CN108132423A - A kind of method for rapidly positioning based on state transition probability power system monitoring data distortion - Google Patents
A kind of method for rapidly positioning based on state transition probability power system monitoring data distortion Download PDFInfo
- Publication number
- CN108132423A CN108132423A CN201711340903.4A CN201711340903A CN108132423A CN 108132423 A CN108132423 A CN 108132423A CN 201711340903 A CN201711340903 A CN 201711340903A CN 108132423 A CN108132423 A CN 108132423A
- Authority
- CN
- China
- Prior art keywords
- data
- state
- equipment
- monitoring
- matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
Abstract
The present invention relates to a kind of method for rapidly positioning based on state transition probability power system monitoring data distortion, this method pays close attention to the state change probability of monitoring data, it is compared by the threshold value of the feature and equipment runtime data normal variation of the probability distribution after repeatedly shifting, export distortion positional matrix, the quick position for positioning distortion data.Entirety is divided into four step 1. monitoring data attribute entities and divides, the monitoring data transition probability of 2. power equipments, and the multiple monitoring transfer matrix of power equipment, 4. distortion datas positioning is measured with subsystem distortion level.Various data formats or data structure can be unified into state transition probability by data of the present invention acquisition rank, therefore evaded different data format in multi-source heterogeneous data and influenced caused by data analysis, reduce the complexity of analysis system.
Description
Technical field
The invention belongs to the application study that Electric Power Communication Data and big data technology blend, by by the shape of detection data
State variation probability words, by the uniform format of data each in power communication system into the transfering probability distribution of data mode, then pass through
Qie Puman-kolmogorov equation is repeatedly shifted to initially shifting distribution, and the theoretical distribution repeatedly shifted passes through
The change threshold of each data limits, and calculates the distortion positional matrix that can directly position distortion data.
Background technology
Power communication system is the concept of a broad sense, refer to each subsystem related to electric power network and they generate
Data information, with the continuous development of China's electric power network, the continuous expansion of power demand, the number generated in power communication system
According to also increasingly huge, while the speed that data generate also is getting faster, and the data structure between different sub-systems also has very big
Difference, the data that power communication system generates become typical big data.
Power communication system is to ensure the important system of electric system normal operation, and equipment is carried out by various kinds of sensors
Monitoring, provides decision for equipment fault, foundation is provided for maintenance of equipment.Large-scale power communication system generates the monitoring number of magnanimity
According to these data inevitably will appear data distortion phenomenon during acquisition, typing, transmission, exchange and storage.Reality
In, these distortion datas have become positioning with analyzing the important hindering factor of electrical equipment fault.Improve power communication system
The quality of data be to improve the important link of electric power network system.Domestic and international expert proposes the detection of distortion data in electric system
A variety of solutions are gone out, have occurred the reason of data distortion, then from reason setting about in certain research Energy Management System, solve number
The problem of according to distortion.Certain research is to set about attempting to improve the quality of data from data platform.Certain research comes from the angle of interpolation fitting
High speed model of the prediction data quality researchs based on common information model (Common Information Model, CIM) is handed over
The data check technology between the different system that form CIM/E texts are carrier is changed, is screened using improved multi-source data more high-quality
The means of data, and the method fed back according to main website state estimation to field data are measured, improves power network dispatching system
Overall data quality.
Raw data detection above based on status of electric power estimation is in the Quality advance for treating office system local data
When have a certain effect, but the multi-source heterogeneous big data generated for entire power communication system does not have good be applicable in still
Property, and it is relatively high to establish the cost of corresponding knowledge base for each data distortion.The present invention is proposed based on state
The quick positioning of transition probability power system monitoring data distortion is not concerned with the multi-source heterogeneous form of electric power monitoring data, then
The state change of data is paid close attention to, whether the probability of high spot reviews monitoring data variation is consistent with equipment actual change, passes through
National grid truthful data collection verification algorithm, the experimental results showed that this method is suitable for the power communication system under big data environment
The quick positioning of the multi-source heterogeneous data distortion generated.
Invention content
Power communication system is to ensure the necessary information Transmission system of electric system normal table operation, passes through all kinds of sensings
Device monitors electrical equipment in real time, and abnormal data is reported to ensure equipment stable operation in time, decision is provided for equipment fault
It supports, basis on location is provided for maintenance of equipment.At present, as the scale of network system is increasing and all kinds of power equipments are supervised
Survey sensor type gradually increase, power communication system daily all generate magnanimity monitoring data, these data acquisition,
It inevitably will appear data distortion phenomenon during typing, transmission, exchange and storage.In fact, in the big data epoch, this
Class distortion data has become one of the important hindering factor of positioning with analyzing electrical equipment fault.Occur in power communication system
Data distortion mainly include following two aspects：
1. violating monitoring data consistency, monitoring data unanimously refer to, system physical record to data whether meet one
Fixed functional dependence or logical relation, if having the data beyond attribute definition domain.
2. violating monitoring data integrality, monitoring data integrality is that the data that power information system is actually typing are to exist
Missing, if the completely recorded total data by design requirement record.
It is a kind of automatic right the present invention is directed to establish for the relatively low problem of the quality of data occurred in current power communication system
The random process quick discrimination method that electric power data is identified into line distortion, positioning is measured with distortion level, the concern weight of this method
Point is the consistency of the variation and the variation of physical device operating status of electric power data state.By data each in power communication system
Uniform format into data mode transfering probability distribution, then by Qie Puman-kolmogorov equation to initially shifting distribution
It is repeatedly shifted, the theoretical distribution repeatedly shifted, is limited by the change threshold of each data, calculating directly to determine
The distortion positional matrix of position distortion data.
For completion more than target, the present invention is whole comprising four steps, and overall flow figure is shown in attached drawing 1
Step 1 monitoring data attribute entities divide
Monitoring data acquisition is essentially referred in electric power network system, and various kinds of sensors is monitored equipment and will monitoring
Data are transmitted to the process of designated position storage.Monitoring data have a variety of storage mechanisms, this step in different subsystems
Purpose be so that collected data are divided into solid data set according to the entity device of data source.This step is divided into 3
Sub-step：
Step 1.1 primary monitoring data acquires
Define 1 original data set
D={ d11,d12,...,d21,d22,...,dn1,dn2,...,dnk}
Wherein, in D expressions system collected all kinds of initial data attribute set, dijRepresent the kth category of the i-th equipment
Property, in actual information acquisition, since the data acquisition modes of each subsystem of power grid are not quite similar, so each subsystem is acquired
It is often mixed and disorderly data acquisition system after data summarization afterwards.
Step 1.2 monitoring data are classified by entity source address
In data acquisition, data are classified according to source index field, this step is divided into two kinds of forms.
Form one：The classification of gathered data
For shaped like collected data in defining 1, needing to classify according to the data source index field in data attribute,
One group will be divided into from the monitoring data of identical entity
Form two：Monitoring data are acquired by entity
For the subsystem of entity output monitoring data can be pressed, its monitoring data is directly acquired, and indicated in fact in data
The unique attribute of body.
Define 2 device datas vector
di=(di1,di2,...,dik)
Wherein, diExpression comes from the monitoring data vector of i-th (1≤i≤n) equipment, dijRepresent the jth of the i-th equipment, (1
≤ j≤k) attribute.The data sorted out in this way by the data source of arbitrary entity device i are in vector form by diRecord
Step 1.3 structure attribute extended matrix
In power communication system, electric power network includes each subsystem, and various kinds of equipment is by cooperating to support one
The normal operation of a subsystem, the data variation of set of subsystem can be formalized by attribute extension matrix.
Define 3 attribute extension matrixes
Wherein, M (t) represents the attribute extension matrix of t moment, dij(t) jth of i-th (1≤i≤n) equipment, (1≤j are represented
≤ki≤ k) monitor value of the attribute in moment t.The upper bound k of number of attributes, wherein k=that equipment possesses defined in above formula
max(k1,k2...) and represent the attribute number for possessing most attribute equipment in subsystem, it defines the columns of extended matrix.Electricity
The attribute number of distinct device can be different in power subsystem, and most row vectors in such extended matrix M (t) are not fixed
The property value of justice, such undefined attribute is referred to as extended attribute, their effect is to keep the rectangle knot of extended matrix
Structure is in order to next Mathematical treatment.Extended matrix M (t) is completely contained in power subsystem in the property value of moment t.
The monitoring data transition probability of step 2 power equipment
This step is divided into 3 sub-steps
The state demarcation of step 2.1 data
The value of each attribute of power equipment has a certain range normal operation in normal domain, when a certain property value exceeds its normal operation in normal domain, claims
There is exceptional value in the attribute.For discrete type property value, domain is denumerable discrete point.For continuous type property value,
Its domain is continuous section.According to specific power equipment, the different data values monitored can be defined according to it
Domain is divided into different states.When equipment is in normal condition, the state of monitoring data is known as stable state.For discrete
Different points can be classified as one kind according to specific data attribute feature, form a state by type data value.It can also directly will be every
One point is considered as a state.For continuous data value, serial number section can be divided into segment by specific features, often
A segment is a state.
The effect of the step is that the data of equipment are described with discrete state, with monitoring data turning in the status
In-migration portrays the variation of data.
The state of step 2.2 equipment
Power equipment may operate in different states, and different equipment states represents the phase characteristic of equipment operation.
For example, the operating status of transformer can be divided into normal operation, hot operation, unit exception.Different equipment possesses different
Operating status, and electric power data it is inconsistent essence performance be the operating status of physical device and differing for monitoring data state
It causing, i.e., monitoring data cannot really reflect equipment practical operation situation, and the operating status of equipment is varied, and data mode
Type is more.A usual equipment state corresponds to the specific combination of volume of data state.In the past, find wherein inconsistent
Correspondence be complicated, and the technical approach taken of the present invention is to be not concerned with the states of equipment and data in itself, but logical
The variation for investigating these states is crossed, navigates to inconsistent correspondence.
When the equipment stable operation in electric system, all kinds of monitoring data values of equipment should also be stablized in a certain range
In, while it changes rule also with stability.When equipment, which runs out present condition, to be changed, a part of monitoring data will bigger
Probability deviate original state, into new state, so as to this stability regular before breaking.
The step establishes the channel that data mode transfer is described with probabilistic manner, i.e. the transfer of data mode can be by
It is described according to the mathematical form of probability distribution.
The state transition frequency of step 2.3 statistical history monitoring data
Define 4 data mode transition frequencies
Wherein, fijIt represents to monitoring of equipment Ni+NjData are transferred to the frequency of state j, N from state i after secondaryiExpression is in
The number of state i, NjRepresent the number in state j.
1 Bernoulli Jacob's law of large numbers of law
Wherein, P represents probability, and ε represents a positive number, and Bernoulli Jacob's law of large numbers shows by largely collecting power equipment
The state transfer data of monitoring data, the frequency of state transfer being calculated can convergence in (with)probability.The law shows to lead to
The mode of monitoring current-period data or direct statistical history data is crossed repeatedly to estimate the state transition probability of power equipment.
The multiple monitoring transfer matrix of step 3 power equipment
This step is divided into 4 sub-steps
The Markov property and time homogeneity of step 3.1 data transition probability
Since each monitoring data essence is the sampling on discrete time point, while according to the state demarcation of step 2.2
Method, monitoring data are also divided into discrete state, so the state transfer of electric power monitoring data is substantially a time
The all discrete random process with state.Due to monitoring every time, data state in which is only related to last state in which
(carrying out probability transfer by initial state of state where the last time), so this random process has Markov property, essence is
Markov Chain.To data each time monitoring sampling obtain result with for the first time monitor state in which be it is unrelated, this
Illustrating the state transition probability of data also has time homogeneity.This step establish state transition probability with it is markovian
System, the transfer matrix with Markov property can be built by illustrating the transition probability of monitoring of equipment data in theory.
The multiple transfer matrix of step 3.2 electric power data
Define 6 single state-transition matrixes
Wherein, E represents the single state-transition matrix of electric power data, element pijRepresent that data are primary from state i experience
The probability of state j is transferred to behind monitoring time interval.Its probability value can be come by method statistic described in step 2.3.It is single
Next state transfer matrix E be fully described after the primary monitoring of any electric power data experience stateful transfer probability distribution feelings
Condition.According to Qie Puman-kolmogorov equation
Transfer matrix of the electric power data after the transfer of arbitrary next state can be obtained,
By Qie Puman-kolmogorov equation it is found that need to only pass through the single shape of step 3.2 method construct electric power data
State transfer matrix, just can directly calculate it is m+n times arbitrary after state-transition matrix, E(n+m)The element representation data of middle i rows j row
It is monitored with original state i, after then undergoing m+n monitoring, data are in the probability of state j.It is obtained by this computational methods
The practical significance of multiple transition probability out refers to when power equipment is in stable state, the theoretical shape that data should reach
State transition probability.
Residing theoretical probability distribution after step 3.4 time transfer
Define 7 electric power data initial distributions
Wherein, the initial distribution vector of φ (0) table monitoring registration evidence, element(0≤j≤n) represents data at the beginning
Carve each shape probability of state residing for 0.According to Qie Puman-kolmogorov equation, the theory at arbitrary m+n moment can be obtained
Probability distribution.
The theoretical probability distribution of more than 1 transfering state of theorem
φ (m+n)=φ (0) Em+n
Wherein, residing theoretical probability distribution after the m+n transfer of electric power data experience that φ (m+n) expressions calculate
Step 4 distortion data is positioned measures with subsystem distortion level
Equipment refers to the change of equipment non-generating state within the period in certain period of time in stable state, right
The probability distribution over states for the monitoring data answered has reflected equipment due data mode conceptual schema feature in this state.One
When denier power communication system detects that the state of current power equipment does not meet the distribution characteristics of corresponding state transition probability, then
The system of may infer that has been likely to occur monitoring data distortion.In monitoring cycle, if equipment running status does not change, and phase
Data mode is answered to change, such case is referred to as first kind distortion.If equipment running status changes, and data mode
Respective change does not occur, such case is referred to as the distortion of the second class.This step is divided into 5 sub-steps
The irrelevance of step 4.1 electric power data state transfer distribution
Define the irrelevance of 9 electric power data states transfer distribution
δ ＞ v (m+n)=| | φ ' (m+n)-φ (m+n) | |2
Wherein v (m+n) represents the irrelevance of electric power data state transfer distribution, and φ ' (m+n) is power equipment in experience m+
The practical transfering probability distribution counted after n times monitoring, φ (m+n) are that the theory that power equipment is undergone after m+n monitoring is general
Rate is distributed, and passes through the irrelevance of both distributions of the 2- norm measures of vector.δ ＞ 0, which are one, to be defined according to actual conditions
Deviation threshold value, it is determined by the intrinsic state transfer characteristic of the reality of equipment in subsystem.
As δ ＞ v (m+n), illustrate the virtual condition transfer distribution departure degree of device data in the reasonable scope.
As δ≤v (m+n), illustrate that the actual transfer Deviation of device data exceeds threshold value.
When equipment stable operation, there is δ≤v (m+n), show that first kind distortion sign occurs in device data, because of number
It has been appeared according at certain moment with greater probability in the state seldom occurred.When changing in equipment running process, go out
Existing δ ＞ v (m+n) show that the second class distortion sign occurs in equipment, because of the variation of the operating status with equipment, monitoring data
Corresponding state variation ought to occur, and virtual condition distribution does not vary widely.
The positioning of distortion data in step 4.2 subsystem
Define the monitoring data departure matrix of 10 subsystems
Power grid subsystem is the organic whole for including system relevant device, can quick positioning subsystem by departure matrix
The data of middle distortion.
Wherein, V (t) represents that the monitoring data of subsystem undergo the departure matrix after t normal period, element vij(t) table
Show the irrelevance of the data mode transfer distribution of j-th of attribute of i-th of equipment in subsystem after undergoing t normal period.Root
According to the property of extended matrix in step 2.1, if the attribute, command element value is not present in the corresponding equipment in matrix corresponding position
It is -1.
Step 4.3 is positioned into line distortion, based on the following formula：
Define 11 distortion positional matrixs
Wherein,Represent the distortion positional matrix of a n × k, element is only made of numerical value 0,1 and -1, F representing matrixes
0-1 transforming function transformation functions, its function are that element in matrix is carried out 0-1 transformation, and Δ (t) represents to deviate threshold matrix, element δij(t)
It represents to correspond to irrelevance vij(t) deviation threshold value, function f (zij∈ V (t)-Δ (t)) it represents to carry out 0-1 changes to its independent variable
It changes, is defined as follows and defines 11 element 0-1 transforming function transformation functions
The practical significance of distortion positional matrix is the mark that all rubidium markings beyond threshold value are less than threshold value for 1
0 is denoted as, the position of element in a matrix corresponds the position in data in subsystem again.Such a distortion positional matrix
Just contain the variable condition and location information of distortion data in a subsystem.
The processing of step 4.4 missing data
Shortage of data is a kind of performance of also data distortion, regardless of the true operating status of equipment, as long as monitoring number
According to there is data null value, then it is assumed that be data distortion, perform missing data transformation at this time
Wherein, g (dij) represent missing transforming function transformation function, it by M (t) the element of promising null value be mapped as 1, and enterIt is mapped as when assigning null data lacks transforming function transformation function againIn 1 element when, show its distortion attribute.Completely performing
After primary distortion positioning, need to be updated state transition frequency using the initial data of epicycle as historical data.Thus
It can guarantee that the probability element of single state-transition matrix constructed in step 3.2 has real-time accuracy.
Step 4.5 data distortion positions, and is measured with power subsystem distortion level.
Traversal distortion positional matrix it is primary, record it is each be equal to 1 element institute in a matrix position corresponded one
Distortion data position, rower represent device number.List is for the attribute number of the equipment.Taking out element accounts for all elements simultaneously
Ratio quantitative response power subsystem distortion level.
It should be noted that：It is proved involved in step 3.1 to geneva
It proves：Electric power data state migration procedure { Xn, n=0,1,2 ... } it is Markov Chain
Due to n be it is limited arrange, andAnd state i, j, i0,i1,…,in-1There is always conditional probability P (Xn+1
=j | X0=i0,X1=i1,…,Xn-1=in-1,Xn=i) so that P (Xn+1=j | X0=i0,X1=i1,…,Xn-1=in-1,Xn=
I)=P (Xn+1=j | Xn=i)
I.e. sampling process meets geneva, { Xn, n=0,1,2 ... } it is a Markov chain.
Therefore, the invention has the advantages that：1st, data acquisition rank of the present invention can be by various data formats or data structure
State transition probability is unified into, therefore evaded different data format in multi-source heterogeneous data to influence caused by data analysis,
Reduce the complexity of analysis system.2nd, the distortion of data differentiates and can dynamically carry out in real time, as long as theoretically giving certain
As soon as the data mode initial distribution at moment can calculate data distortion situation at this time.The data being distorted simultaneously can be returned again
Return update historical data.3rd, data distortion differentiates has simultaneity, after data distortion is determined, the position of distortion with positioning
It is determined simultaneously according to distortion positional matrix, the emergency repair time of equipment fault can be shortened.
Description of the drawings
Attached drawing 1 is the overall flow schematic diagram of the present invention.
Specific embodiment
Step 1 monitoring data attribute entities divide
Monitoring data acquisition is essentially referred in electric power network system, and various kinds of sensors is monitored equipment and will monitoring
Data are transmitted to the process of designated position storage.Monitoring data have a variety of storage mechanisms, this step in different subsystems
Purpose be so that collected data are divided into solid data set according to the entity device of data source.This step is divided into 3
Sub-step
Step 1.1 primary monitoring data acquires
Define 1 original data set
D={ d11,d12,...,d21,d22,...,dn1,dn2,...,dnk}
Wherein, in D expressions system collected all kinds of initial data attribute set, dijRepresent the kth category of the i-th equipment
Property, in actual information acquisition, since the data acquisition modes of each subsystem of power grid are not quite similar, so each subsystem is acquired
It is often mixed and disorderly data acquisition system after data summarization afterwards.
Step 1.2 monitoring data are classified by entity source address
In data acquisition, data are classified according to source index field, this step is divided into two kinds of forms.Form
One：The classification of gathered data
For shaped like collected data in defining 1, needing to classify according to the data source index field in data attribute,
One group will be divided into from the monitoring data of identical entity
Form two：Monitoring data are acquired by entity
For the subsystem of entity output monitoring data can be pressed, its monitoring data is directly acquired, and indicated in fact in data
The unique attribute of body.
Define 2 device datas vector
di=(di1,di2,...,dik)
Wherein, diExpression comes from the monitoring data vector of i-th (1≤i≤n) equipment, dijRepresent the jth of the i-th equipment, (1
≤ j≤k) attribute.The data sorted out in this way by the data source of arbitrary entity i equipment are in vector form by diRecord
Step 1.3 structure attribute extended matrix
In power communication system, electric power network includes each subsystem, and various kinds of equipment is by cooperating to support one
The normal operation of a subsystem, the data variation of set of subsystem can be formalized by attribute extension matrix.
Define 3 attribute extension matrixes
Wherein, M (t) represents the attribute extension matrix of t moment, dij(t) jth of i-th (1≤i≤n) equipment, (1≤j are represented
≤ki≤ k) monitor value of the attribute in moment t.The upper bound k of number of attributes, wherein k=that equipment possesses defined in above formula
max(k1,k2...) and represent the attribute number for possessing most attribute equipment in subsystem, it defines the columns of extended matrix.Electricity
The attribute number of distinct device can be different in power subsystem, and most row vectors in such extended matrix M (t) are not fixed
The property value of justice, such undefined attribute is referred to as extended attribute, their effect is to keep the rectangle knot of extended matrix
Structure is in order to next Mathematical treatment.Extended matrix M (t) is completely contained in power subsystem in the property value of moment t.
The monitoring data transition probability of step 2 power equipment
This step is divided into 3 sub-steps
The state demarcation of step 2.1 data
The value of each attribute of power equipment has a certain range normal operation in normal domain, when a certain property value exceeds its normal operation in normal domain, claims
There is exceptional value in the attribute.For discrete type property value, domain is denumerable discrete point.For continuous type property value,
Its domain is continuous section.According to specific power equipment, the different data values monitored can be defined according to it
Domain is divided into different states.When equipment is in normal condition, the state of monitoring data is known as stable state.For discrete
Different points can be classified as one kind according to specific data attribute feature, form a state by type data value.It can also directly will be every
One point is considered as a state.For continuous data value, serial number section can be divided into segment by specific features, often
A segment is a state.
The effect of the step is that the data of equipment are described with discrete state, with monitoring data turning in the status
In-migration portrays the variation of data.
The state of step 2.2 equipment
Power equipment may operate in different states, and different equipment states represents the phase characteristic of equipment operation.
For example, the operating status of transformer can be divided into normal operation, hot operation, unit exception.Different equipment possesses different
Operating status, and electric power data it is inconsistent essence performance be the operating status of physical device and differing for monitoring data state
It causing, i.e., monitoring data cannot really reflect equipment practical operation situation, and the operating status of equipment is varied, and data mode
Type is more.A usual equipment state corresponds to the specific combination of volume of data state.In the past, find wherein inconsistent
Correspondence be complicated, and the technical approach taken of the present invention is to be not concerned with the states of equipment and data in itself, but logical
The variation for investigating these states is crossed, navigates to inconsistent correspondence.
When the equipment stable operation in electric system, all kinds of monitoring data values of equipment should also be stablized in a certain range
In, while it changes rule also with stability.When equipment, which runs out present condition, to be changed, a part of monitoring data will bigger
Probability deviate original state, into new state, so as to this stability regular before breaking.
The step establishes the channel that data mode transfer is described with probabilistic manner, i.e. the transfer of data mode can be by
It is described according to the mathematical form of probability distribution.
The state transition frequency of step 2.3 statistical history monitoring data
Define 4 data mode transition frequencies
Wherein, fijIt represents to monitoring of equipment Ni+NjData are transferred to the frequency of state j, N from state i after secondaryiExpression is in
The number of state i, NjRepresent the number in state j.
1 Bernoulli Jacob's law of large numbers of law
Wherein, P represents probability, and ε represents a positive number, and Bernoulli Jacob's law of large numbers shows by largely collecting power equipment
The state transfer data of monitoring data, the frequency of state transfer being calculated can convergence in (with)probability.The law shows to lead to
The mode of monitoring current-period data or direct statistical history data is crossed repeatedly to estimate the state transition probability of power equipment.
The multiple monitoring transfer matrix of step 3 power equipment
This step is divided into 4 sub-steps
The Markov property and time homogeneity of step 3.1 data transition probability
Since each monitoring data essence is the sampling on discrete time point, while according to the state demarcation of step 2.2
Method, monitoring data are also divided into discrete state, so the state transfer of electric power monitoring data is substantially a time
The all discrete random process with state.Due to monitoring every time, data state in which is only related to last state in which
(carrying out probability transfer by initial state of state where the last time), so this random process has Markov property, essence is
Markov Chain.To data each time monitoring sampling obtain result with for the first time monitor state in which be it is unrelated, this
Illustrating the state transition probability of data also has time homogeneity.This step establish state transition probability with it is markovian
System, the transfer matrix with Markov property can be built by illustrating the transition probability of monitoring of equipment data in theory.
The multiple transfer matrix of step 3.2 electric power data
Define 6 single state-transition matrixes
Wherein, E represents the single state-transition matrix of electric power data, element pijRepresent that data are primary from state i experience
The probability of state j is transferred to behind monitoring time interval.Its probability value can be come by method statistic described in step 2.3.It is single
Next state transfer matrix E be fully described after the primary monitoring of any electric power data experience stateful transfer probability distribution feelings
Condition.According to Qie Puman-kolmogorov equation
Transfer matrix of the electric power data after the transfer of arbitrary next state can be obtained,
By Qie Puman-kolmogorov equation it is found that need to only pass through the single shape of step 3.2 method construct electric power data
State transfer matrix, just can directly calculate it is m+n times arbitrary after state-transition matrix, E(n+m)The element representation data of middle i rows j row
It is monitored with original state i, after then undergoing m+n monitoring, data are in the probability of state j.It is obtained by this computational methods
The practical significance of multiple transition probability out refers to when power equipment is in stable state, the theoretical shape that data should reach
State transition probability.
Residing theoretical probability distribution after step 3.4 time transfer
Define 7 electric power data initial distributions
Wherein, the initial distribution vector of φ (0) table monitoring registration evidence, element(0≤j≤n) represents data at the beginning
Carve each shape probability of state residing for 0.According to Qie Puman-kolmogorov equation, the theory at arbitrary m+n moment can be obtained
Probability distribution.
The theoretical probability distribution of more than 1 transfering state of theorem
φ (m+n)=φ (0) Em+n
Wherein, residing theoretical probability distribution after the m+n transfer of electric power data experience that φ (m+n) expressions calculate.
Step 4 distortion data positions
Equipment refers to the change of equipment non-generating state within the period in certain period of time in stable state, right
The probability distribution over states for the monitoring data answered has reflected equipment due data mode conceptual schema feature in this state.One
When denier power communication system detects that the state of current power equipment does not meet the distribution characteristics of corresponding state transition probability, then
The system of may infer that has been likely to occur monitoring data distortion.In monitoring cycle, if equipment running status does not change, and phase
Data mode is answered to change, such case is referred to as first kind distortion.If equipment running status changes, and data mode
Respective change does not occur, such case is referred to as the distortion of the second class.This step is divided into 5 sub-steps
The irrelevance of step 4.1 electric power data state transfer distribution
Define the irrelevance of 9 electric power data states transfer distribution
δ ＞ v (m+n)=| | φ ' (m+n)-φ (m+n) | |2
Wherein v (m+n) represents the irrelevance of electric power data state transfer distribution, and φ ' (m+n) is power equipment in experience m+
The practical transfering probability distribution counted after n times monitoring, φ (m+n) are that the theory that power equipment is undergone after m+n monitoring is general
Rate is distributed, and passes through the irrelevance of both distributions of the 2- norm measures of vector.δ ＞ 0, which are one, to be defined according to actual conditions
Deviation threshold value, as δ ＞ v (m+n), illustrate device data virtual condition transfer distribution departure degree in the reasonable scope,
As δ≤v (m+n), illustrate that the actual transfer Deviation of device data exceeds threshold value.
When equipment stable operation, there is δ≤v (m+n), show that first kind distortion sign occurs in device data, because of number
It has been appeared according at certain moment with greater probability in the state seldom occurred.When changing in equipment running process, go out
Existing δ ＞ v (m+n) show that the second class distortion sign occurs in equipment, because of the variation of the operating status with equipment, monitoring data
Corresponding state variation ought to occur, and virtual condition distribution does not vary widely.
The positioning of distortion data in step 4.2 subsystem
Define the monitoring data departure matrix of 10 subsystems
Power grid subsystem is the organic whole for including system relevant device, can quick positioning subsystem by departure matrix
The data of middle distortion.
Wherein, V (t) represents that the monitoring data of subsystem undergo the departure matrix after t normal period, element vij(t) table
Show the irrelevance of the data mode transfer distribution of j-th of attribute of i-th of equipment in subsystem after undergoing t normal period.Root
According to the property of extended matrix in step 2.1, if the attribute, command element value is not present in the corresponding equipment in matrix corresponding position
It is -1.
Step 4.3
Define 11 distortion positional matrixs
Wherein,Represent the distortion positional matrix of a n × k, element is only made of numerical value 0,1 and -1, F representing matrixes
0-1 transforming function transformation functions, its function are that element in matrix is carried out 0-1 transformation, and Δ (t) represents to deviate threshold matrix, element δij(t)
It represents to correspond to irrelevance vij(t) deviation threshold value, function f (zij∈ V (t)-Δ (t)) it represents to carry out 0-1 changes to its independent variable
It changes, is defined as follows and defines 11 element 0-1 transforming function transformation functions
The practical significance of distortion positional matrix is the mark that all rubidium markings beyond threshold value are less than threshold value for 1
0 is denoted as, the position of element in a matrix corresponds the position in data in subsystem again.Such a distortion positional matrix
Just contain the variable condition and location information of distortion data in a subsystem.
The processing of step 4.4 missing data
Shortage of data is a kind of performance of also data distortion, regardless of the true operating status of equipment, as long as monitoring number
According to there is data null value, then it is assumed that be data distortion, perform missing data transformation at this time
Wherein, g (dij) represent missing transforming function transformation function, it by M (t) the element of promising null value be mapped as 1, and enterIt is mapped as when assigning null data lacks transforming function transformation function againIn 1 element when, show its distortion attribute.Completely performing
After primary distortion positioning, need to be updated state transition frequency using the initial data of epicycle as historical data.Thus
It can guarantee that the probability element of single state-transition matrix constructed in step 3.2 has real-time accuracy.
Step 4.5 data distortion positions, and is measured with power subsystem distortion level.
Traversal distortion positional matrix it is primary, record it is each be equal to 1 element institute in a matrix position corresponded one
Distortion data position, rower represent device number.List is for the attribute number of the equipment.Taking out element accounts for all elements simultaneously
Ratio quantitative response power subsystem distortion level.
Specific embodiment described herein is only an example for the spirit of the invention.Technology belonging to the present invention is led
The technical staff in domain can do various modifications or additions to described specific embodiment or replace in a similar way
In generation, however, it does not deviate from the spirit of the invention or beyond the scope of the appended claims.
Claims (1)
1. a kind of method for rapidly positioning based on state transition probability power system monitoring data distortion, which is characterized in that including
Following steps：
Step 1, monitoring data attribute entities divide, the primary monitoring data of collecting device, and by collected raw monitored number
Solid data set is divided into according to the entity device according to data source, is specifically included：
Step 1.1, primary monitoring data acquisition, and collected raw sensor data is attributed to according to defined below in set：
Define 1, original data set
D={ d11,d12,...,d21,d22,...,dn1,dn2,...,dnk}
Wherein, in D expressions system collected all kinds of initial data attribute set, dijRepresent the kth attribute of the i-th equipment,
In actual information acquisition, since the data acquisition modes of each subsystem of power grid are not quite similar, so after each subsystem is acquired
It is often mixed and disorderly data acquisition system after data summarization；
Step 1.2 classifies the primary monitoring data of acquisition by entity source address, in data acquisition, by data according to
Source index field is classified, and is classified and following form is selected to classify according to the type of data：
Classification form one：Classification for gathered data for collected data in defining 1, is needed according to data category
Property in data source index field classification, one group will be divided into from the monitoring data of identical entity
Classification form two：By entity acquisition monitoring data classification, for the subsystem of entity output monitoring data can be pressed, directly adopt
Collect its monitoring data, and the unique attribute of entity is indicated in data；
Define 2 device datas vector
di=(di1,di2,...,dik)
Wherein, diExpression comes from the monitoring data vector of i-th (1≤i≤n) equipment, dijRepresent the jth of the i-th equipment, (1≤j
≤ k) attribute；The data sorted out by the data source of arbitrary entity device i are in vector form by diRecord
Step 1.3, structure attribute extended matrix：In power communication system, electric power network includes each subsystem, various kinds of equipment
By cooperating to the normal operation of one subsystem of support, the data variation of set of subsystem can pass through attribute extension square
Battle array formalizes, and wherein attribute extension matrix is based on defined below
Define 3 attribute extension matrixes
Wherein, M (t) represents the attribute extension matrix of t moment, dij(t) jth of i-th (1≤i≤n) equipment, (1≤j≤k are representedi
≤ k) monitor value of the attribute in moment t；The upper bound k of number of attributes, wherein k=max that equipment possesses defined in above formula
(k1,k2...) and represent the attribute number for possessing most attribute equipment in subsystem, it defines the columns of extended matrix；Electric power
The attribute number of distinct device can be different in subsystem, and most row vectors in such extended matrix M (t) do not define
Property value, such undefined attribute is referred to as extended attribute, their effect be keep extended matrix rectangular configuration
In order to next Mathematical treatment；Extended matrix M (t) is completely contained in power subsystem in the property value of moment t；
Step 2, the monitoring data transition probability for obtaining power equipment, specifically include：
The state demarcation of step 2.1 data
The value of each attribute of power equipment has a certain range normal operation in normal domain, when a certain property value exceeds its normal operation in normal domain, claims the category
There is exceptional value in property；For discrete type property value, domain is denumerable discrete point；For continuous type property value, determine
Adopted domain is continuous section；
According to specific power equipment, by the different data values monitored according to its domain, it is divided into different states；
When equipment is in normal condition, the state of monitoring data is known as stable state；
For discrete data value, can different points be classified as by one kind according to specific data attribute feature, form a state or
Each point is directly considered as a state by person；It is drawn for continuous data value or by serial number section by specific features
It is divided into segment, each segment is a state；
The state of step 2.2 equipment
Power equipment may operate in different states, and different equipment states represents the phase characteristic of equipment operation；It is different
Equipment possess different operating statuses, and the inconsistent essence performance of electric power data is operating status and the monitoring of physical device
Data mode it is inconsistent, i.e., monitoring data cannot really reflect equipment practical operation situation, and the operating status of equipment is a variety of more
Sample, and the type of data mode is more；One equipment state corresponds to the specific combination of volume of data state；
When the equipment stable operation in electric system, all kinds of monitoring data values of equipment should also be stablized in a certain range,
It, which changes rule, simultaneously also has stability；When equipment, which runs out present condition, to be changed, a part of monitoring data will bigger
Probability deviates original state, into new state, so as to this stability regular before breaking；
The state transition frequency of step 2.3, statistical history monitoring data, based on acquisition defined below：
Define 4 data mode transition frequencies
Wherein, fijIt represents to monitoring of equipment Ni+NjData are transferred to the frequency of state j, N from state i after secondaryiIt represents in state i
Number, NjRepresent the number in state j；
Then electricity is estimated by way of carrying out repeatedly monitoring current-period data or direct statistical history data based on the following formula
The state transition probability of power equipment
Wherein, P represents probability, and ε represents a positive number, and data are shifted by the state for largely collecting power equipment monitoring data,
The frequency of state transfer being calculated can convergence in (with)probability；It can be by repeatedly monitoring current-period data or direct statistical history number
According to mode estimate the state transition probability of power equipment；
Step 3, the multiple monitoring transfer matrix for obtaining power equipment, specifically include：
The Markov property and time homogeneity of step 3.1 data transition probability
Since each monitoring data essence is the sampling on discrete time point, while according to the state demarcation method of step 2.2,
Monitoring data are also divided into discrete state, so the state transfer of electric power monitoring data is substantially a time and state
All discrete random process；Due to monitoring every time, data state in which is only related to last state in which, so this
Random process has Markov property, and essence is Markov Chain；The result and the obtained to the sampling of monitoring each time of data
Primary monitoring state in which is unrelated, this illustrates that the state transition probability of data also has time homogeneity；
Step 3.2, the multiple transfer matrix for obtaining electric power data, based on defined below：
Define 6 single state-transition matrixes
Wherein, E represents the single state-transition matrix of electric power data, element pijRepresent that data undergo primary monitoring from state i
The probability of state j is transferred to after time interval；Its probability value can be come by method statistic described in step 2.3；Single shape
State transfer matrix E be fully described after the primary monitoring of any electric power data experience stateful transfer probability distribution situation；Root
According to Qie Puman-kolmogorov equation
Transfer matrix of the electric power data after the transfer of arbitrary next state is obtained,
By Qie Puman-kolmogorov equation it is found that need to only be turned by the single state of step 3.2 method construct electric power data
Move matrix, just can directly calculate it is m+n times arbitrary after state-transition matrix, E(n+m)The element representation data of middle i rows j row are with first
Beginning, state i was monitored, and after then undergoing m+n monitoring, data are in the probability of state j；It is drawn by this computational methods
The practical significance of multiple transition probability refer to that when power equipment is in stable state the theory state that data should reach turns
Move probability；
Residing theoretical probability distribution after step 3.4, acquisition time transfer, based on defined below：
Define 7 electric power data initial distributions
Wherein, the initial distribution vector of φ (0) table monitoring registration evidence, element(0≤j≤n) represents that data carve 0 institute at the beginning
Each shape probability of state at place；According to Qie Puman-kolmogorov equation, the theoretical probability at arbitrary m+n moment can be obtained
Distribution；
The theoretical probability distribution of more than 1 transfering state of theorem
φ (m+n)=φ (0) Em+n
Wherein, residing theoretical probability distribution after the m+n transfer of electric power data experience that φ (m+n) expressions calculate
Step 4 obtains distortion data positioning and subsystem distortion level measurement, specifically includes：
Step 4.1, the irrelevance for calculating the transfer distribution of electric power data state, based on defined below：
Define the irrelevance of 9 electric power data states transfer distribution
δ ＞ v (m+n)=| | φ ' (m+n)-φ (m+n) | |2
Wherein v (m+n) represents the irrelevance of electric power data state transfer distribution, and φ ' (m+n) is that power equipment is being undergone m+n times
The practical transfering probability distribution counted after monitoring, φ (m+n) are that power equipment undergoes the theoretical probability point after m+n monitoring
Cloth passes through the irrelevance of both distributions of the 2- norm measures of vector；δ ＞ 0 be one can be defined according to actual conditions it is inclined
From threshold value, as δ ＞ v (m+n), illustrate the virtual condition transfer distribution departure degree of device data in the reasonable scope, when δ≤
During v (m+n), illustrate that the actual transfer Deviation of device data exceeds threshold value；
When equipment stable operation, there is δ≤v (m+n), show that first kind distortion sign occurs in device data, because data exist
Certain moment have been appeared in greater probability in the state seldom occurred；When changing in equipment running process, there are δ ＞
V (m+n) shows that the second class distortion sign occurs in equipment, because of the variation of the operating status with equipment, monitoring data ought to be sent out
Raw corresponding state variation, and virtual condition distribution does not vary widely；
Step 4.2, the positioning for carrying out distortion data in subsystem, based on defined below：
Define the monitoring data departure matrix of 10 subsystems
Power grid subsystem is the organic whole for including system relevant device, can quickly be lost in positioning subsystem by departure matrix
Genuine data；
Wherein, V (t) represents that the monitoring data of subsystem undergo the departure matrix after t normal period, element vij(t) warp is represented
The irrelevance that the data mode transfer of j-th of attribute of i-th of equipment is distributed in subsystem after going through t normal period；According to step
The property of extended matrix in rapid 2.1, if the corresponding equipment in matrix corresponding position be not present the attribute, command element value for-
1；
Step 4.3 is positioned into line distortion, based on the following formula：
Define 11 distortion positional matrixs
Wherein,Represent the distortion positional matrix of a n × k, element is only made of numerical value 0,1 and -1, and F representing matrixes 0-1 becomes
Exchange the letters number, its function are that element in matrix is carried out 0-1 transformation, and Δ (t) represents to deviate threshold matrix, element δij(t) it represents
Corresponding to irrelevance vij(t) deviation threshold value, function f (zij∈ V (t)-Δ (t)) it represents to carry out 0-1 transformation to its independent variable,
It is defined as follows
Define 11 element 0-1 transforming function transformation functions
Distortion positional matrix practical significance be all rubidium markings beyond threshold value for 1 be less than threshold value label be
0, the position of element in a matrix corresponds the position in data in subsystem again；Such a distortion positional matrix just wraps
The variable condition and location information of distortion data in a subsystem are contained；
Step 4.4, the processing for carrying out missing data specifically perform missing data transformation, based on the following formula
Wherein, g (dij) represent missing transforming function transformation function, it by M (t) the element of promising null value be mapped as 1, and enter
It is mapped as when assigning null data lacks transforming function transformation function againIn 1 element when, show its distortion attribute；It is performed once complete
After distortion positioning, need to be updated state transition frequency using the initial data of epicycle as historical data；It can thus protect
Card
The probability element of single state-transition matrix constructed in step 3.2 has real-time accuracy；
Step 4.5 carries out data distortion positioning, and calculates and measured with power subsystem distortion level, specifically：Traversal distortion is fixed
Bit matrix is primary, record it is each be equal to 1 element institute in a matrix position corresponded a distortion data position,
Rower represents device number；List is for the attribute number of the equipment；The ratio quantitative response electricity that element accounts for all elements is taken out simultaneously
Power subsystem distortion level.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711340903.4A CN108132423B (en) | 2017-12-14 | 2017-12-14 | A kind of method for rapidly positioning based on state transition probability power system monitoring data distortion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711340903.4A CN108132423B (en) | 2017-12-14 | 2017-12-14 | A kind of method for rapidly positioning based on state transition probability power system monitoring data distortion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108132423A true CN108132423A (en) | 2018-06-08 |
| CN108132423B CN108132423B (en) | 2019-11-22 |
Family
ID=62389414
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711340903.4A Active CN108132423B (en) | 2017-12-14 | 2017-12-14 | A kind of method for rapidly positioning based on state transition probability power system monitoring data distortion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108132423B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111008238A (en) * | 2019-11-15 | 2020-04-14 | 武汉楚誉科技股份有限公司 | Key mode automatic positioning and early warning method based on associated evolution big data |
| CN112256801A (en) * | 2020-10-10 | 2021-01-22 | 深圳力维智联技术有限公司 | Method, system and storage medium for extracting key entities in entity relationship graph |
| CN113515158A (en) * | 2021-09-13 | 2021-10-19 | 常州旭泰克系统科技有限公司 | Equipment state monitoring method based on probability hybrid finite state machine |
| CN114756722A (en) * | 2022-04-25 | 2022-07-15 | 珠海市鸿瑞信息技术股份有限公司 | Multi-data centralized management control system and method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101674482A (en) * | 2009-09-25 | 2010-03-17 | 上海大学 | Method for optimized dispatching of extension type video flow in partially observational Markovian decision process |
| KR101171027B1 (en) * | 2011-09-20 | 2012-08-08 | 김일동 | An on-line monitoring system for power facilities and its method using circuit parameter measuring |
| US8745641B1 (en) * | 2011-07-14 | 2014-06-03 | Google Inc. | Automatic verification and anomaly detection in a representational state transfer (REST) application programming interface |
| CN103888849A (en) * | 2014-04-11 | 2014-06-25 | 北京工业大学 | Computing and wireless resource cooperative dispatching method in mobile cloud video transmission |
| CN106529715A (en) * | 2016-10-25 | 2017-03-22 | 哈尔滨工业大学 | Aeroengine maintenance strategy optimization method based on Markoff decision process |
-
2017
- 2017-12-14 CN CN201711340903.4A patent/CN108132423B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101674482A (en) * | 2009-09-25 | 2010-03-17 | 上海大学 | Method for optimized dispatching of extension type video flow in partially observational Markovian decision process |
| US8745641B1 (en) * | 2011-07-14 | 2014-06-03 | Google Inc. | Automatic verification and anomaly detection in a representational state transfer (REST) application programming interface |
| KR101171027B1 (en) * | 2011-09-20 | 2012-08-08 | 김일동 | An on-line monitoring system for power facilities and its method using circuit parameter measuring |
| CN103888849A (en) * | 2014-04-11 | 2014-06-25 | 北京工业大学 | Computing and wireless resource cooperative dispatching method in mobile cloud video transmission |
| CN106529715A (en) * | 2016-10-25 | 2017-03-22 | 哈尔滨工业大学 | Aeroengine maintenance strategy optimization method based on Markoff decision process |
Non-Patent Citations (2)
| Title |
|---|
| 刘艳丽，余贻鑫: "大系统马尔可夫模型状态转移概率矩阵的快速形成方法", 《天津大学学报》 * |
| 王宁，李石君: "基于模拟退火算法和隐马尔可夫模型的文本信息抽取", 《微电子学与计算机》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111008238A (en) * | 2019-11-15 | 2020-04-14 | 武汉楚誉科技股份有限公司 | Key mode automatic positioning and early warning method based on associated evolution big data |
| CN111008238B (en) * | 2019-11-15 | 2023-10-10 | 武汉楚誉科技股份有限公司 | Key mode automatic positioning and early warning method based on associated evolution big data |
| CN112256801A (en) * | 2020-10-10 | 2021-01-22 | 深圳力维智联技术有限公司 | Method, system and storage medium for extracting key entities in entity relationship graph |
| CN112256801B (en) * | 2020-10-10 | 2024-04-09 | 深圳力维智联技术有限公司 | Method, system and storage medium for extracting key entity in entity relation diagram |
| CN113515158A (en) * | 2021-09-13 | 2021-10-19 | 常州旭泰克系统科技有限公司 | Equipment state monitoring method based on probability hybrid finite state machine |
| CN114756722A (en) * | 2022-04-25 | 2022-07-15 | 珠海市鸿瑞信息技术股份有限公司 | Multi-data centralized management control system and method |
| CN114756722B (en) * | 2022-04-25 | 2022-10-04 | 珠海市鸿瑞信息技术股份有限公司 | Multi-data centralized management control system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108132423B (en) | 2019-11-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108132423B (en) | A kind of method for rapidly positioning based on state transition probability power system monitoring data distortion | |
| Liu et al. | A hybrid forecasting method for wind power ramp based on orthogonal test and support vector machine (OT-SVM) | |
| CN109543901A (en) | Short-Term Load Forecasting Method based on information fusion convolutional neural networks model | |
| CN103439629B (en) | Fault Diagnosis of Distribution Network systems based on data grids | |
| CN108320043A (en) | A kind of distribution network equipment state diagnosis prediction method based on electric power big data | |
| CN103728507A (en) | Grid fault diagnosis method based on data mining | |
| CN104133143B (en) | A kind of Guangdong power system diagnostic system and method calculating platform based on Hadoop cloud | |
| CN107807860B (en) | Power failure analysis method and system based on matrix decomposition | |
| CN112688431A (en) | Power distribution network load overload visualization method and system based on big data | |
| CN106600145A (en) | Power distribution network reliability analysis method and device | |
| Suo et al. | Computer assistance analysis of power grid relay protection based on data mining | |
| CN109412155B (en) | Power distribution network power supply capacity evaluation method based on graph calculation | |
| Liu et al. | Guest editorial: Special issue on data-analytics for stability analysis, control, and situational awareness of power system with high-penetration of renewable energy | |
| CN111965442A (en) | Energy internet fault diagnosis method and device under digital twin environment | |
| CN103383649A (en) | Design model of complex virtual instrument system | |
| Ren et al. | Application of PSO-LSSVM and hybrid programming to fault diagnosis of refrigeration systems | |
| CN107016410B (en) | Power information acquisition system method for diagnosing faults and trouble-shooter | |
| Zhou et al. | Calculation method of the line loss rate in low-voltage transformer district based on PCA and k-means clustering and support vector machine | |
| CN114385403A (en) | Distributed cooperative fault diagnosis method based on double-layer knowledge graph framework | |
| Hu et al. | Bayesian averaging-enabled transfer learning method for probabilistic wind power forecasting of newly built wind farms | |
| Zhang et al. | Distributed Architecture of Power Grid Asset Management and Future Research Directions | |
| Jingyu et al. | Statistical analysis of distribution network fault information based on multi-source heterogeneous data mining | |
| CN113850017A (en) | System-level fault analysis system and method based on power flow change map | |
| Chen et al. | An analytical method of data mining on voltage sag based on gray target theory and cloud model | |
| CN109978715A (en) | User side distributed generation resource Data Reduction 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |