosening and deformation of the iron core clamp.

In addition, electrical measurement requires transformer power failure detection, which belongs to offline detection, and the real-time performance is not high.

The "power transformer iron core loose fault diagnosis and fault preliminary location system and method" disclosed in the invention patent can only judge the loose fault of the iron core, and cannot identify other types of faults in the iron core

Images