CN108287327A - Metering automation terminal fault diagnostic method based on Bayes's classification - Google Patents
Metering automation terminal fault diagnostic method based on Bayes's classification Download PDFInfo
- Publication number
- CN108287327A CN108287327A CN201711329487.8A CN201711329487A CN108287327A CN 108287327 A CN108287327 A CN 108287327A CN 201711329487 A CN201711329487 A CN 201711329487A CN 108287327 A CN108287327 A CN 108287327A
- Authority
- CN
- China
- Prior art keywords
- metering automation
- automation terminal
- fault
- data
- historical data
- 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
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R35/00—Testing or calibrating of apparatus covered by the other groups of this subclass
- G01R35/02—Testing or calibrating of apparatus covered by the other groups of this subclass of auxiliary devices, e.g. of instrument transformers according to prescribed transformation ratio, phase angle, or wattage rating
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Probability & Statistics with Applications (AREA)
- Power Engineering (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention discloses a kind of metering automation terminal fault diagnostic method based on Bayes's classification, includes the following steps：The historical data of metering automation terminal is collected, and marks the fault diagnosis example historical data after on-site verification；For the fault diagnosis example historical data determination failure modes range to be diagnosed being marked；Bayes classifier is trained by metering automation terminal historical data and fault diagnosis example historical data；Fault diagnosis is carried out to metering automation terminal real time data by the Bayes classifier obtained after training.The present invention as sample set using a large amount of metering automation terminal historical datas and fault diagnosis example historical data by training Bayes classifier, and Bayesian inference is carried out to the real time data that metering automation terminal collects using Bayes classifier, the probability for fast and effeciently obtaining the generation of metering automation terminal fault, realizes the quick diagnosis of failure.
Description
Technical field
The invention belongs to metering device fault diagnosis fields, and specifically the metering automation based on Bayes's classification is whole
Hold method for diagnosing faults.
Background technology
With the construction of metering automation system, system integration, complication and the degree of automation increasingly improve, and metering is certainly
The influence of dynamicization terminal-pair power grid electrical energy measurement stable operation is also more and more important, therefore carries out failure to metering automation terminal
The method for diagnosing faults of diagnostic analysis assessment also becomes more and more important, especially more next in current metering automation terminal quantity
In the case that more, equipment mounting condition becomes increasingly complex, metering automation terminal fault probability of happening is got higher and difficult judgment,
Manpower and materials have been difficult to maintain the on-site maintenance of all terminals.Therefore realize that the failure of quickly and effectively metering automation terminal is examined
It is disconnected, it is beneficial to quickly eliminate fault harm, saves O＆M cost, realize the safe and stable operation of metering automation terminal.
The method of metering automation fault diagnosis generally mainly has three classes：Numerical statistic method, expert experience base method and
Mathematical modelling algorithms.
Numerical statistic method is mainly to measure the history electric energy data that automatization terminal collects by analysis, is obtained
The law characteristic of historical data, and choosing, there is the law characteristic for representing meaning to be diagnosed as fault eigenvalue.This method
Fault signature can be analyzed, by aspect ratio to just realizing which kind of failure is rapidly diagnosed to be is.But Feature Selection rule is difficult
With covering, institute is faulty.Expert experience base method be self study, strong antijamming capability are carried out by certain rule, but input with
Without determining relational expression between output, and diagnosis is required for enough data adaptively to be trained every time, if not receiving
It holds back, is then easily absorbed in the predicament of local extremum.Mathematical modelling algorithms are the mathematics by establishing multiple input output and Trouble Match
Model, and new input value is solved to obtain corresponding failure output.This method in system fault diagnosis also extensively
It is applied, but due to the complexity of Operation of Electric Systems, is typically only capable to obtain approximate mathematical model, error is larger.
Described in above-mentioned 2 points, as the increase of metering automation terminal installation number and installation environment are more and more multiple
It is miscellaneous, therefore the fault category of metering automation terminal and feature are also more diversified, therefore pass through expert experience base and data mould
The difficulty of type algorithm discriminating fault types is larger, and Rulemaking is difficult to adapt to all metering automation terminals.
Invention content
The invention discloses a kind of metering automation terminal fault diagnostic method based on Bayes's classification, can pass through system
Meter analyzes a large amount of metering automation terminal historical data, and has already passed through the true fault example that on-site verification obtains, and comes
It realizes the training of Bayes classifier, and then can metering automation terminal fault fast and effeciently be obtained by Bayesian inference
The probability of generation realizes the quick diagnosis of failure.
To achieve the above object, the technical scheme is that：
Metering automation terminal fault diagnostic method based on Bayes's classification, includes the following steps：
S1. the historical data of metering automation terminal is collected, and marks the fault diagnosis example history number after on-site verification
According to；
S2. it is directed to the fault diagnosis example historical data determination failure modes range to be diagnosed being marked；
S3. Bayes classifier is trained by metering automation terminal historical data and fault diagnosis example historical data；
S4. fault diagnosis is carried out to metering automation terminal real time data by the Bayes classifier obtained after training.
Further, further include step S5：S5. metering automation terminal is checked at operation maintenance personnel scene, if the diagnosis metering is certainly
Dynamicization terminal real time data is true failure, then the metering automation terminal real time data is made corresponding fault type mark
Note, and it is included into fault diagnosis example historical data, Bayes classifier is then trained according to updated fault diagnosis example historical data；
Step S5 and S4 are repeated, realizes the diagnosis of the metering automation terminal fault based on Bayes's classification.
Further, in step S1, the metering automation terminal historical data is acquired back by metering automation terminal
Come, and the history gathered data being stored in metering automation main station system database；The metering automation terminal includes
Plant stand, distribution transforming, negative control, centralized meter reading terminal；
The metering automation historical data includes multinomial different characteristic；The fault diagnosis example historical data is the meter
A part for automatization terminal historical data is measured, also includes multinomial characteristic；The fault diagnosis example historical data all includes
Fault type marks.
Further, the step S3 is specially：
S3-1. a kind of fault type in failure modes range described in step S2 is calculated in entire metering automation terminal history
The probability occurred in data calculates the prior probability of Bayes classifier；
S3-2. the different characteristic data that fault type includes described in step S3-1 are calculated in the fault diagnosis example historical data to go out
Existing probability calculates the conditional probability of Bayes classifier；
S3-3. step S3-1 and step S3-2 is repeated to calculate all fault categories in failure modes range described in step S2
Prior probability and conditional probability.Further, the step S4 is specially：
S4-1., metering automation terminal real time data is input to the Bayes classifier after training；
S4-2. general by the priori being calculated in characteristic in metering automation terminal real time data and step S3
Rate and conditional probability obtain metering automation terminal real time data described in step S4-1 using bayesian algorithm and correspond to generation difference
The probability of fault category；
The metering automation terminal real time data that step S4-2 is calculated is corresponded to the probability for different faults classification occur by S4-3,
It arranges from big to small, by live operation maintenance personnel at the scene to thering is the metering automation terminal of maximum probability failure is practical to be sentenced
It is disconnected, fault diagnosis is made to metering automation terminal according to practical judging result.
Further, the step S5 is specially：
S5-1. metering automation terminal is checked at operation maintenance personnel scene, if it is true to diagnose the metering automation terminal real time data
Real failure, metering automation terminal real time data will be all included into metering automation terminal historical data；
S5-2. obtained fault type is actually judged according to scene, corresponding metering automation terminal is real when breaking down to this
When data carry out fault type label, and be included into fault diagnosis example historical data；
S5-3. Bayes classifier is trained according to updated fault diagnosis example historical data.
Further, the specific calculating process of prior probability for calculating Bayes classifier is：
According to step S2, different metering automation terminal faults is denoted as F={ f1,f2,…,fK, K is to measure certainly
Dynamicization terminal fault species number, while remembering that metering automation terminal normal condition is f0；
According to step S3, the historical data and fault diagnosis example historical data of metering automation terminal, as training sample
Data are trained Bayes classifier, are equipped with training data the C={ (x that quantity is N1,y1),(x2,y2),…,(xN,
yN), wherein xiFor i-th of sample data, y in training dataiFor sample data xiCorresponding data classification marker, i.e. y ∈ { f0,
f1,f2,…,fk, It is j-th of feature of i-th of sample, Wherein ajlIt is first of value of j-th of feature, SjFor jth
The quantity of all values of a feature, n are the total characteristic number of each sample；
It is rightMake normalized；
If per one kind fault category fkPrior probability be P (Y=fk), k=1,2,3 ..., K, prior probability is：
Y is the stochastic variable in data classification marker in formula；I is indicator function, i.e. yi=fkWhen be 1, be otherwise 0.
Further, the specific calculating process of conditional probability for calculating Bayes classifier is：
J=1,2 ..., n；L=1,2 ..., Sj；K=1,2 ..., K
X is a stochastic variable in sample data, X in formula(j)For j-th of feature of the stochastic variable；
Obtaining conditional probability according to Bayes' theorem is：
Further, it calculates real time data using bayesian algorithm and the probability detailed processes of different faults occurs and be：
Assuming that the real time data that metering automation terminal collects is D, D=(d(1),d(2),…,d(n))T, according to the following formula：
It obtains real time data D and f occurskThe probability of class failure.
There is certain a kind of failure f in each real time data D of metering automation terminal being calculated by Bayes classifierk's
Probability P (Y=fk| X=D), by this probability by arranging from big to small, scene is gone to by live operation maintenance personnel and is occurred to maximum probability
The metering automation terminal of failure carries out practical judgement, does to be out of order to metering automation terminal according to practical judging result and examine
It is disconnected.
The above-described metering automation terminal fault diagnostic method based on Bayes's classification, by under true environment
Failure in obtained metering automation terminal historical data, and metering automation terminal historical data after being verified
Instance data, to train Bayes classifier, Bayes classifier uses Bayesian inference principle, whether to judge real time data
Including the probability of fault signature data and generation different faults, to help live operation maintenance personnel quickly to investigate metering automation
The failure of terminal realizes the quick diagnosis of failure.And Bayes classifier can continue update training according to real time data,
The accuracy for further improving fault diagnosis can be quickly accurate by the raising of the rapidity and accuracy of fault diagnosis
Really for metering automation terminal fault diagnosis support is provided, be beneficial to quickly eliminate fault harm, saving power grid O＆M at
This, realizes the safe and stable operation of metering automation terminal.
Description of the drawings
Fig. 1 is flow chart of the method for the present invention.
Fig. 2 is to calculate real time data using bayesian algorithm the detail flowchart of the probability of different faults occur.
Specific implementation mode
Below with reference to specific embodiment, the invention will be further described, but protection scope of the present invention be not limited to it is following
Embodiment.
Metering automation terminal fault diagnostic method based on Bayes's classification, as shown in Figure 1, including the following steps：
S1. the historical data of metering automation terminal is collected, and marks the fault diagnosis example history number after on-site verification
According to；
The metering automation terminal historical data is to be returned by the acquisition of metering automation terminal, and be stored in metering automation
History gathered data in main station system database；Metering automation terminal includes plant stand, distribution transforming, negative control, centralized meter reading terminal.Metering
Difference of the historical data according to metering automation terminal kinds is automated, including multinomial different characteristic.
And fault diagnosis example historical data is a part for metering automation terminal historical data, also includes multinomial characteristic
According to.Simultaneous faults instance histories data are by the corresponding data of autoptic metering automation terminal fault.
Also, the historical data of the metering automation terminal all includes fault category mark.
S2. it is directed to the fault diagnosis example historical data determination failure modes range to be diagnosed being marked；
According to nearly 3 years metering automations terminal fault example, failure can be divided into following a few classes：(1) voltage decompression；(2) voltage
It is uneven；(3) terminal time overproof mistake；(4) current imbalance；(5) deformity thinks that imperfect and presence acquires successfully under terminal
Data item；(6) the ammeter time is overproof；(7) measurement point data item is imperfect and there is the successful data item of acquisition；(8) it measures
Point data item is completed and data point is discontinuous in time；(9) measurement point does not acquire；(10) harmonic wave Threshold Crossing Alert；(11) terminal
Day operation time anomaly；(12) electric voltage over press；(13) voltage error；(14) electric flux is overproof；(15) overcurrent；(16) electric
The disconnected phase of stream；(17) current error；(18) voltage circuit is abnormal；(19) indication declines；(20) electric energy meter flies away；(21) electric energy meter stops
It walks；(22) power data mistake；(23) electricity indication mistake；(24) current loop is abnormal；(25) power factor mistake；(26)
The out-of-limit record of DC analogue quantity；(27) active total electric flux is differential out-of-limit；(28) the out-of-limit record of apparent energy；(29) voltage, electricity
Flow reverse sequence；(30) terminal fault；(31) main station system malfunctions；(32) channel is obstructed.
S3. it combines shown in Fig. 2, pattra leaves is trained by the historical data and fault diagnosis example historical data of metering automation terminal
This grader.
The above 32 class metering automation terminal fault is denoted as F={ f1,f2,…,fK, K=32；This failure modes can root
Increased and decreased individually according to the fault diagnosis example classification in fault diagnosis example historical data, while remembering that metering automation terminal normal condition is
f0。
Historical data for metering automation terminal and fault diagnosis example historical data, will come pair as training sample data
Bayes classifier is trained, and is equipped with training data the C={ (x that quantity is N1,y1),(x2,y2),…,(xN,yN), wherein
xiFor i-th of sample data, y in training dataiFor sample data xiCorresponding data classification marker, i.e. y ∈ { f0,f1,f2,…,
fk}.Further have It is j-th of feature of i-th of sample, Wherein ajlIt is first of value of j-th of feature, SjFor all values of j-th feature
Quantity, n are the total characteristic number of each sample.
Because metering automation terminal-pair answers different electric energy metering devices, collected data correspond to different specified
Value, such as voltage data rated value are divided into 100V and 57.7V according to line connection, it is therefore desirable to metering automation historical data
Certain features do normalized.Normalized is done by taking voltage value as an example, its step are as follows：
According to above-mentioned setting, if A phase voltage values areI.e. A phase voltages are the 1st feature values of i-th of sample, according to meter
Automatization terminal archives are measured, if rated voltage is 57.7V,If rated voltage is
100V, then
Normalized is directed to the data characteristics there are many rated value, does not refer to voltage data singly.To metering automation terminal
After historical data is normalized, starts to calculate prior probability and conditional probability, that is, calculate in fault diagnosis example historical data
Different faults type, in the probability of entire metering automation historical data training sample.
If per one kind fault category fkPrior probability be P (Y=fk), k=1,2,3 ..., K, value fkClass failure
The total number of samples N of sample number divided by training set, that is, metering automation terminal historical data.I.e.
Y is the stochastic variable in data classification marker in formula；I is indicator function, i.e. yi=fkWhen be 1, be otherwise 0；
For a given fault category fk, calculate its conditional probability；Calculate corresponding sample number when such failure occurs
According to feature value, it is all occur such failure sample datas in probability, calculate it is as follows：
J=1,2 ..., n；L=1,2 ..., Sj；K=1,2 ..., K
X is a stochastic variable in sample data, X in formula(j)For j-th of feature of the stochastic variable.
It can be obtained according to Bayes' theorem：
P (Y=fk| X=x) be posterior probability, that is, it corresponds to a certain sample data x and f occurskThe probability of class failure.Pass through above-mentioned step
Suddenly any metrology automatization terminal data X can be calculated, obtains the probability P that the generation of certain class failure occurs.If taking probability
Maximum classification fkIt is marked for the fault category of the data, that is, completes the Bayes's classification to data X.
S4. fault diagnosis is carried out to metering automation terminal real time data by the Bayes classifier obtained after training.
S4-1., metering automation terminal real time data is input to the Bayes classifier after training；
S4-2. there is the probability of different faults using bayesian algorithm calculating real time data in Bayes classifier；
Real time data D=(the d collected for metering automation(1),d(2),…,d(n))T, then can calculate
It obtains real time data D and f occurskThe probability of class failure.
S4-3. each real time data D of metering automation terminal that Bayes classifier judges has been calculated in basis
There is certain a kind of failure fkProbability P (Y=fk| X=D), by this probability by arranging from big to small, gone to by live operation maintenance personnel
Scene carries out practical judgement to the metering automation terminal that maximum probability breaks down, according to practical judging result to metering automation
Terminal makes fault diagnosis.
Further, further include step S5：If S5. metering automation terminal is checked at operation maintenance personnel scene, if described in diagnosis
Metering automation terminal real time data is true failure, then the metering automation terminal real time data is made corresponding failure
Type mark, and it is included into fault diagnosis example historical data, Bayes point is then trained according to updated fault diagnosis example historical data
Class device；
If S5-1. live operation maintenance personnel, which is checked, judges the metering automation terminal real time data for true failure, metering is certainly
Dynamicization terminal real time data will be all included into metering automation terminal historical data；
S5-2. obtained fault type is actually judged according to scene, corresponding data D carries out f when breaking down to thisiClass corresponds to
Data classification marker, and be included into fiThe fault diagnosis example historical data sample of class failure is concentrated；
S5-3. Bayes classifier is trained according to updated fault diagnosis example historical data.
Step S5 and S4 are repeated, realizes the diagnosis of the metering automation terminal fault based on Bayes's classification.
The present embodiment uses Bayesian inference principle, according to every historical data of metering automation terminal, and passes through
The true fault example that on-site verification obtains, it is real to establish fault diagnosis model to realize the training of Bayes classifier
It is uncertain and in the case of lacking Given information to solve situation at the scene for the diagnosis of existing metering automation terminal real-time running state
Metering automation terminal fault diagnosis problem.
It is further below that voltage loses to have collected the historical data of 96 groups of quantitative measure automatization terminals, wherein have 30 groups
For pressing fault diagnosis example historical data, the furtherly calculating process of the present embodiment：
In specific steps S1, the historical data for collecting metering automation terminal includes three-phase voltage Ua、Ub、Uc, data classification marker
Y.In practical applications, the historical data of metering automation terminal and it is not specific to three-phase voltage, can increased according to real data situation
Add data characteristics.
Faulty instance histories data are collected simultaneously, fault diagnosis example historical data is metering automation terminal historical data
A part, fault diagnosis example historical data are same as above comprising data item, and data classification marker is that corresponding failure modes identify.
The present embodiment has collected the historical data of 96 groups of quantitative measure automatization terminals, wherein has 30 groups to be the event of voltage decompression
Hinder instance histories data, the historical data of metering automation terminal is shown in Table 1：
Table 1
Specifically in step s3, pattra leaves is trained by the historical data and fault diagnosis example historical data of 1 metering automation terminal of table
This grader.
32 class metering automation terminal faults are denoted as F={ f1,f2,…,fK, K=32；This failure modes can be according to reality
Border situation increase and decrease, while remembering that metering automation terminal normal condition is f0, voltage decompression is i.e. labeled as f1。
For the historical data of above-mentioned 96 Sets of Measurement automatization terminal, 30 groups of fault diagnosis example historical datas wherein included,
Bayes classifier is trained as training sample data, if xiFor i-th of sample data, y in training dataiFor sample
Notebook data xiCorresponding data classification marker, i.e. y ∈ { f0,f1,f2,…,fk}.In addition have
It is j-th of feature of i-th of sample, feature corresponds to data item described in S2, i.e. n=3 in the present embodiment,
WhereinWherein ajlIt is the l of j-th of feature
A value, SjFor the quantity of all values of j-th of feature.
Because metering automation terminal-pair answers different electric energy metering devices, collected data correspond to different specified
Value, such as voltage data rated value are divided into 100V and 57.7V according to line connection, it is therefore desirable to metering automation historical data
Certain features do normalized.Normalized is done by taking voltage value as an example, its step are as follows：
According to above-mentioned setting, A phase voltage values areI.e. A phase voltages are the 1st feature values of i-th of sample, will be according to meter
Automatization terminal archives are measured, if rated voltage is 57.7V,
The present embodiment is 57.7V according to line connection, by the historical data of above-mentioned 96 Sets of Measurement automatization terminal into
Row normalized obtains table 2：
Table 2
After metering automation terminal historical data is normalized, starts to calculate prior probability and conditional probability, that is, count
Different faults type in fault diagnosis example historical data is calculated, in the probability of entire metering automation historical data training sample.
In the present embodiment, if fault category f1Prior probability be P (Y=f1), value f1The sample number of class failure divided by
The total number of samples N of training set, that is, metering automation terminal historical data.I.e.
In addition it is f to have data classification marker0Prior probability be
Y is the stochastic variable in data classification marker in formula；I is indicator function, i.e., in the calculating of this group, yi=f1When be 1, it is no
It is then 0；For a given fault category f1, calculate its conditional probability；Calculate corresponding sample when such failure occurs
The feature value of data, the probability in all sample datas that such failure occurs, calculates as follows：
X is a stochastic variable in sample data, X in formula(1)For the 1st feature U of the stochastic variablea, X(2)It is random for this
2nd feature U of variableb, X(3)For the 3rd feature U of the stochastic variablec。
Specifically in step s 4, by the Bayes classifier that is obtained after training to metering automation terminal real time data into
Row fault diagnosis.
S4-1. by metering automation terminal real time data D=(0.1,1.0,1.0)TIt is input to the Bayes point after training
Class device；
S4-2. there is the probability of different faults using bayesian algorithm calculating real time data in Bayes classifier；
Real time data D=(the d collected for metering automation(1),d(2),…,d(n))T, then can calculate
From the above equation, we can see that its denominator for all failure modes all, then can only calculate its molecule, i.e.,
Real time data D is calculated according to above formula, f occurs1Class failure, the i.e. probability of voltage no-voltage fault.
Real time data D generations f can similarly be calculated separately to obtain2–f32The probability of class failure.
By the generation f of acquisition1–f32Probability by arranging from big to small, if the maximum probability of voltage no-voltage fault, then by existing
Field operation maintenance personnel goes to scene and first judges whether metering automation terminal occurs voltage no-voltage fault really, if the event of voltage decompression
Data D is then carried out f by barrier1Class fault type marks, and is included into f1Fault diagnosis example historical data in the sample set of class failure, if
Voltage no-voltage fault, then select probability is only second to the failure mode of voltage no-voltage fault and is investigated, and repeats to judge,
The type until investigation is out of order, fault diagnosis is made according to practical judging result to metering automation terminal.
Bayes classifier is trained according to updated fault diagnosis example historical data.It then repeats and realizes based on Bayes point
The diagnosis of the metering automation terminal fault of class.
Claims (6)
1. the metering automation terminal fault diagnostic method based on Bayes's classification, it is characterised in that include the following steps：
S1. the historical data of metering automation terminal is collected, and marks the fault diagnosis example history number after on-site verification
According to；
S2. it is directed to the fault diagnosis example historical data determination failure modes range to be diagnosed being marked；
S3. Bayes classifier is trained by metering automation terminal historical data and fault diagnosis example historical data；
S4. fault diagnosis is carried out to metering automation terminal real time data by the Bayes classifier obtained after training.
2. the metering automation terminal fault diagnostic method according to claim 1 based on Bayes's classification, feature exist
In：
Further include step S5：S5. metering automation terminal is checked at operation maintenance personnel scene, if it is real to diagnose the metering automation terminal
When data be true failure, then the metering automation terminal real time data is made corresponding fault type marks, and is included into
Then fault diagnosis example historical data trains Bayes classifier according to updated fault diagnosis example historical data；
Step S5 and S4 are repeated, realizes the diagnosis of the metering automation terminal fault based on Bayes's classification.
3. the metering automation terminal fault diagnostic method according to claim 1 based on Bayes's classification, feature exist
In：
In step S1, the metering automation terminal historical data is to be returned by the acquisition of metering automation terminal, and be stored in
History gathered data in metering automation main station system database；The metering automation terminal includes plant stand, distribution transforming, bears
Control, centralized meter reading terminal；
The metering automation historical data includes multinomial different characteristic；The fault diagnosis example historical data is the meter
A part for automatization terminal historical data is measured, also includes multinomial characteristic；The historical data of the metering automation terminal
It all include fault category mark.
4. the metering automation terminal fault diagnostic method according to claim 1 based on Bayes's classification, feature exist
In：
The step S3 is specially：
S3-1. a kind of fault type in failure modes range described in step S2 is calculated in entire metering automation terminal history
The probability occurred in data calculates the prior probability of Bayes classifier；
S3-2. the different characteristic data that fault type includes described in step S3-1 are calculated in the fault diagnosis example historical data to go out
Existing probability calculates the conditional probability of Bayes classifier；
S3-3. step S3-1 and step S3-2 is repeated to calculate all fault categories in failure modes range described in step S2
Prior probability and conditional probability.
5. the metering automation terminal fault diagnostic method according to claim 4 based on Bayes's classification, feature exist
In：
The step S4 is specially：
S4-1., metering automation terminal real time data is input to the Bayes classifier after training；
S4-2. general by the priori being calculated in characteristic in metering automation terminal real time data and step S3
Rate and conditional probability obtain metering automation terminal real time data described in step S4-1 using bayesian algorithm and correspond to generation difference
The probability of fault category；
The metering automation terminal real time data that step S4-2 is calculated is corresponded to the probability for different faults classification occur by S4-3,
It arranges from big to small, by live operation maintenance personnel at the scene to thering is the metering automation terminal of maximum probability failure is practical to be sentenced
It is disconnected, fault diagnosis is made to metering automation terminal according to practical judging result.
6. the metering automation terminal fault diagnostic method according to claim 2 based on Bayes's classification, feature exist
In：
The step S5 is specially：
S5-1. metering automation terminal is checked at operation maintenance personnel scene, if it is true to diagnose the metering automation terminal real time data
Real failure, metering automation terminal real time data will be all included into metering automation terminal historical data；
S5-2. obtained fault type is actually judged according to scene, corresponding metering automation terminal is real when breaking down to this
When data carry out fault type label, and be included into fault diagnosis example historical data；
S5-3. Bayes classifier is trained according to updated fault diagnosis example historical data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711329487.8A CN108287327A (en) | 2017-12-13 | 2017-12-13 | Metering automation terminal fault diagnostic method based on Bayes's classification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711329487.8A CN108287327A (en) | 2017-12-13 | 2017-12-13 | Metering automation terminal fault diagnostic method based on Bayes's classification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108287327A true CN108287327A (en) | 2018-07-17 |
Family
ID=62832000
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711329487.8A Pending CN108287327A (en) | 2017-12-13 | 2017-12-13 | Metering automation terminal fault diagnostic method based on Bayes's classification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108287327A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086889A (en) * | 2018-09-30 | 2018-12-25 | 广东电网有限责任公司 | Terminal fault diagnostic method neural network based, device and system |
| CN109145977A (en) * | 2018-08-15 | 2019-01-04 | 河海大学常州校区 | A kind of bone damage type identification method based on naive Bayesian |
| CN109409444A (en) * | 2018-12-26 | 2019-03-01 | 国网陕西省电力公司电力科学研究院 | A kind of method of discrimination of the polynary electric network fault type based on prior probability |
| CN111401431A (en) * | 2020-03-12 | 2020-07-10 | 成都小步创想慧联科技有限公司 | Group renting house identification method and system and storage medium |
| CN111666978A (en) * | 2020-05-11 | 2020-09-15 | 深圳供电局有限公司 | Intelligent fault early warning system for IT system operation and maintenance big data |
| CN113033642A (en) * | 2021-03-17 | 2021-06-25 | 广东电网有限责任公司计量中心 | Intelligent electric energy meter state judgment method and system based on alarm event |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101917297A (en) * | 2010-08-30 | 2010-12-15 | 烽火通信科技股份有限公司 | Method and system for diagnosing faults of core network based on Bayesian network |
| CN103454516A (en) * | 2013-05-06 | 2013-12-18 | 国家电网公司 | Intelligent transformer substation secondary equipment health state diagnosis method |
| CN103699698A (en) * | 2014-01-16 | 2014-04-02 | 北京泰乐德信息技术有限公司 | Method and system for track traffic failure recognition based on improved Bayesian algorithm |
| CN103729804A (en) * | 2014-01-02 | 2014-04-16 | 东南大学 | On-line decision support method for responding to power quality early warning |
| CN104504525A (en) * | 2014-12-26 | 2015-04-08 | 国家电网公司 | Method for realizing power-grid equipment failure prewarning through big data mining technology |
-
2017
- 2017-12-13 CN CN201711329487.8A patent/CN108287327A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101917297A (en) * | 2010-08-30 | 2010-12-15 | 烽火通信科技股份有限公司 | Method and system for diagnosing faults of core network based on Bayesian network |
| CN103454516A (en) * | 2013-05-06 | 2013-12-18 | 国家电网公司 | Intelligent transformer substation secondary equipment health state diagnosis method |
| CN103729804A (en) * | 2014-01-02 | 2014-04-16 | 东南大学 | On-line decision support method for responding to power quality early warning |
| CN103699698A (en) * | 2014-01-16 | 2014-04-02 | 北京泰乐德信息技术有限公司 | Method and system for track traffic failure recognition based on improved Bayesian algorithm |
| CN104504525A (en) * | 2014-12-26 | 2015-04-08 | 国家电网公司 | Method for realizing power-grid equipment failure prewarning through big data mining technology |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109145977A (en) * | 2018-08-15 | 2019-01-04 | 河海大学常州校区 | A kind of bone damage type identification method based on naive Bayesian |
| CN109145977B (en) * | 2018-08-15 | 2021-12-10 | 河海大学常州校区 | Bone damage type discrimination method based on naive Bayes |
| CN109086889A (en) * | 2018-09-30 | 2018-12-25 | 广东电网有限责任公司 | Terminal fault diagnostic method neural network based, device and system |
| CN109086889B (en) * | 2018-09-30 | 2021-05-11 | 广东电网有限责任公司 | Terminal fault diagnosis method, device and system based on neural network |
| CN109409444A (en) * | 2018-12-26 | 2019-03-01 | 国网陕西省电力公司电力科学研究院 | A kind of method of discrimination of the polynary electric network fault type based on prior probability |
| CN109409444B (en) * | 2018-12-26 | 2020-10-23 | 国网陕西省电力公司电力科学研究院 | Multivariate power grid fault type discrimination method based on prior probability |
| CN111401431A (en) * | 2020-03-12 | 2020-07-10 | 成都小步创想慧联科技有限公司 | Group renting house identification method and system and storage medium |
| CN111666978A (en) * | 2020-05-11 | 2020-09-15 | 深圳供电局有限公司 | Intelligent fault early warning system for IT system operation and maintenance big data |
| CN111666978B (en) * | 2020-05-11 | 2023-12-01 | 深圳供电局有限公司 | Intelligent fault early warning system for IT system operation and maintenance big data |
| CN113033642A (en) * | 2021-03-17 | 2021-06-25 | 广东电网有限责任公司计量中心 | Intelligent electric energy meter state judgment method and system based on alarm event |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108287327A (en) | Metering automation terminal fault diagnostic method based on Bayes's classification | |
| CN106199276B (en) | The intelligent diagnosis system and method for exception information in a kind of power information acquisition system | |
| US8868985B2 (en) | Supervised fault learning using rule-generated samples for machine condition monitoring | |
| CN105846780B (en) | A kind of photovoltaic module method for diagnosing faults based on decision-tree model | |
| CN103914064B (en) | Based on the commercial run method for diagnosing faults that multi-categorizer and D-S evidence merge | |
| CN106777984A (en) | A kind of method that photovoltaic array Working state analysis and fault diagnosis are realized based on density clustering algorithm | |
| CN113032454A (en) | Interactive user power consumption abnormity monitoring and early warning management cloud platform based on cloud computing | |
| CN110311709B (en) | Fault judgment method for electricity consumption information acquisition system | |
| Gilanifar et al. | Multi-task logistic low-ranked dirty model for fault detection in power distribution system | |
| CN112464995A (en) | Power grid distribution transformer fault diagnosis method and system based on decision tree algorithm | |
| CN104899353B (en) | A kind of power quality disturbance localization method based on evidence theory | |
| CN107436277A (en) | The single index data quality control method differentiated based on similarity distance | |
| CN105974273A (en) | Power distribution network fault positioning system | |
| CN108445435A (en) | A kind of electric energy meter calibrating apparatus error online evaluation method | |
| CN103544542A (en) | Power system transient stability margin predicting method | |
| CN110443481B (en) | Power distribution automation terminal state evaluation system and method based on hybrid K-nearest neighbor algorithm | |
| CN111999591B (en) | Method for identifying abnormal state of primary equipment of power distribution network | |
| CN109284886A (en) | Electrical Safety management method and device based on artificial intelligence | |
| CN110096723B (en) | High-voltage switch cabinet insulation state analysis method based on operation and maintenance detection big data | |
| CN113721182B (en) | Method and system for evaluating reliability of online performance monitoring result of power transformer | |
| CN111190072A (en) | Centralized meter reading system diagnosis model establishing method, fault diagnosis method and fault diagnosis device | |
| CN109739197A (en) | A kind of multi-state failure prediction method of chemical spent material processing equipment | |
| CN113189513A (en) | Ripple-based redundant power supply current sharing state identification method | |
| CN109117865A (en) | One kind realizing photovoltaic array Working state analysis and method for diagnosing faults based on principal component analysis | |
| CN113010394A (en) | Machine room fault detection method for data center |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180717 |