CN105701543B - A kind of traditional transformer state monitoring method based on probabilistic neural network - Google Patents
A kind of traditional transformer state monitoring method based on probabilistic neural network Download PDFInfo
- Publication number
- CN105701543B CN105701543B CN201610020821.0A CN201610020821A CN105701543B CN 105701543 B CN105701543 B CN 105701543B CN 201610020821 A CN201610020821 A CN 201610020821A CN 105701543 B CN105701543 B CN 105701543B
- Authority
- CN
- China
- Prior art keywords
- traditional
- traditional transformer
- neural network
- transformer
- temperature
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01K—MEASURING TEMPERATURE; MEASURING QUANTITY OF HEAT; THERMALLY-SENSITIVE ELEMENTS NOT OTHERWISE PROVIDED FOR
- G01K13/00—Thermometers specially adapted for specific purposes
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
- Protection Of Transformers (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
A kind of traditional transformer state monitoring method based on probabilistic neural network, it is characterised in that traditional transformer state monitoring method based on probabilistic neural network is：Monitoring is acquired to three bus temperatures of traditional transformer temperature, traditional connector temperature of transformer secondary side outlet two, different traditional transformer inlet wires, analyzing and processing is carried out to temperature data and is used as input feature vector amount input probability neural network classification model；Probabilistic neural network disaggregated model is trained, learnt and classified by above-mentioned input feature vector amount, classification is identified to the state of temperature characteristic quantity of traditional transformer.The present invention can quickly, it is accurate, independently complete the identification to traditional transformer state and fault diagnosis, realize the diagnosis early to traditional mulual inductor malfunction and find failure, ensure the normal operation of traditional transformer, influence of traditional mulual inductor malfunction to power system can be avoided, reduces economic loss.
Description
Technical field
The invention belongs to POWER SYSTEM STATE to monitor field, is related to a kind of state monitoring method of power equipment, particularly
A kind of traditional transformer state monitoring method based on probabilistic neural network.
Background technology
Transformer is one of indispensable capital equipment such as power network normal operation, monitoring, metering, protection, control.It
Presence, high voltage, high current information is applied.Transformer running status is good and bad, directly influence power system measuring,
The accuracy of metering and relay protection, the reliability of automatics, safe operation of power system is influenceed very big.
At present, in power system, the use of traditional transformer is extensive, and for the state monitoring method of traditional transformer
Without research well and solve.Traditional transformer is in itself there is performance issue and failure problems, once traditional transformer is sent out
The abnormal safe operation that power system will be had a strong impact on failure of life running status.Therefore, it is necessary to which a kind of be based on probabilistic neural
Traditional transformer state monitoring method of network can be quickly and accurately mutual to tradition to traditional transformer status real time monitor
The failure of sensor is diagnosed, find traditional transformer early existing for failure, ensure the normal operation of traditional transformer.
The content of the invention
The present invention can not solve the problems, such as that above-mentioned traditional transformer is present for prior art, there is provided one kind is based on probability
Traditional transformer state monitoring method of neutral net.Traditional transformer state monitoring method based on probabilistic neural network passes through
Three traditional transformer temperature, traditional transformer secondary side outlet joint temperature, traditional transformer inlet wire bus temperatures are carried out
Monitoring, said temperature Analysis on monitoring data is handled as input feature vector amount input probability neural network classification model, to tradition
The state of transformer carries out Classification and Identification, so as to realize the diagnosis to traditional mulual inductor malfunction early.
In order to complete above-mentioned purpose, the present invention uses following technical scheme.
A kind of traditional transformer state monitoring method based on probabilistic neural network, it is characterised in that described to be based on probability
Traditional transformer state monitoring method of neutral net is：Traditional transformer temperature, traditional transformer secondary side outlet two are connect
Head temperature, three bus temperatures of different traditional transformer inlet wires are acquired monitoring, and analyzing and processing work is carried out to temperature data
For input feature vector amount input probability neural network classification model；Pass through the input feature vector amount of three bus temperature Data Analysis Services
As the fixed reference feature amount for judging Operation of Electric Systems situation, for excluding current power running situation to traditional transformer
The temperature influence factor of operation；Probabilistic neural network disaggregated model is trained, learnt and divided by above-mentioned input feature vector amount
Class, classification is identified to the state of temperature characteristic quantity of traditional transformer, realizes the diagnosis to traditional mulual inductor malfunction, completion pair
The status monitoring of traditional transformer.
A kind of traditional transformer state monitoring method based on probabilistic neural network, it is characterised in that described to be based on probability
Traditional transformer state monitoring method step of neutral net is as follows.
S01：Gather traditional transformer temperature T0, traditional connector temperature T1 of transformer secondary side outlet two and T2, not simultaneous interpretation
Three bus temperatures T3, T4 and T5 of system transformer inlet wire.
S02：Analyzing and processing is carried out to temperature T0, T1, T2, T3, T4 and T5 and is converted into input feature vector amount.
S03：Construct probabilistic neural network disaggregated model.Probabilistic neural network（PNN）Hierarchical model by input layer, pattern
Layer, summation layer and output layer form for 4 layers totally.Input layer number is equal with the dimension of input feature vector amount, mode layer nerve
The number of member is equal to the summation of various traditional transformer status categories number of training, summation layer neuron number and traditional mutual inductance
Device state classification situation number is identical, and output layer uses Bayes classifying rules, selects the classification of maximum a posteriori probability as output
Classification.An appropriate number of input feature vector amount data are chosen as training sample, Spread values are set by training sample input probability
Neural network classification model carries out network training to obtain the probabilistic neural network disaggregated model of traditional transformer state.Training
During, Spread values are tested, obtaining optimal value makes network output effect optimal.Probabilistic neural network disaggregated model point
Class output is classified according to traditional transformer state status and fault severity level.
S04：The input feature vector amount of traditional transformer state is input to the probabilistic neural network classification mould trained and constructed
Type, probabilistic neural network disaggregated model carry out classification output to input feature vector amount, judge the current operation shape of traditional transformer
State, running state recognition and fault diagnosis to traditional transformer are completed, find the failure of traditional transformer early.
The beneficial effects of the invention are as follows：A kind of traditional transformer status monitoring side based on probabilistic neural network of the present invention
Method, can quickly, it is accurate, independently complete the identification to traditional transformer state and fault diagnosis, realize former to traditional transformer
The diagnosis early of barrier and discovery failure, ensure the normal operation of traditional transformer, can avoid traditional mulual inductor malfunction to electric power
The influence of system, reduce economic loss.
Brief description of the drawings
Fig. 1 is traditional transformer state monitoring method flow chart of the invention based on probabilistic neural network.
Embodiment
The present invention is further illustrated with reference to the accompanying drawings and examples, to be specifically described the technical side of the present invention
Case.
A kind of traditional transformer state monitoring method based on probabilistic neural network, it is characterised in that described to be based on probability
Traditional transformer state monitoring method of neutral net is：Traditional transformer temperature, traditional transformer secondary side outlet two are connect
Head temperature, three bus temperatures of different traditional transformer inlet wires are acquired monitoring, and analyzing and processing work is carried out to temperature data
For input feature vector amount input probability neural network classification model；Pass through the input feature vector amount of three bus temperature Data Analysis Services
As the fixed reference feature amount for judging Operation of Electric Systems situation, for excluding current power running situation to traditional transformer
The temperature influence factor of operation；Probabilistic neural network disaggregated model is trained, learnt and divided by above-mentioned input feature vector amount
Class, classification is identified to the state of temperature characteristic quantity of traditional transformer, realizes the diagnosis to traditional mulual inductor malfunction, completion pair
The status monitoring of traditional transformer.
Embodiment process is as follows：As shown in figure 1, traditional transformer state monitoring method based on probabilistic neural network
Step is as follows.
S01：Gather traditional transformer temperature T0, traditional connector temperature T1 of transformer secondary side outlet two and T2, not simultaneous interpretation
Three bus temperatures T3, T4 and T5 of system transformer inlet wire.
S02：Analyzing and processing is carried out to temperature T0, T1, T2, T3, T4 and T5 and is converted into input feature vector amount.
S03：Construct probabilistic neural network disaggregated model.Probabilistic neural network（PNN）Hierarchical model by input layer, pattern
Layer, summation layer and output layer form for 4 layers totally.Input feature vector amount is by traditional transformer temperature T0, traditional transformer secondary side outlet
Two connector temperature T1 are with T2, Bu Tong three bus temperatures T3, T4 and T5 of traditional transformer inlet wire are formed, probabilistic neural network point
Class model classification output is divided into：Normally, minor failure, moderate failure, heavier failure, catastrophe failure, totally 5 kinds export situation.It is defeated
It is 6 that it is equal with the dimension of input feature vector amount, which to enter layer neuron number, and various traditional transformer status categories number of training are 30,
It is 1.5 to set Spread values, and the number of mode layer neuron is equal to the total of various traditional transformer status categories number of training
With for 150, summation layer neuron number and traditional transformer state classification situation number are all mutually 5, and output layer is using Bayes classification
Rule, select classification of the classification of maximum a posteriori probability as output.Using above-mentioned 150 groups of input feature vector amount data as training sample
This input probability neural network classification model carries out the training, study and classification output of network.
S04：The input feature vector amount of traditional transformer state is input to the probabilistic neural network classification mould trained and constructed
Type, probabilistic neural network disaggregated model carry out classification output to input feature vector amount, judge the current operation shape of traditional transformer
State, running state recognition and fault diagnosis to traditional transformer are completed, find the failure of traditional transformer early.Pass through 100
Classification output effect of the traditional transformer state input feature vector amount test sample of group to probabilistic neural network disaggregated model is tested
Card, classifying quality is accurate, error free.
Traditional transformer state monitoring method of the invention based on probabilistic neural network more than, it is quick, accurate, autonomous
The identification to traditional transformer state and fault diagnosis are completed, the diagnosis early to traditional mulual inductor malfunction is realized and finds event
Barrier, ensure the normal operation of traditional transformer, avoid influence of traditional mulual inductor malfunction to power system, reduce economic loss.
Claims (1)
1. a kind of traditional transformer state monitoring method based on probabilistic neural network, it is characterised in that described based on probability god
Traditional transformer state monitoring method through network is：To traditional transformer temperature, traditional joint of transformer secondary side outlet two
Temperature, three bus temperatures of different traditional transformer inlet wires are acquired monitoring, and analyzing and processing conduct is carried out to temperature data
Input feature vector amount input probability neural network classification model；Made by the input feature vector amount of three bus temperature Data Analysis Services
To judge the fixed reference feature amount of Operation of Electric Systems situation, traditional transformer is transported for excluding current power running situation
Capable temperature influence factor；Probabilistic neural network disaggregated model is trained, learnt and classified by above-mentioned input feature vector amount,
Classification is identified to the state of temperature characteristic quantity of traditional transformer, realizes the diagnosis to traditional mulual inductor malfunction, is completed to passing
The status monitoring of system transformer；
Traditional transformer state monitoring method step based on probabilistic neural network is as follows：
S01：Gather traditional transformer temperature T0, traditional connector temperature T1 of transformer secondary side outlet two with T2, Bu Tong tradition mutually
Three bus temperatures T3, T4 and T5 of sensor inlet wire；
S02：Analyzing and processing is carried out to temperature T0, T1, T2, T3, T4 and T5 and is converted into input feature vector amount；
S03：Construct probabilistic neural network disaggregated model；Probabilistic neural network（PNN）Hierarchical model by input layer, mode layer,
Summation layer and output layer form for 4 layers totally；Input layer number is equal with the dimension of input feature vector amount, mode layer neuron
Number be equal to the summations of various traditional transformer status categories number of training, summation layer neuron number and traditional transformer
State classification situation number is identical, and output layer uses Bayes classifying rules, selects class of the classification of maximum a posteriori probability as output
Not；Input feature vector amount by traditional transformer temperature T0, traditional connector temperature T1 of transformer secondary side outlet two with T2, it is Bu Tong traditional
Three bus temperatures T3, T4 and T5 composition of transformer inlet wire, the classification output of probabilistic neural network disaggregated model are divided into：Normally,
Minor failure, moderate failure, heavier failure, catastrophe failure, totally 5 kinds export situation；Choose an appropriate number of input feature vector amount number
According to as training sample, set Spread values that training sample input probability neural network classification model is carried out into network training to obtain
Obtain the probabilistic neural network disaggregated model of traditional transformer state；In the training process, Spread values are tested, obtained most
The figure of merit makes network output effect optimal；The classification output of probabilistic neural network disaggregated model is according to traditional transformer state status and event
The barrier order of severity is classified；
S04：The input feature vector amount of traditional transformer state is input to the probabilistic neural network disaggregated model trained and constructed, generally
Rate neural network classification model carries out classification output to input feature vector amount, judges the current running status of traditional transformer, complete
The running state recognition and fault diagnosis of paired traditional transformer, the failure of traditional transformer is found early.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610020821.0A CN105701543B (en) | 2016-01-13 | 2016-01-13 | A kind of traditional transformer state monitoring method based on probabilistic neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610020821.0A CN105701543B (en) | 2016-01-13 | 2016-01-13 | A kind of traditional transformer state monitoring method based on probabilistic neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105701543A CN105701543A (en) | 2016-06-22 |
| CN105701543B true CN105701543B (en) | 2018-03-02 |
Family
ID=56226371
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610020821.0A Expired - Fee Related CN105701543B (en) | 2016-01-13 | 2016-01-13 | A kind of traditional transformer state monitoring method based on probabilistic neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105701543B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106251059B (en) * | 2016-07-27 | 2020-02-07 | 中国电力科学研究院 | Cable state evaluation method based on probabilistic neural network algorithm |
| CN106124949B (en) * | 2016-08-30 | 2019-08-13 | 国网山东省电力公司济南供电公司 | One kind is based on thermal infrared imaging technology to insulator breakdown on-line monitoring method |
| CN106408687B (en) * | 2016-11-24 | 2019-04-05 | 沈阳航空航天大学 | A kind of automobile engine fault early warning method based on machine learning method |
| CN112580700B (en) * | 2020-12-04 | 2021-07-30 | 杭州佳速度产业互联网有限公司 | Data correction method, system and storage medium of electric power Internet of things meter |
| CN113420813B (en) * | 2021-06-23 | 2023-11-28 | 北京市机械工业局技术开发研究所 | Diagnostic method for particulate matter filter cotton state of vehicle tail gas detection equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1805234A (en) * | 2005-12-21 | 2006-07-19 | 长沙理工大学 | Pattern matching based security protection method for relay protection information of electric system in network environment |
| CN204241588U (en) * | 2014-11-11 | 2015-04-01 | 国家电网公司 | Current transformer operating state monitoring system |
-
2016
- 2016-01-13 CN CN201610020821.0A patent/CN105701543B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1805234A (en) * | 2005-12-21 | 2006-07-19 | 长沙理工大学 | Pattern matching based security protection method for relay protection information of electric system in network environment |
| CN204241588U (en) * | 2014-11-11 | 2015-04-01 | 国家电网公司 | Current transformer operating state monitoring system |
Non-Patent Citations (3)
| Title |
|---|
| 《Improved Structure of PNN Using PCA in Transformer Fault》;Hossein Paydarnia,et al.;《Arabian Journal for Science and Engineering》;20140630;全文 * |
| 基于概率神经网络的电力变压器故障诊断;曹永刚等;《继电器》;20060201;第34卷(第3期);全文 * |
| 基于神经网络的配电线路综合故障定位方法;严凤等;《电力系统及其自动化学报》;20150531;第27卷(第5期);正文第3节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105701543A (en) | 2016-06-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105701543B (en) | A kind of traditional transformer state monitoring method based on probabilistic neural network | |
| Afrasiabi et al. | Integration of accelerated deep neural network into power transformer differential protection | |
| US11853899B2 (en) | Methods and apparatus for data analysis | |
| Nurmanova et al. | A new transformer FRA measurement technique to reach smart interpretation for inter-disk faults | |
| Bigdeli et al. | Transformer winding faults classification based on transfer function analysis by support vector machine | |
| CN102340811B (en) | Method for carrying out fault diagnosis on wireless sensor networks | |
| CN110297141A (en) | Fault Locating Method and system based on multilayer assessment models | |
| CN104052612B (en) | A kind of Fault Identification of telecommunication service and the method and system of positioning | |
| CN109670242B (en) | Transformer winding deformation unsupervised online monitoring method based on elliptical envelope curve | |
| US20050278597A1 (en) | Methods and apparatus for data analysis | |
| US20100088054A1 (en) | Methods and apparatus for data analysis | |
| CN103838229B (en) | Diagnosis method and device of electric car | |
| CN113139335B (en) | Intelligent rail circuit fault diagnosis method based on BP neural network | |
| CN110118928B (en) | Breaker fault diagnosis method based on error inverse propagation algorithm | |
| WO2012094156A2 (en) | Methods and apparatus for data analysis | |
| Chang et al. | Wire mismatch detection using a convolutional neural network and fault localization based on time–frequency-domain reflectometry | |
| CN108304567A (en) | High-tension transformer regime mode identifies and data classification method and system | |
| CN116893336B (en) | High-low voltage switching device and method thereof | |
| CN111062569A (en) | Low-current fault discrimination method based on BP neural network | |
| CN102426671B (en) | Optimal troubleshooting method based on comprehensive cost | |
| CN102306244B (en) | Fault eliminating method based on evaluation of detecting points | |
| CN113109666A (en) | Track circuit fault diagnosis method based on deep convolutional neural network | |
| CN106250937B (en) | A kind of failure modes diagnostic method based on non-index of similarity | |
| CN117491813A (en) | Insulation abnormality detection method for power battery system of new energy automobile | |
| CN103235206A (en) | Transformer fault diagnosis method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180302 Termination date: 20200113 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |