CN-121980781-A - VFTO distribution characteristic analysis method and system of extra-high voltage GIS based on multi-confidence interval evaluation

CN121980781ACN 121980781 ACN121980781 ACN 121980781ACN-121980781-A

Abstract

A method and a system for analyzing the VFTO distribution characteristics of an extra-high voltage GIS based on multi-confidence interval assessment are provided, wherein breakdown voltage data caused by GIS isolating switch operation is firstly collected, a hierarchical Bayesian model is combined with MCMC sampling of PyMC to obtain posterior distribution of BDV, then upper and lower boundaries of corresponding confidence intervals are extracted under a plurality of confidence levels, upper and lower boundary curves are fitted through parameterized functions to form upper and lower boundary mathematical expressions describing the change of BDV along with time, and the expressions are directly embedded into an EMTP/ATPdraw simulation model to serve as breakdown voltage boundary conditions, so that quick simulation of the VFTO characteristics of different dispersibility conditions is realized on the premise of no random sequence generation.

Inventors

LI ZHENHUA
HE JINGKE
ZHANG CHENGCHENG
LI ZHENXING
ZHANG LEI
ZHAO XIAOZHEN
PI ZHIYONG
XU YANCHUN
ZHANG WENTING

Assignees

三峡大学

Dates

Publication Date: 20260505
Application Date: 20260113

Claims (10)

1. The VFTO distribution characteristic analysis method of the extra-high voltage GIS based on the multi-confidence interval evaluation is characterized by comprising the following steps of: S1, collecting breakdown voltage data generated by GIS isolating switch operation, and sorting the breakdown voltage data according to sequence; S2, setting an observation layer, a parameter layer and a super parameter layer based on a hierarchical Bayesian model, setting prior distribution for parameters of each layer, reconstructing breakdown voltage posterior distribution, and carrying out posterior sampling by using a Markov chain Monte Carlo method in a PyMC framework to obtain posterior estimation values of breakdown voltage distribution parameters and super parameters; s3, calculating upper and lower bounds of confidence intervals under a plurality of confidence levels for the breakdown voltage posterior distribution estimated value obtained in the S2; S4, regarding different confidence intervals as different dispersibility grades, extracting upper and lower boundary discrete points of each confidence interval of breakdown voltage along with time change, and fitting the extracted upper and lower boundary discrete points by using a parameterized function to obtain distributed upper and lower boundary function expressions of the breakdown voltage along with time evolution under different confidence levels; S5, taking the upper and lower boundary function expressions of the breakdown voltage distribution obtained in the S4 as breakdown voltage boundary conditions, importing the breakdown voltage boundary conditions into an EMTP/ATPdraw simulation model, and carrying out VFTO transient simulation on the premise that the power supply phases are kept consistent; and S6, carrying out statistical analysis on the VFTO simulation result to obtain the VFTO amplitude characteristics and the probability distribution characteristics thereof under different dispersion conditions.
2. The method for analyzing the VFTO distribution characteristics of the extra-high voltage GIS based on the multi-confidence interval evaluation of claim 1, wherein in S2, the hierarchical Bayesian model is expressed by the following hierarchical structure: Setting a layer 1, namely an observation layer; The observation layer describes the distribution characteristics of breakdown voltage data through a probability distribution function: (1); in the formula (1): As a model of the probability distribution, The distribution parameters of the probability distribution model, i is the sequence number, Represents the ith breakdown voltage value data, Sequence numbers corresponding to the ith data; Setting a layer 2, namely a parameter layer; setting conditional prior distribution for breakdown voltage distribution parameters, and introducing distribution parameters of different confidence intervals into dispersive superparameter control: (2); in the formula (2), k is a confidence level serial number, As the breakdown voltage distribution parameter corresponding to the kth level, To control the hyper-parameters of the breakdown voltage dispersion degree corresponding to the kth level, A conditional prior distribution function of a parameter layer; Setting a 3 rd layer, namely a super parameter layer, for controlling the dispersibility of the parameter layer: Setting a normal prior distribution for the super-parameters: (3); In the formula (3): a normal distribution is indicated and the distribution is determined, Represents the mean value of the normal prior distribution of the hyper-parameters, Representing the variance of the normal prior distribution of the super-parameters; Constructing breakdown voltage posterior distribution, namely multiplying a likelihood function of observed data with prior distribution of parameters of each layer by using Bayesian theorem, so as to construct a combined posterior distribution model comprising distribution parameters and super parameters: (4); In the formula (4), the amino acid sequence of the compound, For a given observation data Lower parameters And super parameters Is a joint posterior distribution of (2); The representation is proportional; is a likelihood function; The conditional prior probability of the parameter layer; is the prior probability of the super parameter.
3. The method for analyzing the VFTO distribution characteristics of the extra-high voltage GIS based on the multi-confidence interval assessment according to claim 1, wherein in S2, the MCMC algorithm in PyMC, namely the Markov chain Monte Carlo method, is adopted to perform posterior sampling, the MCMC algorithm generates a large number of random sample sequences conforming to the posterior distribution characteristics by performing random walk and sampling in a joint probability space of parameters, and when the sampling times are enough and the Markov chain reaches a convergence state, the statistical characteristics of the sample sequences are the true posterior distribution of the parameters.
4. The method for analyzing the VFTO distribution characteristics of the extra-high voltage GIS based on the multi-confidence interval assessment, which is characterized in that in S3, a confidence interval is calculated by adopting a highest posterior density interval algorithm aiming at posterior distribution, and a definition formula is as follows: (5); And satisfies the following: (6); In the formula (5) and the formula (6): a set of confidence intervals representing a kth level, x representing sample variables in the posterior distribution, As a function of the posterior probability density, Representing a probability density threshold meeting a confidence level requirement, gamma k representing a preset confidence level; Through the above calculation, the time-lapse is obtained Upper and lower bound sequences of confidence interval of variation: (7); In the formula (7): And Respectively at the time And the lower threshold and the upper threshold of the breakdown voltage when the confidence interval is k.
5. The method for analyzing the VFTO distribution characteristics of the extra-high voltage GIS based on the multi-confidence interval assessment according to claim 1, wherein in S4, fitting is performed on the upper boundary and the lower boundary of the confidence interval by a parameterized function and a quadratic polynomial function: Upper bound function: (8); lower bound function: (9); In the formulas (8) and (9), t is a time variable; 、、 fitting coefficients of a quadratic term, a primary term and a constant term of the upper bound function of the kth confidence level respectively; 、、 Fitting coefficients of a quadratic term, a primary term and a constant term of the lower bound function of the kth confidence level respectively; the variation of the coefficient with the confidence level k represents a difference in breakdown voltage dispersion.
6. The method for analyzing the VFTO distribution characteristics of the extra-high voltage GIS based on the multi-confidence interval assessment according to claim 5, wherein in S5, the upper and lower boundaries EMTP/ATPdraw of the confidence interval fitted by the parameterized function are input into a model, and the boundary conditions for limiting the breakdown voltage are input as follows: (10); in the formula (10): represents the value of the breakdown voltage occurring at the simulation time t, Representing a numerical closed interval; In the simulation process, the breakdown voltage value is strictly limited to the lower bound function obtained by fitting And upper bound function Within the envelope formed; the phase of the power supply is kept consistent, the randomness is completely reflected by a boundary function, and a random sample sequence is not required to be generated; The output breakdown voltage expression is: (11); in the formula (11): Represents the instantaneous value of the breakdown voltage actually participating in the simulation, and RAN (1) represents a random number function uniformly distributed in the interval of 0-1.
7. The VFTO distribution characteristic analysis method of the extra-high voltage GIS based on multi-confidence interval assessment, which is characterized in that after introducing a random number function RAN (1) into a breakdown voltage expression, a delay breakdown strategy is adopted to further reflect the randomness of air gap ionization breakdown, and the breakdown moment corresponding to a time step is subjected to randomization treatment, namely after meeting a breakdown criterion, the simulation step does not immediately enter a reburning state, but a plurality of steps are delayed according to random numbers on the basis of the current step to trigger a re-breakdown event, so that the occurrence moment of re-breakdown has probability distribution characteristics; Is provided with For the simulation step length, the moment of the heavy-duty breakdown trigger is expressed as: (12); In the formula (12): As the moment of breakdown occurrence; the moment when the breakdown criterion is satisfied for the first time; Is a step offset determined from the random number.
8. The method for analyzing the VFTO distribution characteristics of the extra-high voltage GIS based on the multi-confidence interval evaluation of claim 1, wherein in S5, for each dispersion condition, repeated simulation is performed for a plurality of times under the same phase condition so as to obtain a VFTO peak sample set with statistical significance.
9. The method for analyzing the VFTO distribution characteristics of the extra-high voltage GIS based on the multi-confidence interval evaluation of claim 1, wherein in S6, each group of samples is respectively subjected to statistical analysis, and the method comprises the steps of calculating the overall distribution characteristics of peak amplitude values and the probability duty ratio of the overall distribution characteristics exceeding a specific threshold value; by comparing the different dispersion conditions, namely the change of the VFTO amplitude under different confidence interval widths, the influence rule of the breakdown voltage dispersion on the VFTO distribution characteristics is revealed.
10. The VFTO distribution characteristic analysis system of the extra-high voltage GIS based on multi-confidence interval assessment is characterized by comprising a data acquisition unit, a layered Bayesian modeling unit, a confidence interval calculation unit, an interval function fitting unit, a simulation input generation unit, a VFTO simulation unit and a result statistics unit; Each unit is sequentially connected with the simulation flow according to the data analysis to complete the whole flow from the breakdown voltage data analysis to the VFTO characteristic statistical output.

Description

VFTO distribution characteristic analysis method and system of extra-high voltage GIS based on multi-confidence interval evaluation Technical Field The invention belongs to the technical field of GIS electromagnetic transient simulation, and particularly relates to a VFTO distribution characteristic analysis method and system of an extra-high voltage GIS based on multi-confidence interval evaluation. Background In the process of switching the isolating switch on the no-load short bus and other operation processes, the gas insulated switchgear GIS can generate an electric arc and trigger extremely fast transient overvoltage VFTO, and the VFTO has the characteristics of high amplitude, high frequency, high rising speed and the like, can cause insulation threat to the GIS body and connected equipment thereof, and can also cause interference to the electromagnetic environment of secondary equipment to influence the safe and stable operation of a power system. The breakdown voltage BDV between the isolating switch contacts is an important factor influencing the generation of electric arcs, and the randomness and the dispersity of the BDV directly influence the simulation accuracy and the statistical characteristics of the VFTO; at present, a random sample is usually generated based on a fixed breakdown voltage curve or given experience distribution, and random calculation is carried out for a plurality of times in a simulation tool so as to evaluate the fluctuation interval of the VFTO; this type of approach has the following major problems: Firstly, the article (Liu Weidong, wang Linsen, chen Weijiang, etc. ultra-high voltage GIS ultra-fast transient overvoltage test repeated breakdown process research [ J ]. High voltage technology, 2011,37 (03): 644-650.) has the existing breakdown voltage modeling mostly adopting linear or empirical fitting relation, often presuming that the breakdown occurrence process has certainty or approximate certainty, neglecting statistical dispersion and probability interval of the breakdown voltage under different operation conditions, resulting in limitation of simulation results to single waveform or lack of sufficient statistical significance; Secondly, although the introduction method is discussed to reflect the randomness of the breakdown voltage, the method relies on external sampling, the randomness is uncontrollable, and a large number of simulation times are required to obtain effective statistical results, and the operation cost is high (Li Zhibing. The GIS isolation switch operation repeated discharge model research [ D ]. Chinese electric science institute, 2014.). Meanwhile, the method is difficult to directly embody the VFTO fluctuation range under different confidence levels, and the quantitative expression of risk controllability in actual engineering is insufficient; In summary, the existing VFTO simulation method generally lacks modeling means for breakdown dispersion and multi-confidence interval evaluation, and cannot reveal the fluctuation range and uncertainty measurement of the VFTO under the condition of the same power phase angle; therefore, a method for modeling the probability interval of the VFTO with controllable randomness and explicit physical meaning is needed to accurately describe the BDV distribution interval difference, so as to improve the simulation efficiency and the engineering application reliability. Disclosure of Invention The invention aims to provide a VFTO distribution characteristic analysis method and a system for an extra-high voltage GIS based on multi-confidence interval evaluation, according to the method, parameterized function fitting is performed by using upper and lower bounds of the confidence interval, and quick analysis of VFTO distribution characteristics under different dispersibility conditions can be realized on the premise that random sequences are not generated. In order to solve the technical problems, the invention adopts the following technical scheme: The VFTO distribution characteristic analysis method of the extra-high voltage GIS based on the multi-confidence interval evaluation comprises the following steps: s1, collecting breakdown voltage data generated by GIS isolating switch operation, and sorting according to time or event sequence; S2, setting an observation layer, a parameter layer and a super parameter layer based on a hierarchical Bayesian model, setting prior distribution for parameters of each layer, reconstructing breakdown voltage posterior distribution, and carrying out posterior sampling by using a Markov chain Monte Carlo method in a PyMC framework to obtain posterior estimation values of breakdown voltage distribution parameters and super parameters; s3, calculating upper and lower bounds of confidence intervals under a plurality of confidence levels for the breakdown voltage posterior distribution estimated value obtained in the S2; S4, regarding different confidence intervals as different di