CN-121995169-A - Partial discharge quantitative calibration method based on modal decomposition and frequency weighting

CN121995169ACN 121995169 ACN121995169 ACN 121995169ACN-121995169-A

Abstract

The invention relates to the field of partial discharge quantitative calibration and electromagnetic simulation analysis, in particular to a partial discharge quantitative calibration method based on modal decomposition and frequency weighting. The method comprises the steps of obtaining structural parameters, defect excitation parameters, ultra-high frequency sensor position parameters and sensitivity direction parameters, establishing an electromagnetic model, carrying out finite difference time domain simulation to obtain frequency domain response spectrums, obtaining frequency domain component sets of each mode through orthogonal projection of mode expansion and waveguide eigenmodes, carrying out inverse transformation and wavelet transformation on the frequency domain component sets of each mode to obtain time-frequency energy distribution, extracting total energy, energy weighting arrival time, dispersion and frequency energy spectrum to generate mode characteristic quantity sets, normalizing the mode characteristic quantity sets to obtain mode quality factor weights, generating frequency related weight functions, synthesizing equivalent calibration transfer functions, and carrying out inversion discharge current and integration by combining measured voltage spectrums to obtain apparent discharge quantities. The invention can effectively improve the quantification reliability of the apparent discharge quantity.

Inventors

ZHAO KE
REN MING
XIAO HANYAN
LI YUJIE
LI XINING
YIN ZE
She Congdong
SUN LEI
WANG LIJIANG
Zhuang Tianxin
LIU JIANJUN
HU CHENGBO
LU YONGLING

Assignees

国网江苏省电力有限公司电力科学研究院
西安交通大学

Dates

Publication Date: 20260508
Application Date: 20260121

Claims (10)

1. The partial discharge quantitative calibration method based on modal decomposition and frequency weighting is characterized by comprising the following steps of: S100, acquiring GIS structure parameters, defect excitation parameters, UHF sensor position parameters and sensitive direction parameters, establishing an electromagnetic model, executing FDTD time domain simulation, extracting sensor electric field time domain signals and transforming to obtain a frequency domain response spectrum; S200, solving a mode coefficient set of a TEM mode, a TE11 mode and a TM01 mode according to orthogonal projection of a waveguide eigenmode based on a frequency domain response spectrum, and generating a frequency domain component set of each mode; S300, performing wavelet transformation to obtain time-frequency energy distribution based on the frequency domain component sets of each mode, extracting total energy, energy weighted arrival time, dispersion and frequency energy spectrum, and generating a mode characteristic quantity set; s400, normalizing the modal feature quantity set to obtain modal quality factor weight, and generating a frequency correlation weight function by combining the frequency energy spectrum; S500, inputting the frequency-dependent weight function into a transfer function synthesis module, weighting and fusing the transfer functions of all modes to generate an equivalent calibration transfer function, collecting the actual measurement voltage time domain signal and transforming to obtain a voltage spectrum, inverting the voltage spectrum and the equivalent calibration transfer function to obtain a discharge current spectrum, inverting the discharge current spectrum to obtain a discharge current time domain signal, and integrating to obtain the apparent discharge quantity.
2. The method of claim 1, wherein S100 the electromagnetic model comprises an inner conductor, a shell, an insulator, a phase separator, and the FDTD time domain simulation sets a PML absorption boundary.
3. The method of claim 1, wherein S100 the defect excitation parameters employ spike defect model parameters, the spike defect model consisting of a conductor surface tip radius of curvature, a tip height, a defect location, and a charge pulse time parameter.
4. The method of claim 1 wherein S100 said defect excitation parameters employ needle plate defect model parameters, the needle plate defect model consisting of needle electrode radius, needle plate pitch, defect location and charge pulse time parameters.
5. The method of claim 1, wherein S200 the mode expansion model constructs a set of sampling points in the GIS waveguide cross section, the set of sampling points covering the inner conductor neighborhood, the housing inner wall neighborhood, and the insulator neighborhood, and performs the eigenmode orthogonal projection based on the set of sampling points to solve the set of mode coefficients.
6. The method of claim 1, wherein the mother wavelet of the wavelet transform of S300 is a Morlet wavelet, and the time-frequency energy distribution is obtained by squaring wavelet coefficients.
7. The method of claim 1, wherein S300 the energy weighted arrival time is obtained by a weighted summation of time-frequency energy distribution over a time axis, and the dispersion is obtained by a second order central moment of time-frequency energy distribution over the time axis.
8. The method of claim 1, wherein S400 said normalizing employs a maximum minimum normalization, and wherein a normalized object comprises said total energy, said energy weighted arrival time, said dispersion.
9. The method according to claim 1, wherein the modal quality factor weights of S400 are obtained by combining a total energy normalization value, an energy weighted arrival time normalization value and a dispersion normalization value, and the combination relation includes an energy positive correlation term, an arrival time penalty term and a dispersion penalty term.
10. The method of claim 1, wherein S500 the frequency dependent weight function is obtained by multiplying the modal quality factor weights by the frequency energy spectrum and performing normalization at each frequency bin, and wherein the voltage spectrum performs masking at a frequency band where the equivalent nominal transfer function magnitude is less than five percent of its magnitude maximum.

Description

Partial discharge quantitative calibration method based on modal decomposition and frequency weighting Technical Field The invention relates to the field of partial discharge quantitative calibration and electromagnetic simulation analysis, in particular to a partial discharge quantitative calibration method based on modal decomposition and frequency weighting. Background In gas-insulated switchgear (gas-insulated switchgear, GIS) equipment, partial discharge is one of the most critical electrical characteristics in the early development process of insulation defects, and electromagnetic, acoustic, chemical and optical signals generated by the partial discharge are often detected before the equipment fails seriously, so that accurate identification and quantitative evaluation of the partial discharge are of great significance in guaranteeing the operation safety of a power grid. Among the existing various detection means, an ultra-high frequency (UHF) detection method has become the partial discharge monitoring technology with the most engineering practicability in running GIS because of its strong anti-interference capability, suitability for on-line monitoring, capability of realizing high time resolution and higher positioning accuracy, and the like. However, UHF detection presents a number of difficulties in achieving quantitative diagnostics. The GIS is internally composed of metal shells, central conductors, basin-type or cylinder-type insulators, partition plates, expansion joints, flange joints, branch structures and other complex components, and the geometric structures and material parameters of the GIS are greatly different due to different voltage grades, manufacturing plant types and installation modes. The structural characteristics obviously change the electromagnetic propagation environment in the cavity, so that the electromagnetic wave excited by partial discharge presents obvious multi-mode propagation behavior, and various modes including a TEM fundamental mode, TE11, TM01 and other high-order modes can simultaneously participate in propagation, and the energy coupling relation, attenuation characteristic and dispersion characteristic are deeply influenced by structural factors. Although the traditional electromagnetic simulation method (such as FDTD, FEM and the like) can be used for carrying out visualization and deep analysis on electromagnetic wave propagation phenomena in GIS, the simulation model is often based on an idealized geometric structure, a uniform material assumption and limited boundary conditions, and is difficult to completely cover various loss mechanisms and fine structure differences in actual GIS equipment. Therefore, the electric field response obtained by simulation is close to the real situation in trend, but is difficult to be directly used for absolute value calibration of the discharge quantity, and particularly uniform quantification cannot be realized across different structural types. On the other hand, there are also obvious limitations to relying entirely on experimental approaches to sensor calibration. Firstly, the establishment of a full-scale experiment platform suitable for GIS with different voltage levels and different structural types is high in cost, and a great deal of time is consumed for performing equipment-by-equipment calibration. Secondly, many GIS equipment is installed under on-site working condition, and its internal structure can't disassemble or change, is difficult to controllable standard defect experiment. Thirdly, the experimental calibration result is generally only applicable to a specific structure and cannot be transferred to other GIS types, so that an industry unified UHF quantitative calibration system cannot be formed for a long time. The quantification aspect of UHF partial discharge detection in the prior art still has the prominent bottleneck that a unified technical framework capable of fusing simulation and experiment is lacking, calibration results are difficult to share among different GIS types, and the sensor output lacks a standardized quantification index of physical traceability. Therefore, a method for quantitatively calibrating UHF, which is universally applied between different GIS structures and fuses simulation data and experimental data, is needed, so that the structure mobility advantage of a simulation model can be fully utilized, simulation defects can be made up through experimental calibration, a standardized mapping relation between the output amplitude of a UHF sensor and the actual discharge quantity is realized, and a solid foundation is provided for quantitative diagnosis, state evaluation and intelligent operation and maintenance of GIS equipment. In addition, in the field of partial discharge quantitative calibration and electromagnetic simulation analysis, a calibration transmission link is generally established around a sensor electric field time domain signal or an acquired actual measurem