Transformer extended Debye model parameter identification method based on K-K transformation and spectral line differentiation

Abstract

The invention provides a transformer extended Debye model parameter identification method based on K-K transformation and spectral line differentiation. The method comprises the following steps: S1, measuring the complex capacitance of an oil-paper insulation sample of a transformer oil-paper insulation system, performing K-K conversion on the real part and imaginary part of the measured complex capacitance to obtain the real part and imaginary part of the calculated complex capacitance, subtracting the measured complex capacitance spectral line from the calculated complex capacitance spectral line to decouple a geometric capacitance spectral line, an insulation resistance spectral line and a relaxation polarization spectral line, calculating geometric capacitance parameters and insulation resistance parameters through least square straight line fitting; S2, acquiring a complex capacitance real part / imaginary part differential spectral line and a micro-molecular spectral line of the polarization equivalent circuit of the polarization complex capacitance real part, and calculating polarization resistance; S3, identifying the number of polarization branches and parameters of the polarization branches based on a spectral line differential method. According to the method, the geometric equivalent branch parameters of the transformer extended Debye model can be effectively solved.

Description

technical field [0001] The invention relates to the technical field of transformers, in particular to a transformer extended Debye model parameter identification method based on K-K transformation and spectral line differentiation. Background technique [0002] At present, intelligent algorithm identification is mostly used for parameter identification of transformer extended Debye model, such as genetic algorithm, particle swarm algorithm, etc., but its randomness is strong, the chance factor is large, and there is a probability of identification error in practical applications. In the existing research, the differential decomposition spectrum method based on the polarization / depolarization current data has achieved good results, but the random selection of the end point leads to non-unique parameter identification results, and the geometrically equivalent branch parameters in the model cannot be solved. Contents of the invention [0003] The invention proposes a transfor...

AI Technical Summary

Problems solved by technology

In the existing related research, the differential decomposition spectrum method based on polarization / depolarization current data has achieved good results, but the random selection of points at the end leads to non-unique parameter identification results, and it is impossible to solve the geometrically equivalent branch parameters in the model

Images