CN109409405A - A kind of active power distribution network bad data recognition method and apparatus - Google Patents
A kind of active power distribution network bad data recognition method and apparatus Download PDFInfo
- Publication number
- CN109409405A CN109409405A CN201811087984.6A CN201811087984A CN109409405A CN 109409405 A CN109409405 A CN 109409405A CN 201811087984 A CN201811087984 A CN 201811087984A CN 109409405 A CN109409405 A CN 109409405A
- Authority
- CN
- China
- Prior art keywords
- measurement
- subregion
- matrix
- power distribution
- distribution network
- 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
- 238000009826 distribution Methods 0.000 title claims abstract description 59
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000005259 measurement Methods 0.000 claims abstract description 312
- 238000005192 partition Methods 0.000 claims abstract description 47
- 238000012360 testing method Methods 0.000 claims abstract description 12
- 239000011159 matrix material Substances 0.000 claims description 121
- 238000000638 solvent extraction Methods 0.000 claims description 57
- 230000035945 sensitivity Effects 0.000 claims description 15
- 230000003247 decreasing effect Effects 0.000 claims description 6
- 230000017105 transposition Effects 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims 1
- 238000011109 contamination Methods 0.000 abstract description 9
- 230000000694 effects Effects 0.000 abstract description 8
- 241001269238 Data Species 0.000 abstract description 7
- 238000006424 Flood reaction Methods 0.000 abstract description 6
- 238000007689 inspection Methods 0.000 abstract description 5
- 230000008569 process Effects 0.000 description 13
- 238000010586 diagram Methods 0.000 description 7
- 238000004590 computer program Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000001514 detection method Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 230000002950 deficient Effects 0.000 description 2
- JEIPFZHSYJVQDO-UHFFFAOYSA-N ferric oxide Chemical compound O=[Fe]O[Fe]=O JEIPFZHSYJVQDO-UHFFFAOYSA-N 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 235000006508 Nelumbo nucifera Nutrition 0.000 description 1
- 240000002853 Nelumbo nucifera Species 0.000 description 1
- 235000006510 Nelumbo pentapetala Nutrition 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000012850 discrimination method Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Water Supply & Treatment (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Public Health (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The present invention provides a kind of active power distribution network bad data recognition method and apparatus, determines optimum partition strategy based on all measurements in the power distribution network of acquisition, and obtain each subregion based on optimum partition strategy；Determine the suspicious measurement of each subregion, and the suspicious measurement based on all subregions constructs suspicious measurement duration set, avoids the occurrence of residual contamination and residual error floods phenomenon, not the case where not will cause erroneous judgement or missing inspection, even if there are multiple bad datas, false judgment will not occur, identification effect is good；Technical solution provided by the invention can correctly pick out bad data by increasing suspicious measurement, in conjunction with hypothesis testing identification method, and the measurement effectively flooded to generation residual error recognizes, and improve the reliability and robustness of identification.
Description
Technical field
The present invention relates to distribution network technology fields, and in particular to a kind of active power distribution network bad data recognition method and dress
It sets.
Background technique
Currently, with large number of intermittently power supply, energy-storage system distributed power supply (Distributed Generation,
DG it) is widely applied in power distribution network, the radial power distribution network of traditional passive is transformed into the active power distribution network of trend two-way flow, passes through
It is unpractiaca for a large amount of statistics meters being configured in power distribution network to reach traditional data acquisition.To complete more preferable power grid
The purpose of planning and scheduling needs to realize real time data tracking and monitoring by state estimation.But from practical consideration, state estimation
Accuracy be built upon on the basis of accurate measuring value, and measure channel error and the external world signal interference etc. because
Element, it may appear that the very big metric data of error in measurement, i.e. bad data, when the mode of being scheduled arranges, schedule workers
The scheduling decision to make mistake is done, therefore, the detection and identification for completing bad data are to realize the important prerequisite of correct status estimation.
Most of in active power distribution network bad data recognition in the prior art is to be using weighted residual or residual
Threshold value is arranged according to certain confidence level in characteristic value, and in the logic judgment for carrying out " one or the other " to measurement, this method can
The case where residual contamination capable of occur and residual error floods phenomenon, causing erroneous judgement or missing inspection, when especially there are multiple bad datas, holds
False judgment easily occurs, identification effect is poor.
Summary of the invention
In order to overcome the shortcomings of that above-mentioned identification effect in the prior art is poor, the present invention provides a kind of active power distribution network umber of defectives
According to discrimination method and device, optimum partition strategy is determined based on all measurements in the power distribution network of acquisition, and be based on optimum partition
Strategy obtains each subregion；Determine the suspicious measurement of each subregion, and the building of the suspicious measurement based on all subregions is suspicious
Duration set is measured, avoids the occurrence of residual contamination and the case where residual error floods phenomenon, not will cause erroneous judgement or missing inspection, even if occurring more
When a bad data, false judgment will not occur, identification effect is good.
In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical scheme that:
On the one hand, the present invention provides a kind of active power distribution network bad data recognition method, comprising:
Optimum partition strategy is determined based on all measurements in the power distribution network of acquisition, and is obtained often based on optimum partition strategy
A subregion；
Determine the suspicious measurement of each subregion, and the suspicious measurement based on all subregions constructs suspicious measurement quantity set
It closes；
Based on suspicious measurement duration set, bad data is recognized using hypothesis testing identification method.
All measurements determine optimum partition strategy in the power distribution network based on acquisition, comprising:
Multiple partitioning strategies are determined based on all measurements in the power distribution network of acquisition；
The multiple partitioning strategies are screened by F statistic, obtain optimum partition strategy.
The determination of the multiple partitioning strategies, comprising:
Measurement residual error vector is standardized；
Standardized residual covariance matrix is calculated based on the measurement residual error vector after standardization；
Covariance matrix based on standardized residual establishes similarity factor matrix between measurement；
Multiple partitioning strategies are obtained based on similarity factor matrix between measurement.
It is described that measurement residuals vector is standardized, such as following formula:
In formula, rNFor the measurement residual error vector after standardization, r is measurement residual error vector, and D is intermediate quantity, and
D=diag (WR), R are weight matrix, and W is residual sensitivity matrix, and v is measurement error vector, WNTurn for standardized residual
Change matrix；
The measurement residual error vector r is determined as the following formula:
Wherein, z is measurement actual value, and x is quantity of state actual value,For measurement estimated value,Represent quantity of state estimation
Value, H are constant state estimation matrix, and R is weight matrix, and W is residual sensitivity matrix, and T is transposition, and I is unit matrix.
The measurement residual error vector based on after standardization calculates standardized residual covariance matrix, such as following formula:
In formula, RNFor standardized residual covariance matrix, E indicates expectation.
The covariance matrix based on standardized residual establishes similarity factor matrix between measurement, comprising:
By the similarity degree r between measurement i and measurement jijAs the i-th row between measurement in similarity factor matrix, jth
Column element establishes similarity factor matrix between measurement；
The rijIt determines as the following formula:
In formula, rijFor the similarity degree between measurement i and measurement j, the m amount of being characterized sum, M is distance coefficient, xikFor
K-th of characteristic value of measurement i, xjkFor k-th of characteristic value of measurement j, and xikAnd xjkBy RNIt determines.
It is described that multiple partitioning strategies are obtained based on similarity factor matrix between measurement, comprising:
Based on similarity factor matrix between measurement, fuzzy equivalent matrix is determined using Transitive Closure Method as the following formula:
T (R)=(rij(λ))n*n
In formula, t (R) is the fuzzy equivalent matrix of n*n, rij(λ) is the i-th row, the jth column element of t (R), andλ is default partitioning strategies threshold value；
λ is gradually successively decreased according to preset step-length by 1, multiple partitioning strategies are obtained.
It is described that the multiple partitioning strategies are screened by F statistic, obtain optimum partition strategy, comprising:
Based on default significance, the default corresponding critical value F of significance is obtained by looking into F tables of critical valuesα；
F statistic is selected to be greater than FαCorresponding partitioning strategies are as optimum partition strategy；
The F statistic is calculated as follows:
In formula, F is F statistic, and r is subregion number, and n is measurement number, npFor measurement number in p-th of subregion,For the average value of k-th of characteristic value of measurement in p-th of subregion,It is averaged for k-th characteristic value of all measurements
Value, xpjkFor k-th of characteristic value of p-th of subregion, j-th of measurement.
The suspicious measurement of each subregion of determination, and the suspicious measurement based on all subregions constructs suspicious measurement
Set, comprising:
By the corresponding each subregion Playsization of optimum partition strategy treated measurement residual error vector and default residual error
Measurement more than preset threshold residual value is determined as the suspicious measurement in subregion, forms suspicious measurement by threshold comparison
Set；
Judge, if so, terminating, otherwise to calculate each subregion whether comprising the not suspicious measurement of only one in some subregion
Interior all normal measurements are pressed in the normal measurement and cluster at a distance from cluster centre, and by measurement normal in subregion
The sequence of the distance of the heart from small to large is added in suspicious measurement duration set.
On the other hand, the present invention also provides a kind of active power distribution network bad data recognition devices, comprising:
Determining module determines optimum partition strategy for measurements all in the power distribution network based on acquisition, and based on optimal
Partitioning strategies obtain each subregion；
Module is constructed, the suspicious measurement building for determining the suspicious measurement of each subregion, and based on all subregions
Suspicious measurement duration set；
Module is recognized, for being based on suspicious measurement duration set, bad data is recognized using hypothesis testing identification method.
The determining module includes:
Partitioning strategies determination unit determines multiple partitioning strategies for measurements all in the power distribution network based on acquisition；
Optimum partition policy determining unit obtains most for being screened by F statistic to the multiple partitioning strategies
Excellent partitioning strategies.
Partitioning strategies determination unit includes:
Standardization unit, for being standardized to measurement residual error vector；
Computing unit, for calculating standardized residual covariance square based on the measurement residual error vector after standardization
Battle array；
Unit is established, establishes similarity factor matrix between measurement for the covariance matrix based on standardized residual；
First determination unit, for obtaining multiple partitioning strategies based on similarity factor matrix between measurement.
The standardization unit is as the following formula standardized measurement residuals vector:
In formula, rNFor the measurement residual error vector after standardization, r is measurement residual error vector, and D is intermediate quantity, and
D=diag (WR), R are weight matrix, and W is residual sensitivity matrix, and v is measurement error vector, WNTurn for standardized residual
Change matrix；
The measurement residual error vector r is determined as the following formula:
Wherein, z is measurement actual value, and x is quantity of state actual value,For measurement estimated value,Represent quantity of state estimation
Value, H are constant state estimation matrix, and R is weight matrix, and W is residual sensitivity matrix, and T is transposition, and I is unit matrix.
Standardized residual covariance matrix is calculated as follows in the computing unit:
In formula, RNFor standardized residual covariance matrix, E indicates expectation.
The unit of establishing is specifically used for:
The similarity degree between measurement i and measurement j is determined as the following formula:
In formula, rijFor the similarity degree between measurement i and measurement j, the m amount of being characterized sum, M is distance coefficient, xikFor
K-th of characteristic value of measurement i, xjkFor k-th of characteristic value of measurement j, and xikAnd xjkBy RNIt determines；
By the similarity degree r between measurement i and measurement jijAs the i-th row between measurement in similarity factor matrix, jth
Column element establishes similarity factor matrix between measurement.
First determination unit is specifically used for:
Based on similarity factor matrix between measurement, fuzzy equivalent matrix is determined using Transitive Closure Method as the following formula:
T (R)=(rij(λ))n*n
In formula, t (R) is the fuzzy equivalent matrix of n*n, rij(λ) is the i-th row, the jth column element of t (R), andλ is default partitioning strategies threshold value；
λ is gradually successively decreased according to preset step-length by 1, multiple partitioning strategies are obtained.
The optimum partition policy determining unit is specifically used for:
Based on default significance, the default corresponding critical value F of significance is obtained by looking into F tables of critical valuesα；
F statistic is calculated as follows:
In formula, F is F statistic, and r is subregion number, and n is measurement number, npFor measurement number in p-th of subregion,For the average value of k-th of characteristic value of measurement in p-th of subregion,It is averaged for k-th characteristic value of all measurements
Value, xpjkFor k-th of characteristic value of p-th of subregion, j-th of measurement；
F statistic is selected to be greater than FαCorresponding partitioning strategies are as optimum partition strategy.
The building module is specifically used for:
By the corresponding each subregion Playsization of optimum partition strategy treated measurement residual error vector and default residual error
Measurement more than preset threshold residual value is determined as the suspicious measurement in subregion, forms suspicious measurement by threshold comparison
Set；
Judge, if so, terminating, otherwise to calculate each subregion whether comprising the not suspicious measurement of only one in some subregion
Interior all normal measurements are pressed in the normal measurement and cluster at a distance from cluster centre, and by measurement normal in subregion
The sequence of the distance of the heart from small to large is added in suspicious measurement duration set.
Compared with the immediate prior art, technical solution provided by the invention is had the advantages that
In active power distribution network bad data recognition method provided by the invention, first all measurements in the power distribution network based on acquisition
It measures and determines optimum partition strategy, and each subregion is obtained based on optimum partition strategy；Determine the suspicious measurement of each subregion, and
Suspicious measurement based on all subregions constructs suspicious measurement duration set, avoids the occurrence of residual contamination and residual error floods phenomenon, no
Even if there are multiple bad datas false judgment will not occur for the case where will cause erroneous judgement or missing inspection, and identification effect is good；
Active power distribution network bad data recognition device provided by the invention includes determining module, building module and identification mould
Block, determining module determine optimum partition strategy for measurements all in the power distribution network based on acquisition, and are based on optimum partition plan
Slightly obtain each subregion；Module is constructed, for determining the suspicious measurement of each subregion, and the suspicious measurement based on all subregions
Amount constructs suspicious measurement duration set；Module is recognized, for being based on suspicious measurement duration set, using hypothesis testing identification method to bad
Data are recognized, and avoid the occurrence of residual contamination and the case where residual error floods phenomenon, not will cause erroneous judgement or missing inspection, even if occurring
When multiple bad datas, false judgment will not occur, identification effect is good；
Technical solution provided by the invention can be distinguished correctly by increasing suspicious measurement in conjunction with hypothesis testing identification method
Know bad data out, the measurement effectively flooded to generation residual error recognizes, and improves the reliability and robustness of identification.
Detailed description of the invention
Fig. 1 is active power distribution network bad data recognition method flow diagram in the embodiment of the present invention 1；
Fig. 2 is 33 Node power distribution system structure chart of IEEE in the embodiment of the present invention 3.
Specific embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
Embodiment 1
The embodiment of the present invention 1 provides a kind of active power distribution network bad data recognition method, specific flow chart such as Fig. 1 institute
Show, detailed process is as follows:
S101: optimum partition strategy is determined based on all measurements in the power distribution network of acquisition, and based on optimum partition strategy
Obtain each subregion；
S102: determining the suspicious measurement of each subregion, and the suspicious measurement based on all subregions constructs suspicious measurement
Duration set；
S103: it is based on suspicious measurement duration set, using hypothesis testing identification method (HTI, Hypothesis Testing
Identification) bad data is recognized.
In above-mentioned S101, optimum partition strategy is determined based on all measurements in the power distribution network of acquisition, comprising:
Multiple partitioning strategies are determined based on all measurements in the power distribution network of acquisition；
The multiple partitioning strategies are screened by F statistic, obtain optimum partition strategy.
The specific determination process of multiple partitioning strategies is as follows:
Measurement residual error vector is standardized；
Standardized residual covariance matrix is calculated based on the measurement residual error vector after standardization；
Covariance matrix based on standardized residual establishes similarity factor matrix between measurement；
Multiple partitioning strategies are obtained based on similarity factor matrix between measurement.
It is above-mentioned that measurement residuals vector is standardized as the following formula:
In formula, rNFor the measurement residual error vector after standardization, r is measurement residual error vector, and D is intermediate quantity, and
D=diag (WR), R are weight matrix, and W is residual sensitivity matrix, and v is measurement error vector, WNTurn for standardized residual
Change matrix；
Measurement residual error vector r therein is determined as the following formula:
Wherein, z is measurement actual value, and x is quantity of state actual value,For measurement estimated value,Represent quantity of state estimation
Value, H are constant state estimation matrix, and R is weight matrix, and W is residual sensitivity matrix, and T is transposition, and I is unit matrix.
The expression formula of measurement residual error vector r reflects the relationship between measurement residual error and measurement error, from measurement
Measuring can be seen that in the expression formula of residual error vector r, configuration, network structure, network parameter, the operation of residual sensitivity matrix and measurement
State etc. is related.In addition, showing residual sensitivity matrix and operating status relationship and little by a large amount of actual operation, bear
Lotus variation and the variation of individual bad datas can only cause the minor change of residual sensitivity matrix.Therefore, in electric system net
In the case that network structure and measure configuration point are constant, constant matrix can be considered as by W gusts.
The above-mentioned measurement residual error vector based on after standardization, is calculated as follows standardized residual covariance matrix:
In formula, RNFor standardized residual covariance matrix, E indicates expectation.
By above-mentioned rNAnd RNRespective expression formula can be seen that standardization covariance diagonal element is 1, the member in matrix
Plain absolute value is bigger, and the correlation represented between node residual error is bigger.When there is single bad data, assisted using standardization
Variance method can obtain better effects to carry out raw data detection.But when the multiple bad datas of generation or bad data are larger
When, then it influences whether the residual error of remaining measuring point, residual contamination occurs or floods phenomenon.And when the mutual pass between measurement residuals
When being bigger, residual contamination is also more easy to happen between these measurements and is flooded with residual error.
The above-mentioned covariance matrix based on standardized residual establishes similarity factor matrix between measurement, and detailed process is as follows:
By the similarity degree r between measurement i and measurement jijAs the i-th row between measurement in similarity factor matrix, jth
Column element establishes similarity factor matrix between measurement；
R thereinijIt determines as the following formula:
In formula, rijFor the similarity degree between measurement i and measurement j, rijBetween [0,1]；The m amount of being characterized sum,
M is distance coefficient, M desirable 0.02；xikFor k-th of characteristic value of measurement i, xjkFor k-th of characteristic value of measurement j, and
xikAnd xjkBy RNIt determines.
Above-mentioned to obtain multiple partitioning strategies based on similarity factor matrix between measurement, detailed process is as follows:
Based on similarity factor matrix between measurement, fuzzy equivalent matrix is determined using Transitive Closure Method as the following formula:
T (R)=(rij(λ))n*n
In formula, t (R) is the fuzzy equivalent matrix of n*n, rij(λ) is the i-th row, the jth column element of t (R), andλ is default partitioning strategies threshold value；
λ is gradually successively decreased according to preset step-length (can be 0.05) by 1, multiple partitioning strategies are obtained.
In above-mentioned S101, multiple partitioning strategies are screened by F statistic, obtain optimum partition strategy, F statistic
Application further improve identification effect, detailed process is as follows:
Based on default significance, the default corresponding critical value F of significance is obtained by looking into F tables of critical valuesα；
F statistic is calculated as follows:
In formula, F is F statistic, and r is subregion number, and n is measurement number, npFor measurement number in p-th of subregion,For the average value of k-th of characteristic value of measurement in p-th of subregion,It is averaged for k-th characteristic value of all measurements
Value, xpjkFor k-th of characteristic value of p-th of subregion, j-th of measurement；
F statistic is selected to be greater than FαCorresponding partitioning strategies are as optimum partition strategy.If only having one in subregion at this time
Measurement then needs to calculate measurement at a distance from each subregion average value, and the measurement is incorporated to measurement distance most
In short region.Finally, classification schemes can determine.
In above-mentioned S102, the suspicious measurement of each subregion is determined, and the building of the suspicious measurement based on all subregions can
It doubts and measures duration set, detailed process is as follows:
By the corresponding each subregion Playsization of optimum partition strategy treated measurement residual error vector and default residual error
Threshold value (desirable 2.5) comparison, is determined as the suspicious measurement in subregion for the measurement more than preset threshold residual value, formation can
It doubts and measures duration set (if having subregion includes the not suspicious measurement of only one, it is possible to residual error occur and flood phenomenon, need to increase suspicious
Measure collection)；
Judge, if so, terminating, otherwise to calculate each subregion whether comprising the not suspicious measurement of only one in some subregion
Interior all normal measurements are pressed in the normal measurement and cluster at a distance from cluster centre, and by measurement normal in subregion
The sequence of the distance of the heart from small to large is added in suspicious measurement duration set.
Embodiment 2
Based on the same inventive concept, the embodiment of the present invention 2 also provides a kind of active power distribution network bad data recognition device, packet
Determining module, building module and identification module are included, the function of above-mentioned several modules is described in detail below:
Determining module therein determines optimum partition strategy, and base for measurements all in the power distribution network based on acquisition
In optimum partition, strategy obtains each subregion；
Building module therein, for determining the suspicious measurement of each subregion, and the suspicious measurement based on all subregions
Amount constructs suspicious measurement duration set；
Identification module therein, for be based on suspicious measurement duration set, using hypothesis testing identification method (HTI,
Hypothesis Testing Identification) bad data is recognized.
Above-mentioned determining module specifically includes:
Partitioning strategies determination unit determines multiple partitioning strategies for measurements all in the power distribution network based on acquisition；
Optimum partition policy determining unit obtains most for being screened by F statistic to the multiple partitioning strategies
Excellent partitioning strategies.
Above-mentioned partitioning strategies determination unit is used to determine multiple partitioning strategies, specifically include:
Standardization unit, for being standardized to measurement residual error vector；
Computing unit, for calculating standardized residual covariance square based on the measurement residual error vector after standardization
Battle array；
Unit is established, establishes similarity factor matrix between measurement for the covariance matrix based on standardized residual；
First determination unit, for obtaining multiple partitioning strategies based on similarity factor matrix between measurement.
Above-mentioned standardization unit is as the following formula standardized measurement residuals vector:
In formula, rNFor the measurement residual error vector after standardization, r is measurement residual error vector, and D is intermediate quantity, and
D=diag (WR), R are weight matrix, and W is residual sensitivity matrix, and v is measurement error vector, WNTurn for standardized residual
Change matrix；
Measurement residual error vector r therein is determined as the following formula:
Wherein, z is measurement actual value, and x is quantity of state actual value,For measurement estimated value,Represent quantity of state estimation
Value, H are constant state estimation matrix, and R is weight matrix, and W is residual sensitivity matrix, and T is transposition, and I is unit matrix.
Standardized residual covariance matrix is calculated as follows in above-mentioned computing unit:
In formula, RNFor standardized residual covariance matrix, E indicates expectation.
It is above-mentioned to establish covariance matrix of the unit based on standardized residual, similarity factor between measurement is established by following processes
Matrix:
The similarity degree between measurement i and measurement j is determined as the following formula:
In formula, rijFor the similarity degree between measurement i and measurement j, the m amount of being characterized sum, M is distance coefficient, xikFor
K-th of characteristic value of measurement i, xjkFor k-th of characteristic value of measurement j, and xikAnd xjkBy RNIt determines；
By the similarity degree r between measurement i and measurement jijAs the i-th row between measurement in similarity factor matrix, jth
Column element establishes similarity factor matrix between measurement.
The first above-mentioned determination unit is based on similarity factor matrix between measurement, obtains multiple subregion plans by following processes
Slightly:
Based on similarity factor matrix between measurement, fuzzy equivalent matrix is determined using Transitive Closure Method as the following formula:
T (R)=(rij(λ))n*n
In formula, t (R) is the fuzzy equivalent matrix of n*n, rij(λ) is the i-th row, the jth column element of t (R), andλ is default partitioning strategies threshold value；
λ is gradually successively decreased according to preset step-length by 1, multiple partitioning strategies are obtained.
Above-mentioned optimum partition policy determining unit screens multiple partitioning strategies by F statistic, obtains optimal
Partitioning strategies, detailed process is as follows:
Based on default significance, the default corresponding critical value F of significance is obtained by looking into F tables of critical valuesα；
F statistic is calculated as follows:
In formula, F is F statistic, and r is subregion number, and n is measurement number, npFor measurement number in p-th of subregion,For the average value of k-th of characteristic value of measurement in p-th of subregion,It is averaged for k-th characteristic value of all measurements
Value, xpjkFor k-th of characteristic value of p-th of subregion, j-th of measurement；
F statistic is selected to be greater than FαCorresponding partitioning strategies are as optimum partition strategy.
Above-mentioned building module determines the suspicious measurement of each subregion, and the building of the suspicious measurement based on all subregions
Suspicious measurement duration set, detailed process is as follows:
By the corresponding each subregion Playsization of optimum partition strategy treated measurement residual error vector and default residual error
Measurement more than preset threshold residual value is determined as the suspicious measurement in subregion, forms suspicious measurement by threshold comparison
Set；
Judge, if so, terminating, otherwise to calculate each subregion whether comprising the not suspicious measurement of only one in some subregion
Interior all normal measurements are pressed in the normal measurement and cluster at a distance from cluster centre, and by measurement normal in subregion
The sequence of the distance of the heart from small to large is added in suspicious measurement duration set.
Embodiment 3
The embodiment of the present invention 3 is based on 33 node power distribution system of IEEE by taking 33 Node power distribution system of IEEE shown in Fig. 2 as an example
System carries out bad data and debates knowledge.In Fig. 2, photovoltaic power generation system is accessed in total 8 positions of node 5,7,13,17,19,26,29,24
System accesses photovoltaic generating system PV1 in node 5, access photovoltaic generating system PV2 in node 7, accesses photovoltaic in node 13
Electricity generation system PV3, accesses photovoltaic generating system PV4 in node 17, photovoltaic generating system PV5 is accessed in node 19, in node 26
Photovoltaic generating system PV7 is accessed, accesses photovoltaic generating system PV8 in node 29, accesses photovoltaic generating system PV6 in node 24,
Access capacity is respectively as follows: 10MW, 20MW, 5MW, 15MW, 20MW, 10MW, 15MW, 5MW.
To distributed generation resource in 33 Node power distribution system of IEEE, the data acquired when using t=9h are tested, used
Measure includes all node injecting powers and branch power.It is as shown in the table for its active measurement number, including node injecting power, branch
Road power and voltage modulus value three classes, interior joint injecting power PiIt indicates, branch active power PijIt indicates, voltage modulus value is used
Vi, then active measurement measuring point allocation table and idle measurement measuring point allocation table are respectively Tables 1 and 2.
Table 1
Table 2
In conjunction with Fuzzy classification, the F and F when the number of partitions is 90.05Difference it is maximum, conspicuousness with higher.Selection
This partitioned mode, optimally partitioned policy accounting table are as shown in table 3:
Table 3
| Area code | Measurement | Area code | Measurement |
| 1 | 1,34,35,56,61 | 6 | 7-10,43-47 |
| 2 | 2,21,22,36-39 | 7 | 11-15,48-50,59 |
| 3 | 3,23-25,37,57 | 8 | 16-18,51-53,63 |
| 4 | 4,5,38-40,58,62 | 9 | 19,20,54,55,60 |
| 5 | 6,26-33,41,42 |
Two kinds of situations on weak related and strong correlation node occur from bad data respectively and (increase umber of defectives in certain measuring point
According to whether in a subregion judging whether correlative measuring point belongs to strong correlation or weak phase according to it after analyzing by partitioning algorithm
It closes.Belong to strong correlation if dividing in an area, otherwise as weak correlation；) the bad data recognition scheme proposed is carried out
Test:
1) bad data occurs on weak interdependent node
The bad data of 10MW is added at measurement number 3,15, as shown in table 1, they are located in different subregions,
Correlation is weaker.Metric data is sent into state estimating unit, while being standardized residual detection, it is bad between weak interdependent node
Data debate knowledge, and the results are shown in Table 4:
Table 4
| Measuring point | 3 | 15 |
| Residual error | 3.725 | 3.663 |
As can be seen from Table 4, two measurements are in different subregions, so obscuring there is no residual contamination occurs
The method combination HTI of cluster can effectively debate knowledge bad data.
2) bad data occurs on strong correlation node
Measuring the bad data that 10MW is added at number measurement for being 6,33.It is standardized residual detection, and is combined
HTI, bad data debates knowledge the results are shown in Table 5 between strong correlation node.
Table 5
| Measuring point | 6 | 26 | 33 | 41 |
| Residual error | 4.933 | 7.135 | 6.170 | 9.411 |
| It whether is bad data | It is | It is no | It is | It is no |
As can be seen from Table 5, residual contamination directly can occur using residual covariance detection, it is bad cannot correctly debates knowledge
Data.In conjunction with residual error herein cluster and HTI, due to that can be located in same subregion with data, method regards all suspicious datas
For one kind, cluster centre is calculated, remaining is then calculated and measures at a distance from the center, under the premise of not influencing observability
By distance from the near to the distant in the way of measurement is added into suspicious measurement duration set, then carry out debating knowledge.It can be with by data in table
Find out, after being increased using context of methods suspicious measurement collection, HTI can correctly pick out bad data, effectively
Judge the data that residual error is flooded occur.
For convenience of description, each section of apparatus described above is divided into various modules with function or unit describes respectively.
Certainly, each module or the function of unit can be realized in same or multiple softwares or hardware when implementing the application.
It should be understood by those skilled in the art that, embodiments herein can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the application
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the application, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The application is referring to method, the process of equipment (system) and computer program product according to the embodiment of the present application
Figure and/or block diagram describe.It should be understood that can be realized by computer program instructions each in flowchart and/or the block diagram
The combination of process and/or box in process and/or box and flowchart and/or the block diagram.It can provide these computers
Processor of the program instruction to general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices
To generate a machine, so that generating use by the instruction that computer or the processor of other programmable data processing devices execute
In the dress for realizing the function of specifying in one or more flows of the flowchart and/or one or more blocks of the block diagram
It sets.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Finally it should be noted that: the above embodiments are merely illustrative of the technical scheme of the present invention and are not intended to be limiting thereof, institute
The those of ordinary skill in category field can still modify to a specific embodiment of the invention referring to above-described embodiment or
Equivalent replacement, these are applying for this pending hair without departing from any modification of spirit and scope of the invention or equivalent replacement
Within bright claims.
Claims (18)
1. a kind of active power distribution network bad data recognition method characterized by comprising
Optimum partition strategy is determined based on all measurements in the power distribution network of acquisition, and obtains each point based on optimum partition strategy
Area；
Determine the suspicious measurement of each subregion, and the suspicious measurement based on all subregions constructs suspicious measurement duration set；
Based on suspicious measurement duration set, bad data is recognized using hypothesis testing identification method.
2. active power distribution network bad data recognition method according to claim 1, which is characterized in that described based on acquisition
All measurements determine optimum partition strategy in power distribution network, comprising:
Multiple partitioning strategies are determined based on all measurements in the power distribution network of acquisition；
The multiple partitioning strategies are screened by F statistic, obtain optimum partition strategy.
3. active power distribution network bad data recognition method according to claim 2, which is characterized in that the multiple subregion plan
Determination slightly, comprising:
Measurement residual error vector is standardized；
Standardized residual covariance matrix is calculated based on the measurement residual error vector after standardization；
Covariance matrix based on standardized residual establishes similarity factor matrix between measurement；
Multiple partitioning strategies are obtained based on similarity factor matrix between measurement.
4. active power distribution network bad data recognition method according to claim 3, which is characterized in that described to measurement residuals
Vector is standardized, such as following formula:
In formula, rNFor the measurement residual error vector after standardization, r is measurement residual error vector, and D is intermediate quantity, and D=
Diag (WR), R are weight matrix, and W is residual sensitivity matrix, and v is measurement error vector, WNSquare is converted for standardized residual
Battle array；
The measurement residual error vector r is determined as the following formula:
Wherein, z is measurement actual value, and x is quantity of state actual value,For measurement estimated value,Represent quantity of state estimated value, H
For constant state estimation matrix, R is weight matrix, and W is residual sensitivity matrix, and T is transposition, and I is unit matrix.
5. active power distribution network bad data recognition method according to claim 4, which is characterized in that described based on standardization
Treated, and measurement residual error vector calculates standardized residual covariance matrix, such as following formula:
In formula, RNFor standardized residual covariance matrix, E indicates expectation.
6. active power distribution network bad data recognition method according to claim 5, which is characterized in that described based on standardization
The covariance matrix of residual error establishes similarity factor matrix between measurement, comprising:
By the similarity degree r between measurement i and measurement jijAs the i-th row between measurement in similarity factor matrix, jth column member
Element establishes similarity factor matrix between measurement；
The rijIt determines as the following formula:
In formula, rijFor the similarity degree between measurement i and measurement j, the m amount of being characterized sum, M is distance coefficient, xikTo measure
Measure k-th of characteristic value of i, xjkFor k-th of characteristic value of measurement j, and xikAnd xjkBy RNIt determines.
7. active power distribution network bad data recognition method according to claim 6, which is characterized in that described to be based on measurement
Between similarity factor matrix obtain multiple partitioning strategies, comprising:
Based on similarity factor matrix between measurement, fuzzy equivalent matrix is determined using Transitive Closure Method as the following formula:
T (R)=(rij(λ))n*n
In formula, t (R) is the fuzzy equivalent matrix of n*n, rij(λ) is the i-th row, the jth column element of t (R), andλ is default partitioning strategies threshold value；
λ is gradually successively decreased according to preset step-length by 1, multiple partitioning strategies are obtained.
8. active power distribution network bad data recognition method according to claim 6, which is characterized in that described to be counted by F
Amount screens the multiple partitioning strategies, obtains optimum partition strategy, comprising:
Based on default significance, the default corresponding critical value F of significance is obtained by looking into F tables of critical valuesα；
F statistic is selected to be greater than FαCorresponding partitioning strategies are as optimum partition strategy；
The F statistic is calculated as follows:
In formula, F is F statistic, and r is subregion number, and n is measurement number, npFor measurement number in p-th of subregion,For
The average value of k-th of characteristic value of measurement in p-th of subregion,For the average value of k-th of characteristic value of all measurements,
xpjkFor k-th of characteristic value of p-th of subregion, j-th of measurement.
9. active power distribution network bad data recognition method according to claim 1, which is characterized in that the determination each divides
The suspicious measurement in area, and the suspicious measurement based on all subregions constructs suspicious measurement duration set, comprising:
By the corresponding each subregion Playsization of optimum partition strategy treated measurement residual error vector and default threshold residual value
Measurement more than preset threshold residual value is determined as the suspicious measurement in subregion, forms suspicious measurement duration set by comparison；
Judge, if so, terminating, otherwise to calculate institute in each subregion whether comprising the not suspicious measurement of only one in some subregion
There is normal measurement at a distance from cluster centre, and by measurement normal in subregion by the normal measurement and cluster centre
The sequence of distance from small to large is added in suspicious measurement duration set.
10. a kind of active power distribution network bad data recognition device characterized by comprising
Determining module determines optimum partition strategy for measurements all in the power distribution network based on acquisition, and is based on optimum partition
Strategy obtains each subregion；
Module is constructed, the suspicious measurement building for determining the suspicious measurement of each subregion, and based on all subregions is suspicious
Measure duration set；
Module is recognized, for being based on suspicious measurement duration set, bad data is recognized using hypothesis testing identification method.
11. active power distribution network bad data recognition device according to claim 10, which is characterized in that the determining module
Include:
Partitioning strategies determination unit determines multiple partitioning strategies for measurements all in the power distribution network based on acquisition；
Optimum partition policy determining unit obtains most optimal sorting for screening by F statistic to the multiple partitioning strategies
Area's strategy.
12. active power distribution network bad data recognition device according to claim 11, which is characterized in that the partitioning strategies
Determination unit includes:
Standardization unit, for being standardized to measurement residual error vector；
Computing unit, for calculating standardized residual covariance matrix based on the measurement residual error vector after standardization；
Unit is established, establishes similarity factor matrix between measurement for the covariance matrix based on standardized residual；
First determination unit, for obtaining multiple partitioning strategies based on similarity factor matrix between measurement.
13. active power distribution network bad data recognition device according to claim 12, which is characterized in that at the standardization
Reason unit is as the following formula standardized measurement residuals vector:
In formula, rNFor the measurement residual error vector after standardization, r is measurement residual error vector, and D is intermediate quantity, and D=
Diag (WR), R are weight matrix, and W is residual sensitivity matrix, and v is measurement error vector, WNSquare is converted for standardized residual
Battle array；
The measurement residual error vector r is determined as the following formula:
Wherein, z is measurement actual value, and x is quantity of state actual value,For measurement estimated value,Represent quantity of state estimated value, H
For constant state estimation matrix, R is weight matrix, and W is residual sensitivity matrix, and T is transposition, and I is unit matrix.
14. active power distribution network bad data recognition device according to claim 13, which is characterized in that the computing unit
Standardized residual covariance matrix is calculated as follows:
In formula, RNFor standardized residual covariance matrix, E indicates expectation.
15. active power distribution network bad data recognition device according to claim 14, which is characterized in that described to establish unit
It is specifically used for:
The similarity degree between measurement i and measurement j is determined as the following formula:
In formula, rijFor the similarity degree between measurement i and measurement j, the m amount of being characterized sum, M is distance coefficient, xikTo measure
Measure k-th of characteristic value of i, xjkFor k-th of characteristic value of measurement j, and xikAnd xjkBy RNIt determines；
By the similarity degree r between measurement i and measurement jijAs the i-th row between measurement in similarity factor matrix, jth column member
Element establishes similarity factor matrix between measurement.
16. active power distribution network bad data recognition device according to claim 15, which is characterized in that described first determines
Unit is specifically used for:
Based on similarity factor matrix between measurement, fuzzy equivalent matrix is determined using Transitive Closure Method as the following formula:
T (R)=(rij(λ))n*n
In formula, t (R) is the fuzzy equivalent matrix of n*n, rij(λ) is the i-th row, the jth column element of t (R), andλ is default partitioning strategies threshold value；
λ is gradually successively decreased according to preset step-length by 1, multiple partitioning strategies are obtained.
17. active power distribution network bad data recognition device according to claim 15, which is characterized in that the optimum partition
Policy determining unit is specifically used for:
Based on default significance, the default corresponding critical value F of significance is obtained by looking into F tables of critical valuesα；
F statistic is calculated as follows:
In formula, F is F statistic, and r is subregion number, and n is measurement number, npFor measurement number in p-th of subregion,For
The average value of k-th of characteristic value of measurement in p-th of subregion,For the average value of k-th of characteristic value of all measurements,
xpjkFor k-th of characteristic value of p-th of subregion, j-th of measurement；
F statistic is selected to be greater than FαCorresponding partitioning strategies are as optimum partition strategy.
18. active power distribution network bad data recognition device according to claim 10, which is characterized in that the building module
It is specifically used for:
By the corresponding each subregion Playsization of optimum partition strategy treated measurement residual error vector and default threshold residual value
Measurement more than preset threshold residual value is determined as the suspicious measurement in subregion, forms suspicious measurement duration set by comparison；
Judge, if so, terminating, otherwise to calculate institute in each subregion whether comprising the not suspicious measurement of only one in some subregion
There is normal measurement at a distance from cluster centre, and by measurement normal in subregion by the normal measurement and cluster centre
The sequence of distance from small to large is added in suspicious measurement duration set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811087984.6A CN109409405A (en) | 2018-09-18 | 2018-09-18 | A kind of active power distribution network bad data recognition method and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811087984.6A CN109409405A (en) | 2018-09-18 | 2018-09-18 | A kind of active power distribution network bad data recognition method and apparatus |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109409405A true CN109409405A (en) | 2019-03-01 |
Family
ID=65465573
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811087984.6A Pending CN109409405A (en) | 2018-09-18 | 2018-09-18 | A kind of active power distribution network bad data recognition method and apparatus |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109409405A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110995153A (en) * | 2019-12-18 | 2020-04-10 | 国网电子商务有限公司 | Abnormal data detection method and device for photovoltaic power station and electronic equipment |
| CN113702769A (en) * | 2021-08-30 | 2021-11-26 | 国家电网有限公司 | Power distribution network abnormity monitoring and positioning method based on monitoring data space-time correlation |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103154934A (en) * | 2010-06-11 | 2013-06-12 | Abb研究有限公司 | Detecting state estimation network model data errors |
| CN107506824A (en) * | 2017-08-31 | 2017-12-22 | 广东工业大学 | Bad the observation data detection method and device of a kind of power distribution network |
-
2018
- 2018-09-18 CN CN201811087984.6A patent/CN109409405A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103154934A (en) * | 2010-06-11 | 2013-06-12 | Abb研究有限公司 | Detecting state estimation network model data errors |
| CN107506824A (en) * | 2017-08-31 | 2017-12-22 | 广东工业大学 | Bad the observation data detection method and device of a kind of power distribution network |
Non-Patent Citations (1)
| Title |
|---|
| 张永超: "电力系统不良数据检测和辨识方法研究", 工程科技Ⅱ辑, 16 January 2010 (2010-01-16), pages 7 - 50 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110995153A (en) * | 2019-12-18 | 2020-04-10 | 国网电子商务有限公司 | Abnormal data detection method and device for photovoltaic power station and electronic equipment |
| CN113702769A (en) * | 2021-08-30 | 2021-11-26 | 国家电网有限公司 | Power distribution network abnormity monitoring and positioning method based on monitoring data space-time correlation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104794206B (en) | A kind of substation data QA system and method | |
| Banerjee et al. | Validating clusters using the Hopkins statistic | |
| CN111693931A (en) | Intelligent electric energy meter error remote calculation method and device and computer equipment | |
| CN104280612B (en) | Distributed harmonic source identification method based on single-frequency current transmission characteristics | |
| CN109409628A (en) | Acquisition terminal production firm evaluation method based on metering big data Clustering Model | |
| Li et al. | The lithium-ion battery state-of-charge estimation using random forest regression | |
| CN107632262A (en) | A kind of detection method and device of power battery pack cycle life | |
| CN110348665A (en) | A kind of low-voltage platform area electric power system data quality evaluating method and device | |
| CN109613440A (en) | Stage division, device, equipment and the storage medium of battery | |
| CN109039503A (en) | A kind of frequency spectrum sensing method, device, equipment and computer readable storage medium | |
| CN109359665A (en) | A kind of family's electric load recognition methods and device based on support vector machines | |
| CN109272190A (en) | A kind of power cable state evaluation method based on momentum BP neural network | |
| CN109409405A (en) | A kind of active power distribution network bad data recognition method and apparatus | |
| CN114048819A (en) | Power distribution network topology identification method based on attention mechanism and convolutional neural network | |
| CN110348508A (en) | Examine the data checking method and its system, electronic equipment of exceptional value | |
| CN113406555A (en) | Electric quantity metering error monitoring method of alternating current charging station and charging station | |
| CN114678962B (en) | Distributed array temperature measurement abnormal data transmission monitoring system based on power internet of things | |
| CN109639736A (en) | A kind of Power system state estimation malicious attack detection and localization method based on OPTICS | |
| CN114330486A (en) | Power system bad data identification method based on improved Wasserstein GAN | |
| CN108830006A (en) | Linear-nonlinear industrial processes fault detection method based on the linear evaluation factor | |
| CN115248906A (en) | State error identification method and system for double current transformers on outgoing line of generator | |
| CN109165665A (en) | A kind of category analysis method and system | |
| Yang et al. | A detection method for group fixed ratio electricity thieves based on correlation analysis of non-technical loss | |
| CN103914482A (en) | CMDB (Configuration Management Date Base) based centralized monitoring event influence determination method | |
| CN115166625A (en) | Intelligent ammeter error estimation method and device |
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 |