CN107609769A - A kind of intelligent distribution network fault early warning method based on failure gene table - Google Patents

A kind of intelligent distribution network fault early warning method based on failure gene table Download PDF

Info

Publication number
CN107609769A
CN107609769A CN201710796634.6A CN201710796634A CN107609769A CN 107609769 A CN107609769 A CN 107609769A CN 201710796634 A CN201710796634 A CN 201710796634A CN 107609769 A CN107609769 A CN 107609769A
Authority
CN
China
Prior art keywords
msub
mrow
distribution network
mtr
mtd
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
Application number
CN201710796634.6A
Other languages
Chinese (zh)
Other versions
CN107609769B (en
Inventor
向敏
闵杰
屈琴芹
王在乾
高盼
陈诚
于祥春
许珑璋
孙永民
谭童
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University of Post and Telecommunications
Original Assignee
Chongqing University of Post and Telecommunications
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chongqing University of Post and Telecommunications filed Critical Chongqing University of Post and Telecommunications
Priority to CN201710796634.6A priority Critical patent/CN107609769B/en
Publication of CN107609769A publication Critical patent/CN107609769A/en
Application granted granted Critical
Publication of CN107609769B publication Critical patent/CN107609769B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present invention relates to a kind of intelligent distribution network fault early warning method based on failure gene table, belong to intelligent distribution network fault pre-alarming field.The running status of intelligent distribution network is divided into first excellent, it is good, neutralize poor four kinds of states, then state estimation and the failure with reference to corresponding to are carried out to the historical data of intelligent power distribution network operation using BP neural network algorithm, obtain the mapping relations between each section of state transfer time sequence and failure, so as to construct the gene table that is out of order, then the periodically online state transfer time sequence for obtaining intelligent distribution network, and it is matched with all genes in failure gene table by Smith Waterman algorithms, if maximum matching value reaches the threshold value of setting, the then corresponding failure of early warning.The problem of present invention can become increasingly complex in face of intelligent distribution network well, improve the accuracy rate of intelligent distribution network fault pre-alarming, guidance and help is provided for maintenance of the related management personnel to intelligent distribution network, effectively raises the science and foresight of operation of power networks decision-making.

Description

A kind of intelligent distribution network fault early warning method based on failure gene table
Technical field
The invention belongs to intelligent distribution network fault pre-alarming field, the running status for relating to realization to intelligent distribution network is whole A kind of intelligent distribution network fault early warning method based on failure gene table that body is held.
Background technology
In recent years, the whole world has started research and the construction upsurge of intelligent grid.Intelligent distribution network in intelligent grid as connecting The important component of major network and user oriented power supply is connect, the electric power whether its running status normally directly affects huge numbers of families supplies Should.Meanwhile with the access of distributed power source, the popularization of electric automobile and the increase of user interaction electric power so that power distribution network Dynamic behaviour becomes complicated, and operation risk greatly increases, once occur power distribution network power outage social life can all be caused it is huge Big influence and loss.Therefore, need badly and more in-depth study is carried out to the fault pre-alarming of intelligent distribution network, be related management people Maintenance of the member to intelligent distribution network provides guidance and help.
At present, domestic and foreign scholars propose various solutions from different angles for intelligent distribution network fault pre-alarming. Research shows power distribution network major part failure, and before destructive malfunction generation, power distribution network just comes into ill operation, has Gesture and cumulative effect.But current most solution is to utilize the local parameter such as harmonic current of power distribution network and short circuit electricity Stream, or component home such as transformer, or associated with external factor such as thundery sky gas phase, to reach to intelligent power distribution The effect of net fault pre-alarming.These solutions fail on the whole to hold the running status of intelligent distribution network, and In the scheme of early warning, current a certain factor is only considered so as to judge whether to need early warning, fails to make full use of power distribution network The tendency and cumulative effect of failure, so that in the intelligent distribution network in face of becoming increasingly complex, in the accuracy rate of fault pre-alarming A bit deficient in.Therefore, be badly in need of a kind of can realize at present integrally to hold the running status of intelligent distribution network, and can from current and Conventional running status come together decide on whether the method for early warning that will be broken down.
The content of the invention
In view of this, it is an object of the invention to provide a kind of intelligent distribution network fault pre-alarming side based on failure gene table Method, tendency and cumulative effect feature that intelligent distribution network failure has are made full use of, solve intelligent distribution network fault pre-alarming Problem, and improve the accuracy rate of its early warning.
To reach above-mentioned purpose, the present invention provides following technical scheme：
A kind of intelligent distribution network fault early warning method based on failure gene table, is specifically comprised the following steps：
S1：The running status of intelligent distribution network is divided, and builds intelligent distribution network State Assessment Index System；
S2：Structure includes the BP neural network model of input layer, hidden layer and output layer；
S3：Intelligent grid historical failure data is divided into two parts, to Part I intelligent distribution network historical failure data The service data of bus in source is clustered to build training sample；
S4：Distribution Running State corresponding to the training sample and training sample of Part I intelligent grid is input to BP neural network model is trained, and obtains BP neural network state estimation model；
S5：The service data of bus in Part II intelligent distribution network historical failure data source is input to BP nerve nets Network state estimation model obtains the mapping relations of intelligent power distribution net state transfer time sequence and failure, so as to build the base that is out of order Because of table；
S6：The running state data of the bus of current intelligent distribution network is periodically obtained, and running state data is inputted The state transfer time sequence of current intelligent grid is obtained to BP neural network state estimation model；
S7：By Simith-Waterman gene order alignment algorithms, the state of the current intelligent grid of acquisition is shifted Time series is matched with all genes in failure gene table, asks for matching value, by maximum matching value with it is set Threshold value is compared, if exceeding threshold value, the failure corresponding to the early warning gene.
Further, in step sl, the structure intelligent distribution network State Assessment Index System is specially：
Intelligent distribution network state estimation index is defined as to the state score of each bus of intelligent distribution network, the state of each bus The calculation formula of score is as follows：
Wherein GbFor bus state score, value is [0,1], represents that state is better closer to 1, state is represented closer to 0 It is poorer；λVRepresent the weights of busbar voltage, gVRepresent busbar voltage score, λPRepresent the weights of bus active power, gPRepresent female Line active power score, λQRepresent the weights of bus reactive power rate, gQRepresent bus reactive power rate score.
Further, the busbar voltage score calculation formula is as follows：
Wherein V is the magnitude of voltage of bus,The average value of corresponding busbar voltage, V are concentrated for historical datamaxFor historical data Concentrate the maximum of corresponding busbar voltage, VminThe minimum value of corresponding busbar voltage is concentrated for historical data.
Further, the bus active power score calculation formula is as follows：
Wherein P is the active power value of bus,The average value of corresponding bus active power, P are concentrated for historical datamaxFor Historical data concentrates the maximum of corresponding bus active power, PminThe minimum of corresponding bus active power is concentrated for historical data Value.
Further, the bus reactive power rate score calculation formula is as follows：
Wherein Q is the reactive power value of bus,The average value of corresponding bus reactive power rate, Q are concentrated for historical datamaxFor Historical data concentrates the maximum of corresponding bus reactive power rate, QminThe minimum of corresponding bus reactive power rate is concentrated for historical data Value.
Further, in step S3, the structure training sample is specially：By Part I intelligent distribution network historical data source In bus service data cluster for four class different conditions sample,
Wherein Tn×m(k) it is training sample matrix, k=1,2,3,4 represent different classes of training sample, and n represents training sample This quantity, m represent bus bar number, tnmRepresent the state score of the m articles bus of n-th of training sample.
Further, in step S4, the process of the training is specially：
S41：If the connection weight of input layer to hidden layer is wij, the connection weight of hidden layer to output layer is wj, input The threshold value of layer to hidden layer is γj, the threshold value of hidden layer to output layer isThe number of nodes of hidden layer is N, learning rate η, Desired output is
S42：Sigmoid type functions are selected as hidden layer activation primitive f1With output layer activation primitive f2；
S43：The input and output of hidden layer each unit are calculated, with the input t of input layernm, input layer to hidden layer company Connect weight wijWith input layer to hidden layer threshold value γj, the input h of calculating hidden layer each unitj, then use hjPass through activation primitive f1 Calculate the output b of hidden layer each unitj；
Wherein r=1,2 ..., n, represent that r-th of sample is trained to；
S44：According to the output b of hidden layerj, hidden layer to output layer connection weight wjWith the threshold of hidden layer to output layer ValueCalculate output result y；
S45：Calculation error；
ehj=wj×eo×bj×(1-bj)
Wherein e is output error, and eo is output layer vague generalization error, ehjFor hidden layer each unit vague generalization error；
S46：Hidden layer is adjusted to the connection weight w of output layerjAnd threshold value
Wherein w 'jWithFor the connection weight and threshold value of the hidden layer after adjustment to output layer；
S47：Input layer is adjusted to the connection weight w of hidden layerijAnd threshold gammaj；
Wherein w 'ijWith γ 'jFor the connection weight and threshold value of the input layer after adjustment to hidden layer；
S48：Variable r is got into n from 1, then all training samples have been trained to, then by the error of each training sample E accumulation calculatings go out global error E, judge whether E reaches in the range of specification error, if so, then terminating the current company of training record Connect weights and threshold value；If it is not, global error E is set into zero, and go to the training of step S43 repetitive learnings.
Further, the step S5 is specially：
S51：Evaluating matrix is built with Part II intelligent distribution network historical failure data source：
WhereinFor evaluating matrix, n*The quantity of assessment sample is represented,Represent n-th*The individual m for assessing sample The assessment score of bar bus；
S52：By evaluating matrixBP neural network state estimation model is inputted, draws each section of state of intelligent distribution network Mapping between transfer time sequence and failure, so as to build the gene table that is out of order.
Further, the step S7 is specially：
S71：Set gene order to be compared asLength is lS, the sequence in failure gene table ForLength is lU；
S72：Set the replacement matrix that gene order compares；
S73：Construction size is (lS+1)×(lU+ 1) matrix D of size, for depositing comparison result；
Wherein D (li,lj) represent gene order to be matched, D (li, 0) and=D (0, lj)=0,1≤li≤lS, 1≤lj≤ lU；Represent the l in gene order SiIndividual elementary state,Represent the l in gene order UjThe state of individual element；
S74：L is found out in matrix Di *And lj *, meet：
Wherein D (li *,lj *) represent sequence S and U high specific to score value；
S75：Until gene order to be compared and all gene orders of failure gene table are compared, try to achieve all High specific to score value, by high specific to score value compared with set threshold value, if exceeding threshold value, the early warning gene institute Corresponding failure.
Further, the Simith-Waterman gene orders alignment algorithm of the step S7 meets：
Wherein σ represents the importance of state.
The beneficial effects of the present invention are：Method proposed by the invention has merged BP neural network algorithm and Smith- Waterman gene order alignment algorithms, the fault signature of intelligent distribution network is fully excavated, be the fault pre-alarming of intelligent distribution network New thinking and solution are provided.This method has very strong versatility and applicability, for the intelligent power distribution of different scales Net, this method can establish different failure gene tables and correspond to therewith, and have what is mutually used for reference between different faults gene table Meaning.The problem of this method can become increasingly complex in face of intelligent distribution network well, the accuracy rate of fault pre-alarming is improved, for correlation Maintenance of the administrative staff to intelligent distribution network provides guidance and help, effectively raises the scientific and pre- of operation of power networks decision-making Opinion property.
Brief description of the drawings
In order that the purpose of the present invention, technical scheme and beneficial effect are clearer, the present invention provides drawings described below and carried out Explanation：
Fig. 1 is the structure schematic flow sheet of failure gene table of the present invention；
Fig. 2 is three layers of BP neural network illustraton of model of the present invention；
Fig. 3 is intended to for distribution network failure gene representation of the present invention；
Fig. 4 is the fault pre-alarming block diagram of the invention based on failure gene table；
Fig. 5 is overall flow schematic diagram of the present invention.
Embodiment
Below in conjunction with accompanying drawing, the preferred embodiments of the present invention are described in detail.
1. failure gene table builds the stage
Service data before the failure and failure that occur in view of intelligent distribution network has certain mapping relations, can based on this To regard the service data of failure for the previous period as the gene for characterizing the failure.Four in analogy construction human gene of the present invention Kind base, the running status of intelligent distribution network is divided into excellent, good, the poor four kinds of states of neutralization, then commented using BP neural network Estimate model, the service data of intelligent distribution network is changed into orderly state transfer time sequence, the as base of intelligent distribution network The input of cause, wherein BP neural network is the state score values of each bus of intelligent distribution network.Gone through from the failure of intelligent distribution network In history service data source, the mapping relations of each section of shape body transfer time sequence and failure can be obtained, so as to build the base that is out of order Because of table.
As shown in figure 1, the flow of failure gene table structure of the present invention is：
S1 determines the running status division of intelligent distribution network, and builds intelligent distribution network State Assessment Index System.
The running status of intelligent distribution network is divided into excellent, good, the poor four kinds of states of neutralization, and is denoted as E, G, M and B, such as Shown in table 1.
The running status division of the intelligent distribution network of table 1
The running status of intelligent distribution network Mark
It is excellent E
It is good G
In M
Difference B
The state estimation index of intelligent distribution network is defined as the state score of each bus of intelligent distribution network, the shape of each bus Shown in the calculation formula of state score such as formula (1)：
Wherein GbFor bus state score, value is [0,1], represents that state is better closer to 1, state is represented closer to 0 It is poorer；λVRepresent the weights of busbar voltage, gVRepresent busbar voltage score, λPRepresent the weights of bus active power, gPRepresent female Line active power score, λQRepresent the weights of bus reactive power rate, gQRepresent bus reactive power rate score.
Shown in the calculation formula of voltage score such as formula (2)：
Wherein V is the magnitude of voltage of bus,The average value of corresponding busbar voltage, V are concentrated for historical datamaxFor historical data Concentrate the maximum of corresponding busbar voltage, VminThe minimum value of corresponding busbar voltage is concentrated for historical data.
Shown in the calculation formula of active power score such as formula (3)：
Wherein P is the active power value of bus,The average value of corresponding bus active power, P are concentrated for historical datamaxFor Historical data concentrates the maximum of corresponding bus active power, PminThe minimum of corresponding bus active power is concentrated for historical data Value.
Shown in the calculation formula of reactive power score such as formula (4)：
Wherein Q is the reactive power value of bus,The average value of corresponding bus reactive power rate, Q are concentrated for historical datamax The maximum of corresponding bus reactive power rate, Q are concentrated for historical dataminCorresponding bus reactive power rate is concentrated for historical data most Small value.
S2 designs 3 layers of BP neural network model for including input layer, hidden layer and output layer, as shown in Figure 2.
The number of input layer is equal to the quantity of power distribution network bus, and the quantity of hidden layer can be according to the effect of Mathlab emulation Fruit chooses most suitable hidden layer node quantity, and output layer is the state score of power distribution network, and power distribution network is determined according to state score The state of operation, as shown in table 2.
Table 2BP neutral nets output result and state demarcation rule
Output result State demarcation
0≤y ＜ 0.25 Poor (B)
0.25≤y ＜ 0.5 In (M)
0.5≤y ＜ 0.75 Good (G)
0.75≤y≤1 Excellent (E)
S3 is clustered the service data of each bus of intelligent distribution network in partial history fault data source to build training Sample.
It is four class different conditions by the state score data clusters of all buses in intelligent distribution network historical data source Sample, as training sample matrix Tn×m(k), as shown in formula (5)：
Wherein Tn×m(k) it is training sample matrix, k=1,2,3,4 represent the training sample of different conditions classification, and n represents instruction Practice the quantity of sample, m represents bus bar number, tnmRepresent the state score of the m articles bus of n-th of training sample.
Training sample and corresponding distribution Running State are input in BP neural network and are trained by S4, obtain BP god Through network state assessment models.
By training sample matrix Tn×m(k) input BP neural network is trained, and determines the weights and threshold of each neuron Value, training process are as follows：
A. input layer is set to the connection weight of hidden layer as wij, the connection weight of hidden layer to output layer is wj, input layer Threshold value to hidden layer is γj, the threshold value of hidden layer to output layer isThe number of nodes of hidden layer is N, learning rate η, the phase Hope that output is
B. sigmoid type functions are selected as hidden layer activation primitive f1With output layer activation primitive f2, as shown in formula (6)：
C. the input and output of hidden layer each unit are calculated, with the input t of input layernm, input layer to hidden layer connection Weight wijWith input layer to hidden layer threshold value γj.Calculate the input h of hidden layer each unitj, then use hjPass through activation primitive f1Meter Calculate the output b of hidden layer each unitj, as shown in formula (7)：
Wherein r=1,2 ..., n, represent that r-th of sample is trained to.
D. according to the output b of hidden layerj, hidden layer to output layer connection weight wjWith the threshold value of hidden layer to output layerOutput result y is calculated, as shown in formula (8)：
E. calculation error, the error e of each training sample is obtained according to formula (9), output layer vague generalization error is calculated, such as formula (10) shown in, hidden layer each unit vague generalization error is calculated, as shown in formula (11)：
ehj=wj×eo×bj×(1-bj) (11)
Wherein e is output error, and eo is output layer vague generalization error, ehjFor hidden layer each unit vague generalization error.
F. hidden layer is adjusted to the connection weight w of output layerjAnd threshold valueAs shown in formula (12)：
Wherein w 'jWithFor the connection weight and threshold value of the hidden layer after adjustment to output layer.
G. input layer is adjusted to the connection weight w of hidden layerijAnd threshold gammaj, as shown in formula (13)：
Wherein w 'ijWith γ 'jFor the connection weight and threshold value of the input layer after adjustment to hidden layer.
H. variable r is got into n from 1, then all training samples have been trained to, then by the error e of each training sample Accumulation calculating goes out global error E, judges whether E reaches in the range of specification error, if so, then terminating the current connection of training record Weights and threshold value；If it is not, global error E is set into zero, and go to the training of step c repetitive learnings.
The service data of each bus in remaining historical failure data source is input in the BP neural network trained by S5 The mapping relations of state of electric distribution network transfer time sequence and failure are obtained, so as to build the gene table that is out of order.
Remaining historical failure data source is configured to evaluating matrixAs shown in formula (14)：
WhereinFor evaluating matrix, n*The quantity of assessment sample is represented,Represent n-th*The individual m for assessing sample The assessment score of bar bus.
By evaluating matrixInput the BP neural network trained to be assessed, draw each section of state transfer of power distribution network Mapping between time series and failure, so as to build the gene table that is out of order, as shown in Figure 3.
2. the fault pre-alarming stage based on failure gene table
After the failure gene table of intelligent distribution network is built, the operation number of intelligent distribution network is periodically obtained in real time According to gene being converted into by BP neural network, then by Smith-Waterman gene orders alignment algorithm by itself and failure Gene in gene table is matched, and asks for matching value, if matching value reaches the threshold value of setting, starting early warning will occur Corresponding failure.
For Smith-Waterman algorithms when the gene order in for biology compares, the significance level of four bases is phase With, therefore score design when substituting matrix for identical Mismatching is identical, when being applied in intelligent distribution network Fault pre-alarming field when, because the running status of intelligent distribution network is divided into four kinds, respectively E that performance successively decreases successively, G, M, B, therefore during gene order matches, the significance level of these four states is different.
In order to adapt to the online fault pre-alarming demand of intelligent distribution network, it is necessary to enter to traditional Smith-Waterman algorithms Row improves, and design needs to follow two principles when replacing matrix：
For two states of I in matching, the performance corresponding to two states is poorer, then the score obtained is higher；
When two states of II mismatch, performance difference is bigger between two states, then dividing for being detained is more.
First principle is enabled to when state matches, and the running status as much as possible for allowing poor performance matches, prominent Its significance level, improve the accuracy rate and speed of early warning；Second principle enables to, when state mismatches, reduce it to pre- The negative effect of alert accuracy rate and speed.
According to the significance level of each state and the design principle of replacement matrix, it can be deduced that substitute the design formula of matrix As shown in formula (15)：
Wherein, σ represents the importance of state,L in the S gene orders that expression is mentioned hereinafteriIndividual elementary state, L in the U gene orders that expression is mentioned hereinafterjThe state of individual element.
As shown in figure 4, fault pre-alarming of the present invention is divided into following steps：
S1 periodically obtains the running status score of each bus of current power distribution network, is input in BP neural network and obtains shape State transfer time sequence；
The service data of intelligent distribution network is periodically obtained in real time, is then inputted in BP neural network the state that is converted into and is turned Shift time sequence, gene as to be compared, and the length of the gene is progressively longer over time.
S2 is by Simith-Waterman gene order alignment algorithms, by the state transfer time sequence periodicity of acquisition Matched with the gene in failure gene table, ask for matching value.
Periodically gene to be compared and the gene in failure gene table are calculated by improved Smith-Waterman Method is matched, and matching flow is as follows：
A. set gene order to be compared asLength is lS, the sequence in failure gene table isLength is lU。
B. the replacement matrix that gene order compares is set, sets four kinds of states E, G, M and B of power distribution network importance difference For 1,2,3 and 4, it can calculate that to substitute matrix as shown in table 3 according to formula (15).
The replacement matrix that the gene order of table 3 compares
E G M B
E 1 -1 -2 -3
G -1 4 -1 -2
M -2 -1 9 -1
B -3 -2 -1 16
C. according to the method for Dynamic Programming, construction size is (lS+1)×(lU+ 1) matrix D of size, compared for depositing As a result, matrix D can lead to formula (16) and (17) are calculated.
D(li, 0) and=D (0, lj)=0 (16)
Wherein 1≤li≤lS, 1≤lj≤lU；It can be obtained according to formula (15),ForTake sky Point penalty during position,ForTake point penalty during room.
So D (li,lj) mean that gene order to be matchedWith the sequence in failure gene tableBetween all possible comparison score value.
D. l is found out in Dynamic Programming matrix Di *And lj *, meet：
Wherein D (li *,lj *) represent sequence S and U high specific to score value.
E. until gene order to be compared and all gene orders of failure gene table are compared, try to achieve all High specific to score value, by high specific to score value compared with set threshold value, if exceeding threshold value, early warning gene institute is right The failure answered.
During gene order compares, the setting of threshold value of warning is very crucial, and given threshold of the present invention is event Hinder the percent value of gene full marks, for failure gene different in gene table, although threshold value is identical, corresponding to threshold value Score it is different, therefore there is more preferable adaptability, choose different percent values and emulated, choosing makes early warning accuracy rate most Percent value when excellent is as threshold value of warning.
Finally illustrate, preferred embodiment above only to illustrate invention technical scheme and it is unrestricted, although passing through The present invention is described in detail for above preferred embodiment, it is to be understood by those skilled in the art that can be in shape Various changes are made in formula and to it in details, without departing from claims of the present invention limited range.

Claims (10)

  1. A kind of 1. intelligent distribution network fault early warning method based on failure gene table, it is characterised in that：Specifically comprise the following steps：
    S1：The running status of intelligent distribution network is divided, and builds intelligent distribution network State Assessment Index System；
    S2：Structure includes the BP neural network model of input layer, hidden layer and output layer；
    S3：Intelligent grid historical failure data is divided into two parts, in Part I intelligent distribution network historical failure data source The service data of bus clustered to build training sample；
    S4：Distribution Running State corresponding to the training sample and training sample of Part I intelligent grid is input to BP god It is trained through network model, obtains BP neural network state estimation model；
    S5：The service data of bus in Part II intelligent distribution network historical failure data source is input to BP neural network shape State assessment models obtain the mapping relations of intelligent power distribution net state transfer time sequence and failure, so as to build the gene that is out of order Table；
    S6：The running state data of the bus of current intelligent distribution network is periodically obtained, and running state data is input to BP Neutral net state estimation model obtains the state transfer time sequence of current intelligent grid；
    S7：By Simith-Waterman gene order alignment algorithms, by the state transfer time of the current intelligent grid of acquisition Sequence is matched with all genes in failure gene table, asks for matching value, by the matching value of maximum and set threshold value It is compared, if exceeding threshold value, the failure corresponding to the early warning gene.
  2. 2. a kind of intelligent distribution network fault early warning method based on failure gene table according to claim 1, its feature exist In：In step sl, the structure intelligent distribution network State Assessment Index System is specially：
    Intelligent distribution network state estimation index is defined as to the state score of each bus of intelligent distribution network, the state score of each bus Calculation formula it is as follows：
    <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>G</mi> <mi>b</mi> </msub> <mo>=</mo> <msub> <mi>&amp;lambda;</mi> <mi>V</mi> </msub> <msub> <mi>g</mi> <mi>V</mi> </msub> <mo>+</mo> <msub> <mi>&amp;lambda;</mi> <mi>P</mi> </msub> <msub> <mi>g</mi> <mi>P</mi> </msub> <mo>+</mo> <msub> <mi>&amp;lambda;</mi> <mi>Q</mi> </msub> <msub> <mi>g</mi> <mi>Q</mi> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>&amp;lambda;</mi> <mi>V</mi> </msub> <mo>+</mo> <msub> <mi>&amp;lambda;</mi> <mi>P</mi> </msub> <mo>+</mo> <msub> <mi>&amp;lambda;</mi> <mi>Q</mi> </msub> <mo>=</mo> <mn>1</mn> </mrow> </mtd> </mtr> </mtable> </mfenced>
    Wherein GbFor bus state score, value is [0,1], represents that state is better closer to 1, represents that state is poorer closer to 0； λVRepresent the weights of busbar voltage, gVRepresent busbar voltage score, λPRepresent the weights of bus active power, gPRepresent that bus has Work(power score, λQRepresent the weights of bus reactive power rate, gQRepresent bus reactive power rate score.
  3. 3. a kind of intelligent distribution network fault early warning method based on failure gene table according to claim 2, its feature exist In：The busbar voltage score calculation formula is as follows：
    <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>g</mi> <mi>V</mi> </msub> <mo>=</mo> <mn>1</mn> <mo>-</mo> <mfrac> <mrow> <mi>V</mi> <mo>-</mo> <mover> <mi>V</mi> <mo>&amp;OverBar;</mo> </mover> </mrow> <mrow> <msub> <mi>V</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>V</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> </msub> </mrow> </mfrac> <mo>,</mo> <mi>i</mi> <mi>f</mi> <mi> </mi> <msub> <mi>V</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>&amp;NotEqual;</mo> <msub> <mi>V</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>g</mi> <mi>V</mi> </msub> <mo>=</mo> <mn>0</mn> <mo>,</mo> <mi>o</mi> <mi>t</mi> <mi>h</mi> <mi>e</mi> <mi>r</mi> <mi>w</mi> <mi>i</mi> <mi>s</mi> <mi>e</mi> </mrow> </mtd> </mtr> </mtable> </mfenced>
    Wherein V is the magnitude of voltage of bus,The average value of corresponding busbar voltage, V are concentrated for historical datamaxConcentrated for historical data The maximum of corresponding busbar voltage, VminThe minimum value of corresponding busbar voltage is concentrated for historical data.
  4. 4. a kind of intelligent distribution network fault early warning method based on failure gene table according to claim 2, its feature exist In：The bus active power score calculation formula is as follows：
    <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>g</mi> <mi>P</mi> </msub> <mo>=</mo> <mn>1</mn> <mo>-</mo> <mfrac> <mrow> <mi>P</mi> <mo>-</mo> <mover> <mi>P</mi> <mo>&amp;OverBar;</mo> </mover> </mrow> <mrow> <msub> <mi>P</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>P</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> </msub> </mrow> </mfrac> <mo>,</mo> <mi>i</mi> <mi>f</mi> <mi> </mi> <msub> <mi>P</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>&amp;NotEqual;</mo> <msub> <mi>P</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>g</mi> <mi>P</mi> </msub> <mo>=</mo> <mn>0</mn> <mo>,</mo> <mi>o</mi> <mi>t</mi> <mi>h</mi> <mi>e</mi> <mi>r</mi> <mi>w</mi> <mi>i</mi> <mi>s</mi> <mi>e</mi> </mrow> </mtd> </mtr> </mtable> </mfenced>
    Wherein P is the active power value of bus,The average value of corresponding bus active power, P are concentrated for historical datamaxFor history The maximum of bus active power, P are corresponded in data setminThe minimum value of corresponding bus active power is concentrated for historical data.
  5. 5. a kind of intelligent distribution network fault early warning method based on failure gene table according to claim 2, its feature exist In：The bus reactive power rate score calculation formula is as follows：
    <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>g</mi> <mi>Q</mi> </msub> <mo>=</mo> <mn>1</mn> <mo>-</mo> <mfrac> <mrow> <mi>Q</mi> <mo>-</mo> <mover> <mi>Q</mi> <mo>&amp;OverBar;</mo> </mover> </mrow> <mrow> <msub> <mi>Q</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>Q</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> </msub> </mrow> </mfrac> <mo>,</mo> <mi>i</mi> <mi>f</mi> <mi> </mi> <msub> <mi>Q</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>&amp;NotEqual;</mo> <msub> <mi>Q</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>g</mi> <mi>Q</mi> </msub> <mo>=</mo> <mn>0</mn> <mo>,</mo> <mi>o</mi> <mi>t</mi> <mi>h</mi> <mi>e</mi> <mi>r</mi> <mi>w</mi> <mi>i</mi> <mi>s</mi> <mi>e</mi> </mrow> </mtd> </mtr> </mtable> </mfenced>
    Wherein Q is the reactive power value of bus,The average value of corresponding bus reactive power rate, Q are concentrated for historical datamaxFor history The maximum of bus reactive power rate, Q are corresponded in data setminThe minimum value of corresponding bus reactive power rate is concentrated for historical data.
  6. 6. a kind of intelligent distribution network fault early warning method based on failure gene table according to claim 2, its feature exist In：In step S3, the structure training sample is specially：By the fortune of the bus in Part I intelligent distribution network historical data source Row data clusters are the sample of four class different conditions,
    Wherein Tn×m(k) it is training sample matrix, k=1,2,3,4 represent different classes of training sample, and n represents training sample Quantity, m represent bus bar number, tnmRepresent the state score of the m articles bus of n-th of training sample.
  7. 7. a kind of intelligent distribution network fault early warning method based on failure gene table according to claim 6, its feature exist In：In step S4, the process of the training is specially：
    S41：If the connection weight of input layer to hidden layer is wij, the connection weight of hidden layer to output layer is wj, input layer arrives The threshold value of hidden layer is γj, the threshold value of hidden layer to output layer isThe number of nodes of hidden layer is N, learning rate η, it is expected Export and be
    S42：Sigmoid type functions are selected as hidden layer activation primitive f1With output layer activation primitive f2；
    <mrow> <msub> <mi>f</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <msub> <mi>f</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mrow> <mn>1</mn> <mo>+</mo> <msup> <mi>e</mi> <mrow> <mo>-</mo> <mi>x</mi> </mrow> </msup> </mrow> </mfrac> </mrow>
    S43：The input and output of hidden layer each unit are calculated, with the input t of input layernm, input layer to hidden layer connection weight Value wijWith input layer to hidden layer threshold value γj, the input h of calculating hidden layer each unitj, then use hjPass through activation primitive f1Calculate The output b of hidden layer each unitj；
    <mrow> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>h</mi> <mi>j</mi> </msub> <mo>=</mo> <munderover> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </munderover> <msub> <mi>w</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <msub> <mi>t</mi> <mrow> <mi>r</mi> <mi>i</mi> </mrow> </msub> <mo>+</mo> <msub> <mi>&amp;gamma;</mi> <mi>j</mi> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>b</mi> <mi>j</mi> </msub> <mo>=</mo> <msub> <mi>f</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>h</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>,</mo> <mrow> <mo>(</mo> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mo>...</mo> <mi>N</mi> <mo>)</mo> </mrow> </mrow>
    Wherein r=1,2 ..., n, represent that r-th of sample is trained to；
    S44：According to the output b of hidden layerj, hidden layer to output layer connection weight wjWith the threshold value of hidden layer to output layerMeter Calculate output result y；
    S45：Calculation error；
    <mrow> <mi>e</mi> <mo>=</mo> <mfrac> <mn>1</mn> <mn>2</mn> </mfrac> <msup> <mrow> <mo>(</mo> <mi>y</mi> <mo>-</mo> <mover> <mi>y</mi> <mo>^</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow>
    <mrow> <mi>e</mi> <mi>o</mi> <mo>=</mo> <mrow> <mo>(</mo> <mi>y</mi> <mo>-</mo> <mover> <mi>y</mi> <mo>^</mo> </mover> <mo>)</mo> </mrow> <mo>&amp;times;</mo> <mi>y</mi> <mo>&amp;times;</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>-</mo> <mi>y</mi> <mo>)</mo> </mrow> </mrow>
    ehj=wj×eo×bj×(1-bj)
    Wherein e is output error, and eo is output layer vague generalization error, ehjFor hidden layer each unit vague generalization error；
    S46：Hidden layer is adjusted to the connection weight w of output layerjAnd threshold value
    Wherein w 'jWithFor the connection weight and threshold value of the hidden layer after adjustment to output layer；
    S47：Input layer is adjusted to the connection weight w of hidden layerijAnd threshold gammaj；
    <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>&amp;Delta;w</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>=</mo> <mi>&amp;eta;</mi> <mo>&amp;times;</mo> <msub> <mi>t</mi> <mrow> <mi>r</mi> <mi>i</mi> </mrow> </msub> <mo>&amp;times;</mo> <msub> <mi>eh</mi> <mi>j</mi> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msubsup> <mi>w</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> <mo>&amp;prime;</mo> </msubsup> <mo>=</mo> <msub> <mi>w</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>+</mo> <msub> <mi>&amp;Delta;w</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>&amp;Delta;&amp;gamma;</mi> <mi>j</mi> </msub> <mo>=</mo> <mi>&amp;eta;</mi> <mo>&amp;times;</mo> <msub> <mi>eh</mi> <mi>j</mi> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msubsup> <mi>&amp;gamma;</mi> <mi>j</mi> <mo>&amp;prime;</mo> </msubsup> <mo>=</mo> <msub> <mi>&amp;gamma;</mi> <mi>j</mi> </msub> <mo>+</mo> <msub> <mi>&amp;Delta;&amp;gamma;</mi> <mi>j</mi> </msub> </mrow> </mtd> </mtr> </mtable> </mfenced>
    Wherein w 'ijWith γ 'jFor the connection weight and threshold value of the input layer after adjustment to hidden layer；
    S48：Variable r is got into n from 1, then all training samples have been trained to, and then tire out the error e of each training sample Add and calculate global error E, judge whether E reaches in the range of specification error, if so, then terminating the current connection weight of training record Value and threshold value；If it is not, global error E is set into zero, and go to the training of step S43 repetitive learnings.
  8. 8. a kind of intelligent distribution network fault early warning method based on failure gene table according to claim 7, its feature exist In：The step S5 is specially：
    S51：Evaluating matrix is built with Part II intelligent distribution network historical failure data source：
    Wherein An*×mFor evaluating matrix, n*Represent the quantity of assessment sample, an*mRepresent n-th*Individual the m articles bus for assessing sample Assess score；
    S52：By evaluating matrix An*×mBP neural network state estimation model is inputted, when drawing each section of state transfer of intelligent distribution network Between mapping between sequence and failure, so as to build the gene table that is out of order.
  9. 9. a kind of intelligent distribution network fault early warning method based on failure gene table according to claim 8, its feature exist In：The step S7 is specially：
    S71：Set gene order to be compared asLength is lS, the sequence in failure gene table isLength is lU；
    S72：Set the replacement matrix that gene order compares；
    S73：Construction size is (lS+1)×(lU+ 1) matrix D of size, for depositing comparison result；
    <mrow> <mi>D</mi> <mrow> <mo>(</mo> <msub> <mi>l</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>l</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mi>m</mi> <mi>a</mi> <mi>x</mi> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mn>0</mn> </mtd> </mtr> <mtr> <mtd> <mi>D</mi> <mo>(</mo> <msub> <mi>l</mi> <mi>i</mi> </msub> <mo>-</mo> <mn>1</mn> <mo>,</mo> <msub> <mi>l</mi> <mi>j</mi> </msub> <mo>-</mo> <mn>1</mn> <mo>)</mo> <mo>+</mo> <mi>S</mi> <mo>(</mo> <msub> <mi>s</mi> <msub> <mi>l</mi> <mi>i</mi> </msub> </msub> <mo>,</mo> <msub> <mi>u</mi> <msub> <mi>l</mi> <mi>j</mi> </msub> </msub> <mo>)</mo> </mtd> </mtr> <mtr> <mtd> <mi>D</mi> <mo>(</mo> <msub> <mi>l</mi> <mi>i</mi> </msub> <mo>-</mo> <mn>1</mn> <mo>,</mo> <msub> <mi>l</mi> <mi>j</mi> </msub> <mo>)</mo> <mo>-</mo> <mi>S</mi> <mo>(</mo> <msub> <mi>s</mi> <msub> <mi>l</mi> <mi>i</mi> </msub> </msub> <mo>,</mo> <mn>0</mn> <mo>)</mo> </mtd> </mtr> <mtr> <mtd> <mi>D</mi> <mo>(</mo> <msub> <mi>l</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>l</mi> <mi>j</mi> </msub> <mo>-</mo> <mn>1</mn> <mo>)</mo> <mo>-</mo> <mi>S</mi> <mo>(</mo> <mn>0</mn> <mo>,</mo> <msub> <mi>u</mi> <msub> <mi>l</mi> <mi>j</mi> </msub> </msub> <mo>)</mo> </mtd> </mtr> </mtable> </mfenced> </mrow>
    Wherein D (li,lj) represent gene order to be matched, D (li, 0) and=D (0, lj)=0,1≤li≤lS, 1≤lj≤lU； Represent the l in gene order SiIndividual elementary state,Represent the l in gene order UjThe state of individual element；
    S74：L is found out in matrix Di *And lj *, meet：
    <mrow> <mi>D</mi> <mrow> <mo>(</mo> <msup> <msub> <mi>l</mi> <mi>i</mi> </msub> <mo>*</mo> </msup> <mo>,</mo> <msup> <msub> <mi>l</mi> <mi>j</mi> </msub> <mo>*</mo> </msup> <mo>)</mo> </mrow> <mo>=</mo> <munder> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> <mrow> <mn>1</mn> <mo>&amp;le;</mo> <msub> <mi>l</mi> <mi>i</mi> </msub> <mo>&amp;le;</mo> <msub> <mi>l</mi> <mi>S</mi> </msub> <mo>,</mo> <mn>1</mn> <mo>&amp;le;</mo> <msub> <mi>l</mi> <mi>j</mi> </msub> <mo>&amp;le;</mo> <msub> <mi>l</mi> <mi>U</mi> </msub> </mrow> </munder> <mi>D</mi> <mrow> <mo>(</mo> <msub> <mi>l</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>l</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mrow>
    Wherein D (li *,lj *) represent sequence S and U high specific to score value；
    S75：Until gene order to be compared and all gene orders of failure gene table are compared, try to achieve it is all most It is big to compare score value, by high specific to score value compared with set threshold value, if exceeding threshold value, corresponding to the early warning gene Failure.
  10. 10. a kind of intelligent distribution network fault early warning method based on failure gene table according to claim 9, its feature exist In：The Simith-Waterman gene orders alignment algorithm of the step S7 meets：
    Wherein σ represents the importance of state.
CN201710796634.6A 2017-09-06 2017-09-06 Intelligent power distribution network fault early warning method based on fault gene table Active CN107609769B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710796634.6A CN107609769B (en) 2017-09-06 2017-09-06 Intelligent power distribution network fault early warning method based on fault gene table

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710796634.6A CN107609769B (en) 2017-09-06 2017-09-06 Intelligent power distribution network fault early warning method based on fault gene table

Publications (2)

Publication Number Publication Date
CN107609769A true CN107609769A (en) 2018-01-19
CN107609769B CN107609769B (en) 2020-07-31

Family

ID=61057457

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710796634.6A Active CN107609769B (en) 2017-09-06 2017-09-06 Intelligent power distribution network fault early warning method based on fault gene table

Country Status (1)

Country Link
CN (1) CN107609769B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108537415A (en) * 2018-03-20 2018-09-14 深圳市泰和安科技有限公司 A kind of distribution method, the apparatus and system of online safety utilization of electric power
CN112649118A (en) * 2020-11-10 2021-04-13 许继集团有限公司 Transmission line temperature sensing system based on edge calculation
CN114692480A (en) * 2020-12-30 2022-07-01 上海宏力达信息技术股份有限公司 Intelligent switch reliability analysis method and system based on radial basis function neural network
CN115048965A (en) * 2022-08-15 2022-09-13 南方医科大学珠江医院 Method, device, equipment and medium for detecting fault of ultrasonic medical equipment
CN115372759A (en) * 2022-10-19 2022-11-22 国网浙江省电力有限公司温州供电公司 Cable fault diagnosis system and method for secondary circuit of transformer substation

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105225020A (en) * 2015-11-11 2016-01-06 国家电网公司 A kind of running status Forecasting Methodology based on BP neural network algorithm and system
CN106483947A (en) * 2016-09-21 2017-03-08 国网江苏省电力公司南通供电公司 Distribution Running State assessment based on big data and method for early warning
CN106874963A (en) * 2017-03-17 2017-06-20 南京邮电大学 A kind of Fault Diagnosis Method for Distribution Networks and system based on big data technology
CN106991531A (en) * 2017-03-30 2017-07-28 国家电网公司 Power distribution network project construction effect Post-assessment Method based on BP neural network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105225020A (en) * 2015-11-11 2016-01-06 国家电网公司 A kind of running status Forecasting Methodology based on BP neural network algorithm and system
CN106483947A (en) * 2016-09-21 2017-03-08 国网江苏省电力公司南通供电公司 Distribution Running State assessment based on big data and method for early warning
CN106874963A (en) * 2017-03-17 2017-06-20 南京邮电大学 A kind of Fault Diagnosis Method for Distribution Networks and system based on big data technology
CN106991531A (en) * 2017-03-30 2017-07-28 国家电网公司 Power distribution network project construction effect Post-assessment Method based on BP neural network

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108537415A (en) * 2018-03-20 2018-09-14 深圳市泰和安科技有限公司 A kind of distribution method, the apparatus and system of online safety utilization of electric power
CN112649118A (en) * 2020-11-10 2021-04-13 许继集团有限公司 Transmission line temperature sensing system based on edge calculation
CN114692480A (en) * 2020-12-30 2022-07-01 上海宏力达信息技术股份有限公司 Intelligent switch reliability analysis method and system based on radial basis function neural network
CN115048965A (en) * 2022-08-15 2022-09-13 南方医科大学珠江医院 Method, device, equipment and medium for detecting fault of ultrasonic medical equipment
CN115048965B (en) * 2022-08-15 2022-12-13 南方医科大学珠江医院 Method, device, equipment and medium for detecting fault of ultrasonic medical equipment
CN115372759A (en) * 2022-10-19 2022-11-22 国网浙江省电力有限公司温州供电公司 Cable fault diagnosis system and method for secondary circuit of transformer substation
CN115372759B (en) * 2022-10-19 2023-04-11 国网浙江省电力有限公司温州供电公司 Cable fault diagnosis system and method for secondary circuit of transformer substation

Also Published As

Publication number Publication date
CN107609769B (en) 2020-07-31

Similar Documents

Publication Publication Date Title
CN107609769A (en) A kind of intelligent distribution network fault early warning method based on failure gene table
CN112217202B (en) Distributed new energy, energy storage and power distribution network planning method considering flexibility investment
CN105930901B (en) A kind of Diagnosis Method of Transformer Faults based on RBPNN
CN107656154A (en) Based on the Diagnosis Method of Transformer Faults for improving Fuzzy C-Means Cluster Algorithm
CN107367693B (en) SOC detection system for power battery of electric vehicle
CN104636801A (en) Transmission line audible noise prediction method based on BP neural network optimization
CN105226638A (en) Based on the electric power system Calculation of Available Transfer Capability method improving learning aid algorithm
CN104636823A (en) Wind power prediction method
CN111490542B (en) Site selection and volume fixing method of multi-end flexible multi-state switch
CN109858798B (en) Power grid investment decision modeling method and device for correlating transformation measures with voltage indexes
CN104123684A (en) Power distribution network economy reconstructing method considering network loss and reliability
CN107168292A (en) Submarine navigation device circuit failure diagnosis method based on ELM algorithms
CN111079972A (en) Method, device and medium for planning reliability of active power distribution network
CN107622360A (en) A kind of critical circuits recognition methods for considering subjective and objective factor
Huang et al. Resilient power network structure for stable operation of energy systems: A transfer learning approach
CN107171339A (en) A kind of distribution network voltage idle work optimization method containing microgrid
Yuan et al. A multi-timescale smart grid energy management system based on adaptive dynamic programming and Multi-NN Fusion prediction method
CN113283786A (en) Electricity/gas/heat/storage and distribution network collaborative planning method based on GRU and demand response
CN104021315A (en) Method for calculating station service power consumption rate of power station on basis of BP neutral network
CN115603310A (en) Energy system network topology generation method based on intelligent evolution and stability evaluation
CN111756049B (en) Data-driven reactive power optimization method considering loss of real-time measurement information of power distribution network
CN112564189A (en) Active and reactive power coordinated optimization control method
CN105184383A (en) Urban mobile emergency power supply optimal scheduling method based on intelligent optimization method
CN107169607B (en) Energy efficiency power plant optimization configuration and plant network coordination planning method based on cost
CN112152523B (en) NN/GA-based energy-saving speed regulation method for direct current motor

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