Transformer fault diagnosis method and system using induced ordered weighted evidence reasoning

Patent metadata

Field	Value
Publication number	US-11921169-B2
Application number	US-202117496779-A
Country	US
Kind code	B2
Filing date	Oct 8, 2021
Priority date	Dec 18, 2020
Publication date	Mar 5, 2024
Grant date	Mar 5, 2024

Abstract

A transformer fault diagnosis method and system using induced ordered weighted evidence reasoning is provided. The method includes the following steps. A typical data sample of transformer sweep frequency response analysis is loaded and a diagnostic label is set as an identification framework. Test data of a device to be diagnosed is loaded. Basic probability assignment is calculated and a reliability decision matrix is constructed. An induced ordered weighted averaging operator and its induction vector are calculated according to a sample source of the data. An index weight vector is calculated. All evidence is fused by the induced ordered weighted evidence theory and reliability of comprehensive evaluation is calculated, so as to determine a diagnosis result. The disclosure realizes fault identification, fault type distinction and fault position of power equipment by interpreting detection waveforms.

First claim

What is claimed is:

1. A transformer fault diagnosis method using induced ordered weighted evidence reasoning, comprising:
loading a typical data sample of a transformer FRA, and setting a diagnostic label as an identification framework; wherein the typical data sample of the transformer FRA is detected by voltage detectors, which are connected to end terminals of a transformer;
loading test data of a device to be diagnosed;
calculating basic probability assignment of a detection data curve of the device to be diagnosed and all characteristic data curves in the identification framework, and constructing a reliability decision matrix;
calculating an induced ordered weighted averaging operator and its induction vector according to a sample source of the test data of the device to be diagnosed;
calculating an index weight vector; and
fusing all evidence by the induced ordered weighted evidence theory and calculating reliability of comprehensive evaluation, so as to determine a diagnosis result;
wherein calculating the basic probability assignment of the detection data curve of the device to be diagnosed and all the characteristic data curves in the identification framework comprises:
dividing a detection data curve X 0 of the device to be diagnosed and a characteristic data curve X 1 in the identification framework into N SEC segments in a same manner, with X 0 =[X 01 , X 02 , . . . , X 0k , . . . , X 0N SEC ], X 1 =[X l1 , X l2 , . . . , X lk , . . . , X IN SEC ], where X 0k represents a k-th segment after division of X 0 , X lk represents a k-th segment after division of an l-th curve X 1 in the identification framework, where k=1, 2, . . . N SEC ; wherein the identification framework comprises a plurality of fault types and uncertainty result, and the fault types comprises winding short circuit, longitudinal deformation, and axial deformation; and
calculating a distance P k, l between each segment of the detection data curve of the device to be diagnosed and all the characteristic data curves in the identification framework through a curve similarity algorithm, and convert the distance of each curve into a reliability β k,l .

2. The method according to claim 1 , wherein the typical data sample of the transformer FRA loaded comprises:
an actual test sample, comprising historical test data of the device or manufacturer test data, and the data sample type is a most accurate reference data;
a sample of a device of a same model, comprising test data of another device of the same model, and the data is a more accurate reference data;
an accurate simulation sample, comprising a sample of various labels that are obtained after a detailed simulation model is established for the device; and
a fast simulation sample, comprising the sample of various labels that are obtained through simulation after a simplified model is established for the device while taking into consideration various limitations of hardware and timeliness.

3. A computer-readable storage medium with a computer program stored thereon, wherein the computer program realizes the steps of claim 2 when the computer program is executed by a processor.

4. The method according to claim 1 , wherein setting the diagnostic label as the identification framework comprises:
dividing data to be tested first into two types of diagnosis of healthy and unhealthy in a first stage of diagnosis, so that the identification framework is Θ={healthy, unhealthy}, or dividing the data to be tested into different health levels,
wherein when the diagnosis result is that the data to be tested is in an unhealthy state and a fault type has to be further detected, at this time, a second stage of diagnosis is entered, where the fault type is F i (i=1, 2, . . . ), and the identification framework is set to Θ={F 1 , F 2 , . . . , F i , . . . },
wherein when the diagnosis result is that the data to be tested is in the unhealthy state and a fault location has to be further detected, at this time, a third stage of diagnosis is entered, where the fault location is L j (j=1, 2, . . . ), and the identification framework is set to Θ={L 1 , L 2 , . . . , L j , . . . },
wherein to ensure a unified description, all elements in the identification framework are marked as Θ={H 1 , H 2 , . . . , H l , . . . , H N }, where l represents number of a subset in the identification framework, and N represents number of all subsets in the identification framework.

5. A computer-readable storage medium with a computer program stored thereon, wherein the computer program realizes the steps of claim 4 when the computer program is executed by a processor.

6. The method according to claim 1 , wherein constructing the reliability decision matrix comprises:
defining a distribution assessment vector S as S(X k (A t ))={(H l , β k,l (A t )), l=1, . . . , N}, which means that an attribute X k of an evaluated object A t is assessed to be at a level H l , with the reliability of filo, then an evaluation result of N SEC basic attributes of T evaluated objects (t=1, 2, . . . , T) may be expressed as the reliability decision matrix: D g =(s(x k (A t ))) N SEC ×T , where X k represents a k-th segment curve corresponding to the evaluated object A t after division.

7. The method according to claim 6 , wherein calculating the induced ordered weighted averaging operator and its induction vector comprises:
determining an induced variable u k of the device to be diagnosed, and assigning values to the induced variable based on data preliminarily calculated in previous steps;
defining an ordered weighted averaging operator F to satisfy

F
⁡

(

〈

u
1

,

a
1

〉

,
…
⁢

,

⁢

〈

u

N

S
⁢
E
⁢
C

,

a

N

S
⁢
E
⁢
C

〉

)

=

∑

Assignees

• Univ Wuhan

Inventors

• He Yigang
• Duan Jiajun
• Wu Xiaoxin
• He Liulu

Classifications

• G01R31/62 Primary

  Testing of transformers · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N5/048

  Fuzzy inferencing · CPC title