CN109102031A - A kind of oil-immersed transformer fault detection method neural network based - Google Patents

A kind of oil-immersed transformer fault detection method neural network based Download PDF

Info

Publication number
CN109102031A
CN109102031A CN201810991158.8A CN201810991158A CN109102031A CN 109102031 A CN109102031 A CN 109102031A CN 201810991158 A CN201810991158 A CN 201810991158A CN 109102031 A CN109102031 A CN 109102031A
Authority
CN
China
Prior art keywords
transformer
oil
neural network
metal
fault
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.)
Pending
Application number
CN201810991158.8A
Other languages
Chinese (zh)
Inventor
刘喆
杨靖
赵靓玮
余鹏程
张英
吴国卿
牧灏
李军卫
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guizhou Power Grid Co Ltd
Original Assignee
Guizhou Power Grid Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guizhou Power Grid Co Ltd filed Critical Guizhou Power Grid Co Ltd
Priority to CN201810991158.8A priority Critical patent/CN109102031A/en
Publication of CN109102031A publication Critical patent/CN109102031A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Geometry (AREA)
  • Computer Hardware Design (AREA)
  • Housings And Mounting Of Transformers (AREA)
  • Testing Electric Properties And Detecting Electric Faults (AREA)
  • Protection Of Transformers (AREA)

Abstract

The invention discloses a kind of oil-immersed transformer fault detection methods neural network based, it includes the following steps: (1) that the operation troubles of oil-immersed transformer is classified；Step 2, structure choice metal and nonmetalloid according to oil-immersed transformer；Step 3 encodes fault type；Step 4, acquisition metal and metalloid content data；By acquisition failure transformer oil and the metal and metalloid content data of operating transformer oil, the training and test sample of neural network needs are established；Step 5 classifies data；Establish the relation table of metal and metalloid content in transformer fault state and normal operating condition and transformer oil；Step 6 differentiates transformer fault based on neural network；It solves the transformer fault that the analysis of the existing technology by oil dissolved gas is found and there is lag；After dissolved gas the problems such as the abnormal unobvious insulation status that can not accurately judge transformer.

Description

A kind of oil-immersed transformer fault detection method neural network based
Technical field
The invention belongs to transformer fault detection technique more particularly to a kind of oil-immersed transformer events neural network based Hinder detection method.
Background technique
Transformer is one of most important equipment in electric system, carries the weight of voltage transformation, electric energy distribution and transmission Appoint, operating condition is related to the safety and stablization of entire electric system.It is accurate to detect and judge the whether faulty hair of transformer It is raw, it is most important for the security and reliability for improving electric system.With sensor technology, artificial intelligence technology and distribution The continuous development of data processing technique, Intelligent Diagnosis Technology can also find transformer in the fault diagnosis of for transformer Failure.
Currently, transformer fault can be divided into short trouble, discharge fault, insulation fault, iron core failure etc., transformer is detected The method of failure currently mainly has three-ratio method, dissolved gas analysis method, red, orange, green, blue, yellow (ROGBY) etc..When oil-immersed transformer is run, It can be cracked into gas, existed inside power transformer because of the effect gradually aging of many factors such as electricity, heat and local arc When the hot-spot or shelf depreciation of latency, the speed of gas generation can be accelerated.Therefore, the dissolution in analysis oil can be passed through Gas carries out fault diagnosis to power transformer, and relevant quantitative, qualitative analysis there has been considerable warp in practical applications It tests.But property (the excessively hot or electric discharge of power transformer interior fault is detected according to the gas componant dissolved in transformer oil Property) there is also some shortcomings.Such as: the transformer fault many places found by the analysis of oil dissolved gas have gone out in transformer Now after obvious exception, but at this time transformer fault often than more serious；It, can not be quasi- if abnormal unobvious after dissolved gas The true insulation status for judging transformer, it is also possible to which the maintenance of delay faults transformer leads to that more serious failure occurs；Have Transformer discovery failure and hang cover maintenance after, problem occurs again, leads to maintenance repeatedly, can not but solve transformer and ask Topic.
Summary of the invention
The technical problem to be solved by the present invention is providing a kind of oil-immersed transformer fault detection side neural network based Method has been gone out with solving the transformer fault many places that the analysis of the existing technology by oil dissolved gas is found in transformer Now after obvious exception, but at this time transformer fault often than more serious；It, can not be quasi- if abnormal unobvious after dissolved gas The true insulation status for judging transformer, it is also possible to which the maintenance of delay faults transformer leads to that more serious failure occurs；Have Transformer discovery failure and hang cover maintenance after, problem occurs again, leads to the technical problems such as maintenance repeatedly.
The technical scheme is that
A kind of oil-immersed transformer fault detection method neural network based, it includes:
Step 1 classifies the operation troubles of oil-immersed transformer；
Step 2, structure choice metal and nonmetalloid according to oil-immersed transformer；Select Cu, Fe, Al, Mn, Sn gold Category be characterized metal, select Si for it is nonmetallic be characterized it is nonmetallic；
Step 3 encodes fault type；
Step 4, acquisition metal and metalloid content data；Pass through acquisition failure transformer oil and the gold of operating transformer oil Category and metalloid content data establish the training and test sample of neural network needs；
Step 5 classifies data；Establish metal in transformer fault state and normal operating condition and transformer oil With the relation table of metalloid content；
Step 6 differentiates transformer fault based on neural network.
The operation troubles of oil-immersed transformer is classified described in step 1, is divided into: cryogenic overheating, medium temperature overheat, height Warm overheat, shelf depreciation, low energy electric discharge and high-energy discharge failure.
Fault type is encoded described in step 3, the coding are as follows: cryogenic overheating [000001], medium temperature overheat [000010], hyperthermia and superheating [000100], shelf depreciation [001000], low energy electric discharge [010000], high-energy discharge [100000] and normal condition is encoded to [000000].
Metal and metalloid content in transformer fault state and normal operating condition and transformer oil are established described in step 5 Relation table are as follows:
Upper table is the threshold value of metal and metalloid content in transformer, and content is more than that threshold values is failure.
Transformer fault is carried out to sentence method for distinguishing based on neural network described in step 6 are as follows:
Step 6.1 carries out data prediction to training sample data, and the data prediction is that data are normalized Processing is based on sliding average filtering algorithm rejecting abnormalities data；
Step 6.2 establishes BP neural network model, includes input layer, hidden layer and output layer, hidden layer neuron numberL is the number of hidden nodes, and n is input layer number, and m is output layer number of nodes, and one between z is 1~15 is whole Number, n=6, m=7, l 15；
Step 6.3 trains network using Levenberg-Marquardt method；
Step 6.4, test network: performance test sample verifies the effective of the transformer fault diagnosis based on BP neural network Property；
Step 6.5, output as a result, obtain whether faulty judgement, such as it is faulty, export fault type.
The invention has the advantages that:
Transformer fault judgment method according to the present invention judges transformer fault side based on dissolved gas method with existing Method is different, essentially consist in based in transformer oil metal and nonmetalloid content judge failure, oil immersed type can be become Depressor carries out offline or on-line fault diagnosis, has strong real-time high reliability；To guarantee that the safety of transformer can By operation；Transformer fault many places that the analysis of the existing technology by oil dissolved gas is found are solved in transformer After having there is obvious exception, but at this time transformer fault often than more serious；If abnormal unobvious, nothing after dissolved gas Method accurately judges the insulation status of transformer, it is also possible to which the maintenance of delay faults transformer leads to that more serious event occurs Barrier；In discovery failure and after hanging cover maintenance, problem occurs some transformers again, leads to the technical problems such as maintenance repeatedly.
Detailed description of the invention
Fig. 1 is flow diagram of the present invention；
Fig. 2 is diagnostic model schematic diagram of the present invention.
Specific embodiment
A kind of oil-immersed transformer fault detection method neural network based, it includes:
Step Step 1, step the operation troubles of oil-immersed transformer is classified, be divided into: cryogenic overheating, medium temperature overheat, Hyperthermia and superheating, shelf depreciation, low energy electric discharge and high-energy discharge failure；Diagnostic model is designed, is set according to neural network structure Count the detection model that is out of order；
Step Step 2 selects metal and nonmetalloid, by contained in transformer in failure transformer and operation Metal and nonmetalloid are analyzed, and select Cu, Fe, Al, Mn, Sn these types and the closely related metal of fault type for spy Levy metal, select Si this it is a kind of it is closely related with failure it is nonmetallic be characterized it is nonmetallic.
Step Step 3, fault type is encoded, fault type are as follows: cryogenic overheating (T < 300 DEG C), medium temperature overheat (300 DEG C ≤ T < 700 DEG C), hyperthermia and superheating (700 DEG C≤T), shelf depreciation, low energy electric discharge, high-energy discharge.Fault type is compiled Code: cryogenic overheating [000001], medium temperature overheat [000010], hyperthermia and superheating [000100], shelf depreciation [001000], low energy Electric discharge [010000], high-energy discharge [100000], normal condition are encoded to [000000].
Step Step 4 acquires metal and metalloid content data, passes through acquisition failure transformer oil and operating transformer The metal and metalloid content data of oil, establish the training and test sample of neural network needs.
Step Step 5, classifies to data, establishes transformer fault state and normal operating condition and transformer oil The relationship of middle metal and metalloid content.
The following table 1 is the threshold value of the metal and metalloid content in normal operating transformer oil, has been more than threshold value, has shown transformation Device is likely to occur failure.
Upper table is the threshold value of metal and metalloid content in failure transformer.
Step Step 6 diagnoses transformer fault based on neural network.Specifically, it is based on neural network, to step Fault threshold in rapid 4 data and step 5 judges whether that transformer breaks down based on neural network.
Fault distinguishing method neural network based
A. it constructs training sample and carries out data prediction, data prediction is mainly normalized data, keeps away Exempting from data bulk grade difference causes result error excessive, is based on sliding average filtering algorithm, and rejecting abnormalities data in this way can The convergence rate of accelerans network；
B. BP neural network model is established, includes input layer, hidden layer, output layer, wherein hidden layer neuron numberL is the number of hidden nodes, and n is input layer number, and m is output layer number of nodes, z can be taken as 1~15 between one A integer, because of n=6, m=7, so l can be taken as 15；
C. network is trained, using Levenberg-Marquardt method training network；
D. test network, performance test sample verify the validity of the transformer fault diagnosis based on BP neural network；
E. output as a result, obtain whether faulty judgement, such as it is faulty, export fault type.Enable input layer original defeated Enter data vector D={ d1,d2,…,dn, the fault diagnosis vector of neural network is X={ x1,x2,…,xn, O={ o1, o2,…,om, hidden layer neuron number is H={ b1,b2,…,bl, i is sample number；With j group data sample to BP nerve net Network is trained, and is then tested with trained network to i-j group sample, and φ (x) is hidden neuron excitation function, θk (k=1 ..., l) is the threshold value of neuron,For the weight of input layer and hidden layer,For the weight of hidden layer and output layer, have Input/output relationO, that is, physical fault type approximation.

Claims (5)

1. a kind of oil-immersed transformer fault detection method neural network based, it includes:
Step 1 classifies the operation troubles of oil-immersed transformer；
Step 2, structure choice metal and nonmetalloid according to oil-immersed transformer；Select Cu, Fe, Al, Mn, Sn metal for Characteristic metal, select Si for it is nonmetallic be characterized it is nonmetallic；
Step 3 encodes fault type；
Step 4, acquisition metal and metalloid content data；By acquisition failure transformer oil and operating transformer oil metal with Metalloid content data establish the training and test sample of neural network needs；
Step 5 classifies data；Establish transformer fault state and normal operating condition and metal in transformer oil and non- The relation table of tenor；
Step 6 differentiates transformer fault based on neural network.
2. a kind of oil-immersed transformer fault detection method neural network based according to claim 1, feature exist In: the operation troubles of oil-immersed transformer is classified described in step 1, is divided into: cryogenic overheating, medium temperature overheat, high temperature mistake Heat, shelf depreciation, low energy electric discharge and high-energy discharge failure.
3. a kind of oil-immersed transformer fault detection method neural network based according to claim 1, feature exist In: fault type is encoded described in step 3, the coding are as follows: cryogenic overheating [000001], medium temperature overheat [000010], Hyperthermia and superheating [000100], shelf depreciation [001000], low energy electric discharge [010000], high-energy discharge [100000] and normally State encoding is [000000].
4. a kind of oil-immersed transformer fault detection method neural network based according to claim 1, feature exist In: the pass of metal and metalloid content in transformer fault state and normal operating condition and transformer oil is established described in step 5 It is table are as follows:
Upper table is the threshold value of metal and metalloid content in transformer, and content is more than that threshold values is failure.
5. a kind of oil-immersed transformer fault detection method neural network based according to claim 1, feature exist In: transformer fault is carried out to sentence method for distinguishing based on neural network described in step 6 are as follows:
Step 6.1 carries out data prediction to training sample data, and the data prediction is that data are normalized with place Reason is based on sliding average filtering algorithm rejecting abnormalities data；
Step 6.2 establishes BP neural network model, includes input layer, hidden layer and output layer, hidden layer neuron numberL is the number of hidden nodes, and n is input layer number, and m is output layer number of nodes, and one between z is 1~15 is whole Number, n=6, m=7, l 15；
Step 6.3 trains network using Levenberg-Marquardt method；
Step 6.4, test network: the validity of transformer fault diagnosis of the performance test sample verifying based on BP neural network；
Step 6.5, output as a result, obtain whether faulty judgement, such as it is faulty, export fault type.
CN201810991158.8A 2018-08-28 2018-08-28 A kind of oil-immersed transformer fault detection method neural network based Pending CN109102031A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810991158.8A CN109102031A (en) 2018-08-28 2018-08-28 A kind of oil-immersed transformer fault detection method neural network based

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810991158.8A CN109102031A (en) 2018-08-28 2018-08-28 A kind of oil-immersed transformer fault detection method neural network based

Publications (1)

Publication Number Publication Date
CN109102031A true CN109102031A (en) 2018-12-28

Family

ID=64864122

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810991158.8A Pending CN109102031A (en) 2018-08-28 2018-08-28 A kind of oil-immersed transformer fault detection method neural network based

Country Status (1)

Country Link
CN (1) CN109102031A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110689069A (en) * 2019-09-25 2020-01-14 贵州电网有限责任公司 Transformer fault type diagnosis method based on semi-supervised BP network
CN110879373A (en) * 2019-12-12 2020-03-13 国网电力科学研究院武汉南瑞有限责任公司 Oil-immersed transformer fault diagnosis method with neural network and decision fusion
CN111830438A (en) * 2019-04-19 2020-10-27 宁波奥克斯高科技有限公司 Transformer fault detection method and transformer
CN114152872A (en) * 2021-12-01 2022-03-08 湖南大学 Oil-immersed metal equipment health condition diagnosis method based on metal ion detection

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103364357A (en) * 2013-08-05 2013-10-23 国家电网公司 Detection method for content of trace metal elements in transformer oil by using acid liquor dissolution method to treat transformer oil sample
CN103389430A (en) * 2013-08-06 2013-11-13 华北电力大学 Oil-immersed type transformer fault detection method based on Bayesian discrimination theory
CN103576061A (en) * 2013-10-17 2014-02-12 国家电网公司 Method for discharge fault diagnosis of transformer
CN104538926A (en) * 2014-12-23 2015-04-22 黑龙江宏宇电站设备有限公司 Protection device for oil-immersed power transformer and protection method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103364357A (en) * 2013-08-05 2013-10-23 国家电网公司 Detection method for content of trace metal elements in transformer oil by using acid liquor dissolution method to treat transformer oil sample
CN103389430A (en) * 2013-08-06 2013-11-13 华北电力大学 Oil-immersed type transformer fault detection method based on Bayesian discrimination theory
CN103576061A (en) * 2013-10-17 2014-02-12 国家电网公司 Method for discharge fault diagnosis of transformer
CN104538926A (en) * 2014-12-23 2015-04-22 黑龙江宏宇电站设备有限公司 Protection device for oil-immersed power transformer and protection method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
闫晨光 等: "电力变压器油箱内部故障压力特征建模及仿真", 《中国电机工程学报》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111830438A (en) * 2019-04-19 2020-10-27 宁波奥克斯高科技有限公司 Transformer fault detection method and transformer
CN111830438B (en) * 2019-04-19 2022-10-11 宁波奥克斯高科技有限公司 Transformer fault detection method and transformer
CN110689069A (en) * 2019-09-25 2020-01-14 贵州电网有限责任公司 Transformer fault type diagnosis method based on semi-supervised BP network
CN110879373A (en) * 2019-12-12 2020-03-13 国网电力科学研究院武汉南瑞有限责任公司 Oil-immersed transformer fault diagnosis method with neural network and decision fusion
CN114152872A (en) * 2021-12-01 2022-03-08 湖南大学 Oil-immersed metal equipment health condition diagnosis method based on metal ion detection
CN114152872B (en) * 2021-12-01 2022-11-29 湖南大学 Oil-immersed metal equipment health condition diagnosis method based on metal ion detection

Similar Documents

Publication Publication Date Title
CN109102031A (en) A kind of oil-immersed transformer fault detection method neural network based
CN102981108B (en) Transformer internal insulation aging diagnosis method based on multi-feature information fusion technology
Dukarm Transformer oil diagnosis using fuzzy logic and neural networks
CN103389430B (en) A kind of oil-filled transformer fault detection method based on Bayesian discrimination theory
CN105242155A (en) Transformer fault diagnosis method based on entropy weight method and grey correlation analysis
CN103926490B (en) A kind of power transformer error comprehensive diagnosis method with self-learning function
CN112598298A (en) Power transformer health management system and management method
Li et al. An integrated method of set pair analysis and association rule for fault diagnosis of power transformers
CN101251564A (en) Method for diagnosis failure of power transformer using extendible horticulture and inelegance collection theory
CN105467971B (en) A kind of second power equipment monitoring system and method
CN103778575A (en) Transformer state evaluation method and system
CN106569069A (en) Power transformer fault diagnosis method
CN109490685B (en) Early defect early warning method of transformer based on-line monitoring of dissolved gas in oil
CN110188309A (en) Oil-immersed power transformer defect method for early warning based on Hidden Markov Model
CN105954615A (en) State assessment method and assessment system after transformer short circuit
Khan et al. ANFIS based identification and location of paper insulation faults of an oil immersed transformer
CN116227788A (en) Fault diagnosis and evaluation method and system for substation power equipment
CN106291149A (en) A kind of live detection method of 10kV switch cubicle
CN104360194A (en) Fault diagnosis method for smart power grid
CN214310334U (en) Transformer oil multi-dimensional data acquisition on-line monitoring diagnostic system
CN110321520A (en) A kind of transformer state evaluation method based on Weighted distance diagnostic method
CN106644436B (en) A kind of assessment method of breaker mechanic property
Shamsudin et al. Investigation on online DGA monitoring to determine transformer health index using machine learning
CN208488118U (en) A kind of intelligent transformer Integrated Fault Diagnosis System
CN111272222A (en) Transformer fault diagnosis method based on characteristic quantity set

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20181228

RJ01 Rejection of invention patent application after publication