CN106526352A - Method and system for determining power transformer fault types - Google Patents
Method and system for determining power transformer fault types Download PDFInfo
- Publication number
- CN106526352A CN106526352A CN201610875616.2A CN201610875616A CN106526352A CN 106526352 A CN106526352 A CN 106526352A CN 201610875616 A CN201610875616 A CN 201610875616A CN 106526352 A CN106526352 A CN 106526352A
- Authority
- CN
- China
- Prior art keywords
- power transformer
- fault
- fault type
- data
- type
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Housings And Mounting Of Transformers (AREA)
Abstract
The invention discloses a method and a system for determining power transformer fault types. The method comprises steps of collecting and integrating power transformer fault data, and defining fault characteristic quantities and fault types of a power transformer; carrying out discretization pretreatment of the power transformer fault data; based on the pretreated power transformer fault data, using an Apriori mining algorithm to mine the association rules between the fault characteristic quantities and the fault types, and generating a final fault type association rule set; and determining the power transformer fault types according to the final fault type association rule set.
Description
Technical field
The present invention relates to Power Transformer Faults Intelligent Diagnosis Technology, more particularly to a kind of to determine Power Transformer Faults class
The method and system of type.
Background technology
Electric energy plays more and more important effect in modern society, and power network is used as electrical energy transportation and the important base of distribution
Infrastructure, proposes higher and higher requirement to its security, reliability and economy.Power transformer becomes as in power network
One of the capital equipment in electric institute and power plant, not only can bring the voltage up so that electric energy is to be sent to electricity consumption region compared with low-loss,
Voltage can also be reduced to use voltages at different levels, meet the electricity consumption needs of different user.When power transformer works, its insulation
System is easily acted on by each side such as heat, electricity, machinery, and its each side function will tend to deterioration step by step, ultimately cause
The generation of Power Transformer Faults.The failure of power transformer can have a strong impact on the normal operation of power network, be even more when serious
The supply of meeting interrupt power, the life that can not only give people bring impact, can also cause serious economic loss to society.Institute
To monitor the running status of transformer and predict the failure of power transformer to reducing the generation of failure and reducing economical in time
Loss plays very obvious action.
In normal course of operation, equipment long-term fever can cause transformer oil gradually aging, rotten to transformer, this mistake
Journey is a very slow process, although it is accompanied by a small amount of low molecular hydrocarbon, CO, CO2Generation, but which can't be exceeded
The fault alarm value of regulation.And when electric superheating occurs in inside transformer, C-C keys and c h bond can occur fracture, formed not
The free radical of stable hydrogen and its hydrocarbon, and the hydrogen and some hydrocarbon gas that were formed later are then by these hydrogen
What atom and free radical were bound up again, this kind of hydrocarbon gas mainly include methane CH4, acetylene C2H2, ethene C2H4, ethane
C2H6, hydrogen H2Deng.Due to these Gases Dissolved in Transformer Oil species and content and equipment fault between exist to a certain degree
Relevance, therefore based on dissolved gas analysis method (DGA) analyze Gases Dissolved in Transformer Oil type and its content
Can effectively diagnose and assess inside transformer state of insulation.At present, it is analyzed for Gases Dissolved in Transformer Oil
There is certain ambiguity due to the classification of fault type in three-ratio method, not simple between failure volume and fault type
One-to-one relationship, so which is perfect not enough in terms of the malfunction of assessment equipment.Therefore, build how effectively and comprehensively
The correlation rule between Gases Dissolved in Transformer Oil characteristic quantity and fault type and its order of severity is erected for based on transformation
The transformer state diagnosis of device dissolved gas analysis is significant.
The content of the invention
In order to solve the above problems, the invention discloses a kind of method for determining Power Transformer Faults type, the side
Method includes：
Power Transformer Faults data are collected and integration, and is divided fault characteristic value and the failure of power transformer
Type, wherein：
Content and variable quantity of the fault characteristic value for six kinds of dissolved gas in electric power transformer oil, six kinds of solution gas
Body is respectively methane CH4, acetylene C2H2, ethene C2H4, ethane C2H6, hydrogen H2And total hydrocarbon；
The fault type is：Cryogenic overheating, overheated middle temperature, hyperthermia and superheating, low energy electric discharge and high-energy discharge；
Above-mentioned Power Transformer Faults data are pre-processed, the pretreatment is referred to Power Transformer Faults data
Carry out discretization；
Based on pretreated Power Transformer Faults data, power transformer event is excavated using Apriori mining algorithms
Correlation rule between barrier characteristic quantity and fault type, generates the Association Rules of final fault type, and the generation is final
The step of Association Rules of fault type, includes：
For each fault type in pretreated Power Transformer Faults data, according to the minimum support of setting
Degree, determines fault characteristic value of the support more than or equal to minimum support；
For each fault type in pretreated Power Transformer Faults data, according to the minimum trust of setting
Degree, determines reliability in fault characteristic value of the support more than or equal to minimum support more than or equal to minimum trust
The fault characteristic value of degree, so that it is determined that the sample data of final fault characteristic value and fault type；
Relation between the final fault characteristic value for determining and the sample data of fault type is opened up with network
Show, determine min-link number；
After determining min-link number, the sample data of the final fault characteristic value for determining and fault type is divided into into instruction
Practice part and part of detecting, training part is used for excavating the correlation rule between Power Transformer Faults characteristic quantity and fault type,
So as to generate the Association Rules of fault type, and part of detecting is used for the Association Rules of the fault type that detection is generated
The degree of accuracy, it is determined that the Association Rules of final fault type；And
The fault type of power transformer is determined according to the Association Rules of final fault type.
Preferably, it is described that collecting for Power Transformer Faults data is included with integration：The combing of unstructured data, system
The formulation of one coding rule, the typing of unstructured data, data cleansing process and transformer data characteristics statistical analysis.
Preferably, carrying out discretization to Power Transformer Faults data includes：
The discretization of fault characteristic value, the discretization of the fault characteristic value adopt the discretization method of equidistant division,
Fault alarm value size with the content of every kind of gas and variable quantity represents gas content as siding-to-siding block length when centrifugal pump is 0
With variable quantity in normal range (NR), when centrifugal pump is the integer more than 0, represent that gas content and variable quantity exceed fault alarm
Value, and centrifugal pump is equal to the multiple more than fault alarm value；And
The discretization of fault type, the discretization of the fault type adopt boolean's discretization, when power transformer occurs
It is 1 during the failure of the type, is otherwise 0.
Preferably, the association between Power Transformer Faults characteristic quantity and fault type is excavated using Apriori mining algorithms
Rule, correlation rule and the excavation power transformer of content and fault type including excavation Detection Ssytem of Dissolved Gases in Power Transformer Oil Base
The correlation rule of the variable quantity and fault type of oil dissolved gas.
According to a further aspect in the invention, the invention further relates to a kind of system for determining Power Transformer Faults type, institute
The system of stating includes：
Data divide device, and which is used for collecting and integration Power Transformer Faults data, and divides power transformer
The fault characteristic value and fault type of device, wherein：
Content and variable quantity of the fault characteristic value for six kinds of dissolved gas in electric power transformer oil, six kinds of solution gas
Body is respectively methane CH4, acetylene C2H2, ethene C2H4, ethane C2H6, hydrogen H2And total hydrocarbon；
The fault type is：Cryogenic overheating, overheated middle temperature, hyperthermia and superheating, low energy electric discharge and high-energy discharge；Data are located in advance
Reason device, which is used for pre-processing above-mentioned Power Transformer Faults data, and the pretreatment is to power transformer data
Carry out discretization；
Fault type association rule mining device, which is used for based on pretreated Power Transformer Faults data, utilizes
Apriori mining algorithms excavate the correlation rule between Power Transformer Faults characteristic quantity and fault type, generate final failure
The Association Rules of type, the fault type association rule mining device generate the Association Rules of final fault type
Step includes：
For each fault type in pretreated Power Transformer Faults data, according to the minimum support of setting
Degree, determines fault characteristic value of the support more than or equal to minimum support；
For each fault type in pretreated Power Transformer Faults data, according to the minimum trust of setting
Degree, determines reliability in fault characteristic value of the support more than or equal to minimum support more than or equal to minimum trust
The fault characteristic value of degree, so that it is determined that the sample data of final fault characteristic value and fault type；
Relation between the final fault characteristic value for determining and the sample data of fault type is opened up with network
Show, determine min-link number；
After determining min-link number, the sample data of the final fault characteristic value for determining and fault type is divided into into instruction
Practice part and part of detecting, training part is used for excavating the correlation rule between Power Transformer Faults characteristic quantity and fault type,
So as to generate the Association Rules of fault type, and part of detecting is used for the Association Rules of the fault type that detection is generated
The degree of accuracy, it is determined that the Association Rules of final fault type；And
Power Transformer Faults determining device, which is used for determining that electric power becomes according to the Association Rules of final fault type
The fault type of depressor.
Preferably, the data divide device and collecting for Power Transformer Faults data are included with integration：Destructuring
The combing of data, the formulation of Unified coding rule, the typing of unstructured data, data cleansing process and transformer data are special
Levy statistical analysis.
Preferably, data prediction device Power Transformer Faults data are carried out pretreatment refer to power transformer therefore
The discretization of barrier data, including：
The discretization of fault characteristic value, which adopts the discretization method of equidistant division, with the content of every kind of gas and change
The fault alarm value size of change amount is siding-to-siding block length, when centrifugal pump is 0, represents gas content and variable quantity in normal range (NR)
It is interior, when centrifugal pump is the integer more than 0, represents that gas content and variable quantity exceed fault alarm value, and centrifugal pump is equal to and surpasses
Cross the multiple of fault alarm value；
The discretization of fault type, which adopts boolean's discretization, i.e. data to be mapped on Boolean, when power transformer goes out
It is 1 during the failure of existing the type, is otherwise 0.
Preferably, fault correlation rule digging device excavates Power Transformer Faults feature using Apriori mining algorithms
The correlation rule of amount and fault type, including the association rule of the content and fault type for excavating Detection Ssytem of Dissolved Gases in Power Transformer Oil Base
Then with excavate Detection Ssytem of Dissolved Gases in Power Transformer Oil Base variable quantity and fault type correlation rule.
By said method, the content and variable quantity and failure classes of Detection Ssytem of Dissolved Gases in Power Transformer Oil Base is preferably established
The Association Rules of type such that it is able to according to content and the variable quantity of Detection Ssytem of Dissolved Gases in Power Transformer Oil Base, i.e. fault characteristic value
To determine the failure of transformer, in time failure is positioned and analyzed, it is to avoid the generation of Power Transformer Faults, it is perfect
Transformer fault diagnosis system, so that improve the operational reliability of power transformer.
Description of the drawings
By reference to the following drawings, the illustrative embodiments of the present invention can be more fully understood by：
Fig. 1 is the flow chart of the method for the determination Power Transformer Faults type according to embodiment of the present invention；
Fig. 2 is the content of the excavation Detection Ssytem of Dissolved Gases in Power Transformer Oil Base according to embodiment of the present invention and fault type
The flow chart of correlation rule；
Fig. 3 A and 3B are the content of the Detection Ssytem of Dissolved Gases in Power Transformer Oil Base according to embodiment of the present invention and fault type
Network under different linking number；
Fig. 4 is the structure chart of the system of the determination Power Transformer Faults type according to embodiment of the present invention；
Fig. 5 is the flow chart of the method for the determination Power Transformer Faults type according to another embodiment of the invention；
Fig. 6 is the variable quantity of the excavation Detection Ssytem of Dissolved Gases in Power Transformer Oil Base according to another embodiment of the invention and failure
The flow chart of the correlation rule of type；And
Fig. 7 A and 7B are the variable quantity of the Detection Ssytem of Dissolved Gases in Power Transformer Oil Base according to another embodiment of the invention and event
Barrier network of the type under different linking number.
Specific embodiment
With reference now to accompanying drawing, the illustrative embodiments of the present invention are introduced, however, the present invention can be with many different shapes
Formula is implementing, and is not limited to embodiment described herein, there is provided these embodiments are to disclose at large and fully
The present invention, and the scope of the present invention is fully passed on to person of ordinary skill in the field.For showing for being illustrated in the accompanying drawings
Term in example property embodiment is not limitation of the invention.In the accompanying drawings, identical cells/elements are attached using identical
Icon is remembered.
Embodiment one
Fig. 1 is the flow chart of the method for the determination Power Transformer Faults type according to embodiment of the present invention.Such as Fig. 1 institutes
Show, the method 100 of the determination Power Transformer Faults type in the present invention is from the beginning of step S101.
In step S101, Power Transformer Faults data are collected and integration, and divided the failure of power transformer
Characteristic quantity and fault type, wherein, content of the fault characteristic value for six kinds of dissolved gas in electric power transformer oil is described
Six kinds of dissolved gas are respectively methane CH4, acetylene C2H2, ethene C2H4, ethane C2H6, hydrogen H2And total hydrocarbon, the fault type
For：Cryogenic overheating, overheated middle temperature, hyperthermia and superheating, low energy electric discharge and high-energy discharge.Preferably, it is described to Power Transformer Faults
Collecting for data is included with integration：The combing of unstructured data, the formulation of Unified coding rule, the record of unstructured data
Enter, data cleansing process and transformer data characteristics statistical analysis.
In step S102, above-mentioned Power Transformer Faults data are pre-processed, the pretreatment is to power transformer
Device fault data carries out discretization.
Preferably, Power Transformer Faults data are carried out with discretization includes discretization and the fault type of fault characteristic value
Discretization.
The discretization of fault characteristic value refers to the discretization of the content of six kinds of dissolved gas in electric power transformer oil, described
Discretization method of the discretization of fault characteristic value using equidistant division.Due to hydrogen in the normal transformer oil of power transformer
Different with the prohibitive content of hydrocarbon gas, concrete demand value is as shown in table 1, if adopting identical area to all properties value
Between, then lack certain specific aim, therefore after experiment, with the fault alarm value size of the content of every kind of gas as interval long
Degree, when centrifugal pump is 0, represents gas content in normal range (NR), when centrifugal pump is the integer more than 0, represents that gas contains
Amount is beyond fault alarm value, and centrifugal pump is equal to the multiple more than fault alarm value.
The prohibitive content of hydrogen and hydrocarbon gas in 1 normal transformer oil of table
The discretization of fault type, the discretization of the fault type adopt boolean's discretization, when power transformer occurs
It is 1 during the failure of the type, is otherwise 0.
In step S103, based on pretreated Power Transformer Faults data, electricity is excavated using Apriori mining algorithms
The correlation rule of the content of six kinds of dissolved gas and five kinds of fault types in power transformer oil, generates the pass of final fault type
Connection rule set.
In step S104, the fault type of power transformer is determined according to the Association Rules of final fault type.
Fig. 2 is the content of the excavation Detection Ssytem of Dissolved Gases in Power Transformer Oil Base according to embodiment of the present invention and fault type
The flow chart of correlation rule.As shown in Fig. 2 excavating the content of Detection Ssytem of Dissolved Gases in Power Transformer Oil Base and the association rule of fault type
Method 200 then is from the beginning of step S201.In step S201, for pretreated Power Transformer Faults data in it is each
Individual fault type, according to the minimum support of setting, determines fault characteristic value of the support more than or equal to minimum support.
In step S202, for each fault type in pretreated Power Transformer Faults data, foundation sets
Fixed minimum reliability, support more than or equal to minimum support fault characteristic value in determine reliability be more than or
Equal to the fault characteristic value of minimum reliability, so that it is determined that the sample data of final fault characteristic value and fault type.
In step S203, by the relation net between the final fault characteristic value for determining and the sample data of fault type
Network figure is shown, and determines min-link number.
Fig. 3 A and 3B are the content of six kinds of dissolved gas in electric power transformer oil according to embodiment of the present invention and five kinds
Network of the fault type under different linking number.As shown in figs.3 a and 3b, the size of min-link number is adjusted, then power transformer
Relational network figure in device oil between the content of six kinds of dissolved gas and five kinds of fault types can occur respective change therewith, minimum
Link number becomes big, and the quantity of link is reduced.Before the Association Rules of fault type are generated, according to support and reliability
Setting, selects optimal min-link number.
In step S204, after determining min-link number, by the final fault characteristic value for determining and the sample of fault type
Data are divided into training part and part of detecting, and training part is used for excavating between Power Transformer Faults characteristic quantity and fault type
Correlation rule, so as to generate the Association Rules of fault type, and part of detecting is used for detecting the fault type that generates
The degree of accuracy of Association Rules, it is determined that the Association Rules of final fault type.
(1) train part
Training part proportion is 70%, and distribution is random.Training part be used for excavating in electric power transformer oil six kinds it is molten
Solution gas content and five kinds of fault types correlation rule, the content with six kinds of dissolved gas as preceding paragraph, five kinds of fault types
For consequent, minimum support and minimum reliability be set, excavated in electric power transformer oil with the data mining algorithm of Apriori
The correlation rule of the content of six kinds of dissolved gas and five kinds of fault types, selected part are as shown in table 2.
2 association rule mining partial results of table
After excavating the content of six kinds of dissolved gas in electric power transformer oil and the correlation rule of five kinds of fault types, then root
The Association Rules of five kinds of fault types, the correlation rule of the fault type of generation are generated according to the consequent fault type of correlation rule
The size of collection is as shown in table 3.
The number of 3 five kinds of fault type rule set correlation rules of table
(2) part of detecting
Part of detecting proportion is 30%, and distribution is random.Part of detecting is used for five kinds of fault types that inspection is generated
The performance of Association Rules how.In part of detecting, the failure that the Association Rules of five kinds of fault types of Jing are predicted and reality
Border failure is contrasted, if the failure of prediction is identical with physical fault, can draw the reliability of the prediction.With hyperthermia and superheating
As a example by, intercepting partial table analysis as shown in table 4, in figure, three row represent physical fault respectively, predict the letter of failure and its prediction
Lai Du.
Predicting the outcome for 4 hyperthermia and superheating of table is compared with actual result
The accuracy of five kinds of transformer fault diagnosis can be drawn by statistical analysis, as shown in table 5.Set when accuracy meets
During definite value, it is determined that the Association Rules of final fault type.
The statistics of 5 association rule mining method Fault Diagnosis Method of Power Transformer of table
According to a further aspect in the invention, the invention further relates to a kind of system for determining Power Transformer Faults type.Fig. 4
It is the structure chart of the system of determination Power Transformer Faults type according to embodiment of the present invention.As shown in figure 4, determining electric power
The system 400 of transformer fault type includes that data divide device 401, data prediction device 402, fault type correlation rule
Excavating gear 403 and Power Transformer Faults determining device 404.
Data divide device 401, and which is used for collecting and integration Power Transformer Faults data, and divide electric power change
The fault characteristic value and fault type of depressor.Wherein, the fault characteristic value that the data divide device division is power transformer
The content of six kinds of dissolved gas in oil, six kinds of dissolved gas are respectively methane CH4, acetylene C2H2, ethene C2H, ethane
C2H6, hydrogen H2And total hydrocarbon.The data divide the fault type of device division：Cryogenic overheating, overheated middle temperature, hyperthermia and superheating,
Low energy is discharged and high-energy discharge.Preferably, the data divide device collecting and integration bag to Power Transformer Faults data
Include：The combing of unstructured data, Unified coding rule formulation, the typing of unstructured data, data cleansing process and
Transformer data characteristics statistical analysis.
Data prediction device 402, which is used for pre-processing above-mentioned Power Transformer Faults data, the pretreatment
Referring to carries out discretization to Power Transformer Faults data.
Preferably, data prediction device 402 carries out discretization to Power Transformer Faults data, including fault characteristic value
Discretization and fault type discretization.
The discretization of fault characteristic value refers to the discretization of the content of six kinds of dissolved gas in electric power transformer oil, described
Discretization method of the discretization of fault characteristic value using equidistant division.Due to hydrogen in the normal transformer oil of power transformer
Different with the prohibitive content of hydrocarbon gas, concrete demand value is as shown in table 1, if adopting identical area to all properties value
Between, then lack certain specific aim, therefore after experiment, with the fault alarm value size of the content of every kind of gas as interval long
Degree, when centrifugal pump is 0, represents gas content in normal range (NR), when centrifugal pump is the integer more than 0, represents that gas contains
Amount is beyond fault alarm value, and centrifugal pump is equal to the multiple more than fault alarm value.
The prohibitive content of hydrogen and hydrocarbon gas in 1 normal transformer oil of table
The discretization of fault type, the discretization of the fault type adopt boolean's discretization, when power transformer occurs
It is 1 during the failure of the type, is otherwise 0.
Fault type association rule mining device 403, which is used for based on pretreated Power Transformer Faults data, profit
The correlation rule between fault characteristic value and fault type is excavated with Apriori mining algorithms, the pass of final fault type is generated
Connection rule set.
Power Transformer Faults determining device 404, which is used for determining electricity according to the Association Rules of final fault type
The fault type of power transformer.
Fig. 2 is the flow chart of the correlation rule for excavating fault characteristic value and fault type according to embodiment of the present invention.
As shown in Fig. 2 fault type association rule mining device 403 excavates the association of Power Transformer Faults characteristic quantity and fault type
The method of rule is from the beginning of step S201.
In step S201, for each fault type in pretreated Power Transformer Faults data, foundation sets
Fixed minimum support, determines fault characteristic value of the support more than or equal to minimum support.
In step S202, for each fault type in pretreated Power Transformer Faults data, foundation sets
Fixed minimum reliability, support more than or equal to minimum support fault characteristic value in determine reliability be more than or
Equal to the fault characteristic value of minimum reliability, so that it is determined that the sample data of final fault characteristic value and fault type.
In step S203, by the relation net between the final fault characteristic value for determining and the sample data of fault type
Network figure is shown, and determines min-link number.
Fig. 3 A and 3B are the content of six kinds of dissolved gas in electric power transformer oil according to embodiment of the present invention and five kinds
Network of the fault type under different linking number.As shown in figs.3 a and 3b, the size of min-link number is adjusted, then power transformer
Relational network figure in device oil between the content of six kinds of dissolved gas and five kinds of fault types can occur respective change therewith, minimum
Link number becomes big, and the quantity of link is reduced.Before the Association Rules of fault type are generated, according to support and reliability
Setting, selects optimal min-link number.
In step S204, after determining min-link number, by the final fault characteristic value for determining and the sample of fault type
Data are divided into training part and part of detecting, and training part is used for excavating between Power Transformer Faults characteristic quantity and fault type
Correlation rule, so as to generate the Association Rules of fault type, and part of detecting is used for detecting the fault type that generates
The degree of accuracy of Association Rules, it is determined that the Association Rules of final fault type.
(1) train part
Training part proportion is 70%, and distribution is random.Training part be used for excavating in electric power transformer oil six kinds it is molten
Solution gas content and five kinds of fault types correlation rule, the content with six kinds of dissolved gas as preceding paragraph, five kinds of fault types
For consequent, minimum support and minimum reliability are set, power transformer are excavated with the data mining classical algorithm of Apriori
The correlation rule of the content of six kinds of dissolved gas and five kinds of fault types in oil, selected part are as shown in table 2.
2 association rule mining partial results of table
After excavating the content of six kinds of dissolved gas in electric power transformer oil and the correlation rule of five kinds of fault types, then root
The Association Rules of five kinds of fault types, the correlation rule of the fault type of generation are generated according to the consequent fault type of correlation rule
The size of collection is as shown in table 3.
The number of 3 five kinds of fault type rule set correlation rules of table
(2) part of detecting
Part of detecting proportion is 30%, and distribution is random.Part of detecting is used for five kinds of fault types that inspection is generated
The performance of Association Rules how.In part of detecting, the failure that the Association Rules of five kinds of fault types of Jing are predicted and reality
Border failure is contrasted, if the failure of prediction is identical with physical fault, can draw the reliability of the prediction.With hyperthermia and superheating
As a example by, intercepting partial table analysis as shown in table 4, in figure, three row represent physical fault respectively, predict the letter of failure and its prediction
Lai Du.
Predicting the outcome for 4 hyperthermia and superheating of table is compared with actual result
The accuracy of five kinds of transformer fault diagnosis can be drawn by statistical analysis, as shown in table 5.Set when accuracy meets
During definite value, Power Transformer Faults determining device 404 determines the Association Rules of final fault type.
The statistics of 5 association rule mining method Fault Diagnosis Method of Power Transformer of table
Embodiment two
Fig. 5 is the flow chart of the method for the determination Power Transformer Faults type according to another embodiment of the invention.Such as
Shown in Fig. 5, the method 500 of the determination Power Transformer Faults type in the present invention is from the beginning of step S501.
In step S501, Power Transformer Faults data are collected and integration, and divided the failure of power transformer
Characteristic quantity and fault type.Wherein, variable quantity of the fault characteristic value for six kinds of dissolved gas in electric power transformer oil, institute
State six kinds of dissolved gas and be respectively methane CH4, acetylene C2H2, ethene C2H4, ethane C2H6, hydrogen H2And total hydrocarbon, the dissolved gas
Variable quantity refer in units of day the changing value of measured dissolved gas content.The fault type is：Cryogenic overheating, in
Warm overheated, hyperthermia and superheating, low energy electric discharge and high-energy discharge.
Preferably, it is described that collecting for Power Transformer Faults data is included with integration：The combing of unstructured data, system
The formulation of one coding rule, the typing of unstructured data, data cleansing process and transformer data characteristics statistical analysis.
In step S502, above-mentioned Power Transformer Faults data are pre-processed, the pretreatment is to power transformer
Device failure carries out discretization.
Preferably, Power Transformer Faults data are carried out with discretization includes discretization and the fault type of fault characteristic value
Discretization.
The discretization of fault characteristic value refers to the discretization of the variable quantity of six kinds of dissolved gas in electric power transformer oil, institute
State the discretization method of the discretization using equidistant division of fault characteristic value.Due in the normal transformer oil of power transformer
The variable quantity limit value of hydrogen and hydrocarbon gas is different, if interval using identical to all properties value, lacks certain
Specific aim, therefore after experiment, the fault alarm value size with the variable quantity of every kind of gas as siding-to-siding block length, when centrifugal pump is 0
When, gas variation amount is represented in normal range (NR), when centrifugal pump is the integer more than 0, represent that gas variation amount exceeds failure
Alarming value, and centrifugal pump is equal to the multiple more than fault alarm value.
The discretization of fault type, the discretization of the fault type adopt boolean's discretization, when power transformer occurs
It is 1 during the failure of the type, is otherwise 0.
In step S503, based on pretreated Power Transformer Faults data, electricity is excavated using Apriori mining algorithms
The correlation rule of the variable quantity of six kinds of dissolved gas and five kinds of fault types in power transformer oil, generates final fault type
Association Rules.
In step S504, the fault type of power transformer is determined according to the Association Rules of final fault type.
Fig. 6 is the variable quantity of the excavation Detection Ssytem of Dissolved Gases in Power Transformer Oil Base according to another embodiment of the invention and failure
The flow chart of the correlation rule of type.As shown in fig. 6, excavating the variable quantity and fault type of Detection Ssytem of Dissolved Gases in Power Transformer Oil Base
Correlation rule method from the beginning of step S601.
In step S601, for each fault type in pretreated Power Transformer Faults data, foundation sets
Fixed minimum support, determines fault characteristic value of the support more than or equal to minimum support.
In step S602, for each fault type in pretreated Power Transformer Faults data, foundation sets
Fixed minimum reliability, support more than or equal to minimum support fault characteristic value in determine reliability be more than or
Equal to the fault characteristic value of minimum reliability, so that it is determined that the sample data of final fault characteristic value and fault type.
In step S603, by the relation net between the final fault characteristic value for determining and the sample data of fault type
Network figure is shown, and determines min-link number.
Fig. 7 A and 7B are the variable quantities of six kinds of dissolved gas in electric power transformer oil according to another embodiment of the invention
With network of five kinds of fault types under different linking number.As shown in figs. 7 a-b, the size of min-link number is adjusted, then electricity
Relational network figure in power transformer oil between the variable quantity of six kinds of dissolved gas and five kinds of fault types can occur accordingly therewith
Change, min-link number become big, and the quantity of link is reduced.Before the Association Rules of fault type are generated, basis is needed
The setting of degree of holding and reliability, selects optimal min-link number.
In step S604, after determining min-link number, by the final fault characteristic value for determining and the sample of fault type
Data are divided into training part and part of detecting, and training part is used for excavating between Power Transformer Faults characteristic quantity and fault type
Correlation rule, so as to generate the Association Rules of fault type, and part of detecting is used for detecting the fault type that generates
The degree of accuracy of Association Rules, it is determined that the Association Rules of final fault type.
(1) train part
Training part proportion is 70%, and distribution is random.Training part be used for excavating in electric power transformer oil six kinds it is molten
Solution gas variable quantity and five kinds of fault types correlation rule, the variable quantity with six kinds of dissolved gas as preceding paragraph, five kinds of failures
Type is consequent, arranges minimum support and minimum reliability, excavates electric power with the data mining classical algorithm of Apriori and becomes
The variable quantity and the correlation rule of five kinds of fault types of six kinds of dissolved gas in depressor oil.
After excavating the correlation rule of variable quantity and five kinds of fault types of six kinds of dissolved gas in electric power transformer oil, then
The Association Rules of five kinds of fault types are generated according to the consequent fault type of correlation rule.
(2) part of detecting
Part of detecting proportion is 30%, and distribution is random.Part of detecting is used for five kinds of fault types that inspection is generated
The performance of Association Rules how.In part of detecting, the failure that the Association Rules of five kinds of fault types of Jing are predicted and reality
Border failure is contrasted, if the failure of prediction is identical with physical fault, can draw the reliability of the prediction.
The accuracy of five kinds of transformer fault diagnosis can be drawn by statistical analysis, when accuracy meets setting value, really
The Association Rules of fixed final fault type.
According to a further aspect in the invention, the invention further relates to a kind of system for determining Power Transformer Faults type.Institute
State system identical with the system in embodiment one.
Unless otherwise stated, term (including scientific and technical terminology) used herein has to person of ordinary skill in the field
It is common to understand implication.Further it will be understood that the term limited with the dictionary being usually used, is appreciated that and which
The linguistic context of association area has consistent implication, and is not construed as Utopian or excessively formal meaning.
By reference to above embodiments describing the present invention.However, it is known in those skilled in the art, as
What subsidiary Patent right requirement was limited, except the present invention other embodiments disclosed above equally fall the present invention's
In the range of.
Normally, all terms for using in the claims are all solved in the usual implication of technical field according to them
Release, unless it is characterized in that clearly being defined in addition.All of reference " one/described/be somebody's turn to do【Device, component etc.】" all by
At least one of described device, component etc. example is construed to openly, unless otherwise expressly specified.It is disclosed herein any
The step of method, all need not be run with disclosed accurate order, unless explicitly stated otherwise.
Claims (8)
1. a kind of method for determining Power Transformer Faults type, methods described includes：
Power Transformer Faults data are collected and integration, and is divided fault characteristic value and the failure classes of power transformer
Type, wherein：
Content and variable quantity of the fault characteristic value for six kinds of dissolved gas in electric power transformer oil, six kinds of dissolved gas point
Wei not methane CH4, acetylene C2H2, ethene C2H4, ethane C2H6, hydrogen H2And total hydrocarbon；
The fault type is：Cryogenic overheating, overheated middle temperature, hyperthermia and superheating, low energy electric discharge and high-energy discharge；
Above-mentioned Power Transformer Faults data are pre-processed, the pretreatment refers to and Power Transformer Faults data are carried out
Discretization；；
Based on pretreated Power Transformer Faults data, Power Transformer Faults are excavated using Apriori mining algorithms special
Correlation rule between the amount of levying and fault type, generates the Association Rules of final fault type, the final failure of the generation
The method of the Association Rules of type includes：
For each fault type in pretreated Power Transformer Faults data, according to the minimum support of setting,
Determine fault characteristic value of the support more than or equal to minimum support；
For each fault type in pretreated Power Transformer Faults data, according to the minimum reliability of setting,
In fault characteristic value of the support more than or equal to minimum support, determine reliability more than or equal to minimum reliability
Fault characteristic value, so that it is determined that the sample data of final fault characteristic value and fault type；
Relation between the final fault characteristic value for determining and the sample data of fault type is shown with network, really
Determine min-link number；
After determining min-link number, the sample data of the final fault characteristic value for determining and fault type is divided into into training department
Divide and part of detecting, training part is used for excavating the correlation rule between Power Transformer Faults characteristic quantity and fault type, so as to
The Association Rules of fault type are generated, and part of detecting is used for the accurate of the Association Rules of the fault type that detection is generated
Degree, it is determined that the Association Rules of final fault type；And
Power Transformer Faults type is determined according to the Association Rules of final fault type.
2. method according to claim 1, it is characterised in that it is described Power Transformer Faults data are collected with it is whole
Conjunction includes：The combing of unstructured data, the formulation of Unified coding rule, the typing of unstructured data, data cleansing process
And transformer data characteristics statistical analysis.
3. method according to claim 1, it is characterised in that carrying out discretization to Power Transformer Faults data includes：
The discretization of fault characteristic value, the discretization of the fault characteristic value adopt the discretization method of equidistant division, with every
The fault alarm value size of the content and variable quantity of planting gas is siding-to-siding block length, when centrifugal pump is 0, represents gas content and change
Change amount when centrifugal pump is the integer more than 0, represents that gas content and variable quantity exceed fault alarm value in normal range (NR),
And centrifugal pump is equal to the multiple more than fault alarm value；And
The discretization of fault type, the discretization of the fault type adopt boolean's discretization, when such occurs in power transformer
It is 1 during the failure of type, is otherwise 0.
4. method according to claim 1, it is characterised in that excavate power transformer using Apriori mining algorithms therefore
Correlation rule between barrier characteristic quantity and fault type, including the content and fault type of excavating Detection Ssytem of Dissolved Gases in Power Transformer Oil Base
Correlation rule with excavate Detection Ssytem of Dissolved Gases in Power Transformer Oil Base variable quantity and fault type correlation rule.
5. a kind of system for determining Power Transformer Faults type, the system (includes：
Data divide device, and which is used for collecting and integration Power Transformer Faults data, and divides power transformer
Fault characteristic value and fault type, wherein；
Content and variable quantity of the fault characteristic value for six kinds of dissolved gas in electric power transformer oil, six kinds of dissolved gas point
Wei not methane CH4, acetylene C2H2, ethene C2H4, ethane C2H6, hydrogen H2And total hydrocarbon；
The fault type is：Cryogenic overheating, overheated middle temperature, hyperthermia and superheating, low energy electric discharge and high-energy discharge；
Data prediction device, which is used for pre-processing above-mentioned Power Transformer Faults data, and it is right that the pretreatment is referred to
Power Transformer Faults data carry out discretization；
Fault type association rule mining device, which is used for based on pretreated Power Transformer Faults data, utilizes
Apriori mining algorithms excavate the correlation rule between Power Transformer Faults characteristic quantity and fault type, generate final failure
The Association Rules of type, the fault type association rule mining device generate the Association Rules of final fault type
Method includes：For each fault type in pretreated Power Transformer Faults data, according to the most ramuscule of setting
Degree of holding, determines fault characteristic value of the support more than or equal to minimum support；
For each fault type in pretreated Power Transformer Faults data, according to the minimum reliability of setting,
In fault characteristic value of the support more than or equal to minimum support, determine reliability more than or equal to minimum reliability
Fault characteristic value, so that it is determined that the sample data of final fault characteristic value and fault type；
Relation between the final fault characteristic value for determining and the sample data of fault type is shown with network, really
Determine min-link number；
After determining min-link number, the sample data of the final fault characteristic value for determining and fault type is divided into into training department
Divide and part of detecting, training part is used for excavating the correlation rule between Power Transformer Faults characteristic quantity and fault type, so as to
The Association Rules of fault type are generated, and part of detecting is used for the accurate of the Association Rules of the fault type that detection is generated
Degree, it is determined that the Association Rules of final fault type；And
Power Transformer Faults determining device, which is used for determining power transformer according to the Association Rules of final fault type
Fault type.
6. system according to claim 5, it is characterised in that the data divide device to Power Transformer Faults data
Collect and include with integration：The combing of unstructured data, the formulation of Unified coding rule, the typing of unstructured data, number
According to cleaning process and transformer data characteristics statistical analysis.
7. system according to claim 5, it is characterised in that data prediction device enters to Power Transformer Faults data
Row discretization includes：
The discretization of fault characteristic value, which adopts the discretization method of equidistant division, with the content of every kind of gas and variable quantity
Fault alarm value size be siding-to-siding block length, when centrifugal pump is 0, represent gas content and variable quantity in normal range (NR), when
When centrifugal pump is the integer more than 0, represents that gas content and variable quantity exceed fault alarm value, and centrifugal pump is equal to more than failure
The multiple of alarming value；
The discretization of fault type, which adopts boolean's discretization, i.e. data to be mapped on Boolean, when being somebody's turn to do occurs in power transformer
It is 1 during the failure of type, is otherwise 0.
8. system according to claim 5, it is characterised in that fault correlation rule digging device is excavated using Apriori
Algorithm excavates the correlation rule of Power Transformer Faults characteristic quantity and fault type, including solution gas in excavation electric power transformer oil
The correlation rule of the content of body and fault type and the variable quantity and fault type for excavating Detection Ssytem of Dissolved Gases in Power Transformer Oil Base
Correlation rule.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610875616.2A CN106526352B (en) | 2016-09-30 | 2016-09-30 | Method and system for determining fault type of power transformer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610875616.2A CN106526352B (en) | 2016-09-30 | 2016-09-30 | Method and system for determining fault type of power transformer |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106526352A true CN106526352A (en) | 2017-03-22 |
| CN106526352B CN106526352B (en) | 2020-10-13 |
Family
ID=58333050
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610875616.2A Active CN106526352B (en) | 2016-09-30 | 2016-09-30 | Method and system for determining fault type of power transformer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106526352B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107047250A (en) * | 2017-04-27 | 2017-08-18 | 上海博历机械科技有限公司 | A kind of irrigation system of high security |
| CN107907783A (en) * | 2017-12-19 | 2018-04-13 | 西安交通大学 | Transformer fault integrated diagnostic system and diagnostic method based on fuzzy association rules |
| CN108629528A (en) * | 2018-06-20 | 2018-10-09 | 国网江苏省电力有限公司电力科学研究院 | Quality of Transformer problem analysis method based on Apriori algorithm |
| CN109191011A (en) * | 2018-10-12 | 2019-01-11 | 中国电力科学研究院有限公司 | A kind of DC bushing state evaluating method and device based on Apriori algorithm |
| CN109254219A (en) * | 2018-11-22 | 2019-01-22 | 国网湖北省电力有限公司电力科学研究院 | A kind of distribution transforming transfer learning method for diagnosing faults considering multiple factors Situation Evolution |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7370033B1 (en) * | 2002-05-17 | 2008-05-06 | Oracle International Corporation | Method for extracting association rules from transactions in a database |
| CN103020256A (en) * | 2012-12-21 | 2013-04-03 | 电子科技大学 | Association rule mining method of large-scale data |
| CN104730378A (en) * | 2015-02-13 | 2015-06-24 | 国家电网公司 | Internal transformer composite-defect fuzzy diagnostic method based on gas dissolved in oil |
| CN105843210A (en) * | 2016-03-22 | 2016-08-10 | 清华大学 | Power transformer defect information data mining method |
| CN105891629A (en) * | 2016-03-31 | 2016-08-24 | 广西电网有限责任公司电力科学研究院 | Transformer equipment fault identification method |
-
2016
- 2016-09-30 CN CN201610875616.2A patent/CN106526352B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7370033B1 (en) * | 2002-05-17 | 2008-05-06 | Oracle International Corporation | Method for extracting association rules from transactions in a database |
| CN103020256A (en) * | 2012-12-21 | 2013-04-03 | 电子科技大学 | Association rule mining method of large-scale data |
| CN104730378A (en) * | 2015-02-13 | 2015-06-24 | 国家电网公司 | Internal transformer composite-defect fuzzy diagnostic method based on gas dissolved in oil |
| CN105843210A (en) * | 2016-03-22 | 2016-08-10 | 清华大学 | Power transformer defect information data mining method |
| CN105891629A (en) * | 2016-03-31 | 2016-08-24 | 广西电网有限责任公司电力科学研究院 | Transformer equipment fault identification method |
Non-Patent Citations (1)
| Title |
|---|
| 丁茜: "基于关联规则的变压器故障诊断研究", 《中国优秀硕士学位论文全文数据库（电子期刊）》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107047250A (en) * | 2017-04-27 | 2017-08-18 | 上海博历机械科技有限公司 | A kind of irrigation system of high security |
| CN107907783A (en) * | 2017-12-19 | 2018-04-13 | 西安交通大学 | Transformer fault integrated diagnostic system and diagnostic method based on fuzzy association rules |
| CN107907783B (en) * | 2017-12-19 | 2019-08-13 | 西安交通大学 | Transformer fault integrated diagnostic system and diagnostic method based on fuzzy association rules |
| CN108629528A (en) * | 2018-06-20 | 2018-10-09 | 国网江苏省电力有限公司电力科学研究院 | Quality of Transformer problem analysis method based on Apriori algorithm |
| CN109191011A (en) * | 2018-10-12 | 2019-01-11 | 中国电力科学研究院有限公司 | A kind of DC bushing state evaluating method and device based on Apriori algorithm |
| CN109254219A (en) * | 2018-11-22 | 2019-01-22 | 国网湖北省电力有限公司电力科学研究院 | A kind of distribution transforming transfer learning method for diagnosing faults considering multiple factors Situation Evolution |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106526352B (en) | 2020-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106526352A (en) | Method and system for determining power transformer fault types | |
| US20210209263A1 (en) | Tunnel tunneling feasibility prediction method and system based on tbm rock-machine parameter dynamic interaction mechanism | |
| CN106054104A (en) | Intelligent ammeter fault real time prediction method based on decision-making tree | |
| CN104091035B (en) | Health monitoring method for effective loads of space station based on data-driven algorithm | |
| CN103077447B (en) | A kind of subway work Safety management decision making method based on Bayesian network | |
| CN102252843B (en) | Assessment method for rolling bearing performance variation | |
| Zhu et al. | Remaining-useful-lifetime and system-remaining-profit based importance measures for decisions on preventive maintenance | |
| CN106952003B (en) | Method for predicting surrounding rock damage type of underground cavern of stratified rock mass in high ground stress region | |
| CN109254219B (en) | A kind of distribution transforming transfer learning method for diagnosing faults considering multiple factors Situation Evolution | |
| US20220083655A1 (en) | Chip security analysis method based on petri net | |
| Jia et al. | A fast contingency screening technique for generation system reliability evaluation | |
| Meier et al. | Power system data management and analysis using synchrophasor data | |
| CN109781044A (en) | Comprehensive gradual approximation early warning method for slope instability | |
| WO2018138973A1 (en) | System operation decision-making assistance device and method | |
| Sun et al. | Data-driven approach for spatiotemporal distribution prediction of fault events in power transmission systems | |
| Liu et al. | Assessing the nonequilibrium of decaying turbulence with reversed initial fields | |
| Song et al. | Multifractional and long-range dependent characteristics for remaining useful life prediction of cracking gas compressor | |
| CN108829794A (en) | Alert analysis method based on interval graph | |
| CN103675357A (en) | Anemometer fault warning method and system | |
| CN106056305A (en) | Power generation system reliability rapid assessment method based on state clustering | |
| CN106126949A (en) | A kind of steam turbine generator running status appraisal procedure | |
| Huang et al. | Study on viscoelastic deformation monitoring index of an RCC gravity dam in an alpine region using orthogonal test design | |
| Baker et al. | Optimal Control of Transient Flows in Pipeline Networks with Heterogeneous Mixtures of Hydrogen and Natural Gas | |
| CN104537437A (en) | Power equipment state maintaining predicting method based on genetic algorithm | |
| Sorooshian | Alternative method for evaluation of dagang deep drilling applications |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |