CN107271867A - GIS partial discharge fault type recognition method based on D S evidence theories - Google Patents

Abstract

The invention discloses a GIS partial discharge fault type identification method, aiming at the three detection methods of ultrasonic detection method, ultra-high frequency detection method and SF6 decomposition product detection method in the prior art, it cannot completely identify the partial discharge fault type of GIS equipment To solve the technical problem of effective identification, it is proposed to use the D‑S (Dempster‑Shafer) evidence theory to carry out decision-level fusion of the fault type identification results of the three detection methods in the prior art to make up for the shortcomings of the three single methods; a design based on The multi-sensor information fusion fault type identification system structure of D‑S theory, through the simulation calculation of two typical insulation faults, the accuracy and rapidity of GIS insulation fault type identification can be greatly improved. Research on the insulation state detection of GIS equipment It has certain reference value.

Description

GIS Partial Discharge Fault Type Identification Method Based on D-S Evidence Theory

technical field

The invention relates to a method for identifying the insulation state of GIS equipment, in particular to a decision-level data fusion identification method based on D-S (Dempster-Shafer) theory.

Background technique

On-line Partial Discharge (PD) detection of Gas Insulated Switchgear (GIS) can effectively grasp the internal insulation status of GIS and prevent power grid accidents caused by GIS insulation fault tripping. At present, the engineering application of ultrasonic and ultra-high frequency detection methods is relatively common, and the technology is relatively mature, but many problems have been reported in the engineering application, and the SF6 decomposition component detection method is currently a relatively new online detection method with a good development trend.

The ultrasonic detection method detects the PD signal by detecting the ultrasonic and vibration signals generated by the PD with an ultrasonic probe, and the Ultra High Frequency (UHF) method detects the PD signal by receiving the 300-3000MHz frequency band UHF electromagnetic wave signal generated by the PD through an antenna. Partial discharges caused by different insulation defects produce different decomposed compound gases. The SF6 decomposition product detection method detects the PD signal through the content of various characteristic gases generated by the decomposition of SF6 gas inside the GIS caused by PD.

Four types of insulation defects are simulated inside GIS, including protrusion type A defect, attachment type B defect, insulator air gap type C defect and free particle D type defect. Using these three methods for fault detection, the analysis of the detection map shows that: ultrasonic The detection method has the most obvious detection effect on PD caused by class D free metal particle defects, but it is not obvious for the discharge detection of class B insulators attached to pollutant defects; The detection effect of PD is the most obvious, and the detection effect of D-type free metal particle defect discharge is the worst; the SF6 decomposition product component detection method is generally 15 hours after the PD occurs, and the SF6 gas decomposition product content reaches a certain amount before it can be effectively identified. , among them, the PD produced by type A metal protrusions and type B insulators attached to pollutant defects is the most stable, with large gas production and high decomposition rate, and the best identification effect. The PD gas production of type C insulator air gap defects is relatively small, and the identification effect poor.

Therefore, the three single methods of ultrasonic, ultra-high frequency, and SF6 decomposition component detection cannot fully and effectively identify the PD fault type of GIS equipment.

Contents of the invention

The present invention uses the D-S (Dempster-Shafer) theory to carry out decision-level fusion on the data analysis results collected by three different types of sensors, intelligently synthesizes multi-source information from a certain target, and utilizes the complementary characteristics of the three detection methods, Ultrasound, ultra-high frequency and SF6 decomposition component detection are used for joint online detection, and the D-S evidence theory is used to make data fusion decisions from three different types of sensors.

In order to achieve the above object, the present invention adopts the following technical solutions:

A GIS partial discharge fault type identification method is characterized in that three detection methods, namely, ultrasonic detection method, ultra-high frequency detection method and SF6 decomposition product detection method, are combined for data, so as to identify the GIS partial discharge fault type.

The data fusion is based on D-S evidence theory.

The identification method further includes the steps of:

Step 1: The partial discharge results are detected by ultrasonic sensors, ultra-high frequency sensors, and gas sensors;

Step 2: Preprocessing the data detected by each sensor in the step 1, and determining the fault type respectively;

Step 3: Use D-S evidence theory for decision fusion;

Step 4: Output the recognition result.

Described step three further comprises the following steps:

(1) Carry out reliability data fusion of ultrasonic and ultra-high frequency detection methods;

(2) The fused result is fused again with the reliability result of the decomposition product detection method to obtain the fused result.

Described step (1) carries out based on following formula:

Among them, m(A) is the basic credibility distribution function of A; A is the possible result of the proposition; Θ represents the non-empty set of all possible conclusions corresponding to the proposition, and the set contains a limited number of elements, and all elements are mutually exclusive; let Θ There are two reliability functions of Bel ₁ and Bel ₂ on , and m ₁ and m ₂ are the basic reliability distributions corresponding to Bel ₁ and Bel ₂ respectively. If A∈Θ and m(A)>0, the focal elements are respectively For A ₁ , A ₂ ,...,A _k and B ₁ ,B ₂ ...,B _n ; i represents the constant 1, 2,...k; j represents the constant 1, 2,...n; n represents the natural number 1, 2, 3,....

Described step (2) carries out based on following formula:

The beneficial effects of the present invention are: the present invention designs a multi-sensor information fusion fault type identification system structure for the lack of identification of a single type detection method, which greatly improves the accuracy and rapidity of GIS insulation fault type identification.

Description of drawings

Figure 1 is a process diagram of information fusion of D-S evidence theory;

Figure 2 is a structural block diagram of the GIS partial discharge type identification information fusion system.

detailed description

D-S (Dempster-Shafer) evidence theory summarizes and calculates uncertain and incomplete information through mathematical reasoning, and makes scientific and reasonable decisions. The theory proposes the concepts of Basic Probability Assignment (BPA), Belief Function (BEL) and Likelihood Function (P1).

(1) Recognition framework Θ and basic probability assignment BPA

Let Θ represent a non-empty set of propositions corresponding to all possible conclusions, the set contains a finite number of elements, and all elements are mutually exclusive.

Θ={A ₁ ,A ₂ ,...,A _n ,θ}

In the formula, A _i is the possible result of the proposition; θ represents the uncertainty of the result.

There is a mass function m:2 ^Θ →[0,1] in Θ, and it satisfies:

Then m(A) is A's basic probability assignment BPA.

(2) Belief function Bel and likelihood function Pl

The D-S evidence theory gives a confidence interval [Bel(A), P1(A)] to represent the upper and lower limits of the range supported by the fusion result to the event, then the lower limit function bel is defined as:

The online function pl is defined as:

Among them: Bel(A)≤P1(A), Θ represents all subsets, and Bel(A) or P1(A) can be used to represent the degree of trust and support for the proposition during fusion calculation.

Combination rule is the process of fusion, evidence fusion algorithm is two kinds of combination of reliability functions and combination of multiple reliability functions.

(1) Synthesis algorithm of two reliability functions

Assuming that there are two reliability functions Bel ₁ and Bel ₂ on Θ, m ₁ and m ₂ are the basic reliability distributions corresponding to Bel ₁ and Bel ₂ respectively. If A∈Θ and m(A)>0, its focal The elements are respectively A ₁ , A ₂ ,…,A _k and B ₁ ,B ₂ …,B _n , suppose:

Then the basic credibility distribution function after synthesis is:

In formula 2, if k≠1, then m can determine a basic probability assignment; otherwise m ₁ and m ₂ are contradictory, and the assignment of basic probability cannot be combined. i represents constants 1, 2, ...k; j represents constant 1, 2,...n; n represents the natural number 1, 2, 3,....

(2) Synthesis of multiple reliability functions

Assuming that there are reliability functions Bel1, Bel2,..., Beln on the same Θ, the corresponding basic reliability distributions are m1, m2,...,mn, if there are And the basic reliability distribution is m, then:

The above formula is satisfied

The combined basic credibility distribution function m is:

The information fusion process of D-S evidence theory is shown in Figure 1. Various methods such as decision-making based on basic probability assignment can be used. The output rules of decision-making based on basic probability assignment should satisfy:

Also meet:

Then it shows that the decision output result is A ₁ . m(A ₁ ) is the maximum value of the output BPA; m(A ₂ ) is the second maximum value of the output BPA; ε is the preset threshold value, which is 0.25 in the present invention; m(θ) is the uncertainty BPA.

Figure 2 shows the GIS partial discharge fault type identification multi-information fusion system based on D-S evidence theory. The system consists of four parts: 1) A collection of multiple types of sensors, consisting of ultrasonic, ultra-high frequency, and gas sensors; 2) The homogeneous sensor collection information fusion part, which fuses homogeneous data after data preprocessing; 3 ) Judging the type of partial discharge fault, using three single methods to detect the type of fault, using the D-S evidence theory for decision-level fusion; 4) Decision output, outputting the identification result of PD type.

Construct the fault identification framework Θ={F ₁ ,F ₂ ,F ₃ ,F ₄ ,θ}, where F1 is the defect of free conductive particles; F2 is the defect of surface attachment; F3 is the defect of metal protrusion; F4 is the air gap of the insulator defect; θ stands for uncertainty.

Table 1 shows the results of the simulation test for defect F3. The BPA of three kinds of information, including ultrasonic S, UHF P, and decomposition product component Q, is calculated by formula (2) and formula (4), and the fusion result is obtained through the D-S synthesis rule. Firstly, the reliability data of the S and P detection methods are fused by the formula (2), and then the fused result is fused with the reliability result of the decomposition product component Q by the formula (4), and the final result is as follows D-S fusion (S&P&Q) is shown in the table.

First calculate the conflict coefficient k of the evidence system:

Then the credibility distribution function after synthesis:

It can be seen from the above table that after D-S fusion (S&P&Q), m(θ)=0.0376, m(F3)=0.9006, where m(θ) is significantly reduced, that is, the uncertainty of the diagnosis result is reduced, and the possibility of fault F3 The reliability is greatly improved, and the reliability of the corresponding fault diagnosis is also greatly improved. The output conclusions of the three types of identification information are basically the same, that is, they all believe that the probability of failure of the metal protrusion is relatively high.

m(A ₁ )＝0.9006＞m(θ), m(A ₂ )＝0.0337,

m(A ₁ )-m(A ₂ )=0.9006-0.0337=0.8669>ε,

The preset threshold value ε of the present invention is set to 0.25, and the fusion result satisfies formula (5) and formula (6), which conforms to the basic probability assignment decision-making output judgment rule and is determined to be a surface attachment defect, which is consistent with the initially set fault type.

Table 2 shows the simulation test results for surface attachment defects. The identification results of the three single detection methods are not completely consistent. The UHF P detection method has a higher probability of identifying F3 defects and F2 defects. Component Q detection methods all believe that the probability of F2 insulator surface attachment is relatively high, but ultrasonic S can also identify the existence of F1 defects at the same time. Through formula (2) and formula (4) in the D-S evidence theory to calculate the detection results of S, P and Q, the decision output result of D-S fusion (S&P&Q) is obtained. The deterministic m(θ) is reduced to 0.0040, and the result satisfies formula (5) and formula (6), conforms to the decision rule of probability assignment decision output, and is consistent with the actual model setting failure.

Table 1 Three single methods and D-S fusion test results reliability distribution table 1

Table 2 Three single methods and D-S fusion test results reliability distribution Table 2

Claims (6)

1. A GIS partial discharge fault type identification method is characterized in that these three detection methods of ultrasonic detection method, ultra-high frequency detection method and SF decomposed product detection method are carried out data fusion, thereby identifying the GIS partial discharge fault type. 2. The GIS partial discharge fault type identification method according to claim 1, wherein said data fusion is based on D-S evidence theory. 3. GIS partial discharge fault type identification method according to claim 2, is characterized in that, described identification method further comprises the steps: Step 1: The partial discharge results are detected by ultrasonic sensors, ultra-high frequency sensors, and gas sensors; Step 2: Preprocessing the data detected by each sensor in the step 1, and determining the fault type respectively; Step 3: Use D-S evidence theory for decision fusion; Step 4: Output the recognition result. 4. GIS partial discharge fault type identification method according to claim 3, described step 3 further comprises the steps: (1) Carry out reliability data fusion of ultrasonic and ultra-high frequency detection methods; (2) The fused result is fused again with the reliability result of the decomposition product detection method to obtain the fused result. 5. GIS partial discharge fault type recognition method according to claim 4, described step (1) carries out based on the following formula: Among them, m(A) is the basic credibility distribution function of A; A is the possible result of the proposition; Θ represents the non-empty set of all possible conclusions corresponding to the proposition, and the set contains a limited number of elements, and all elements are mutually exclusive; let Θ There are two reliability functions of Bel ₁ and Bel ₂ on , and m ₁ and m ₂ are the basic reliability distributions corresponding to Bel ₁ and Bel ₂ respectively. If A∈Θ and m(A)>0, the focal elements are respectively For A ₁ , A ₂ ,...,A _k and B ₁ ,B ₂ ...,B _n ; i represents the constant 1, 2,...k; j represents the constant 1, 2,...n; n represents the natural number 1, 2, 3,.... 6. GIS partial discharge fault type recognition method according to claim 4, described step (2) is carried out based on the following formula: m(A) is the basic credibility distribution function of A; A is the possible result of the proposition; Θ represents the non-empty set of all possible conclusions corresponding to the proposition, the set contains a limited number of elements, and all elements are mutually exclusive; There are two reliability functions of Bel ₁ and Bel ₂ , m ₁ and m ₂ are the basic reliability distributions corresponding to Bel ₁ and Bel ₂ respectively, if A∈Θ and m(A)>0, the focal elements are A ₁ ,A ₂ ,...,A _k and B ₁ ,B ₂ ...,B _n ;