CN109858663A - Distribution network failure power failure INFLUENCING FACTORS analysis based on BP neural network - Google Patents
Distribution network failure power failure INFLUENCING FACTORS analysis based on BP neural network Download PDFInfo
- Publication number
- CN109858663A CN109858663A CN201811373125.3A CN201811373125A CN109858663A CN 109858663 A CN109858663 A CN 109858663A CN 201811373125 A CN201811373125 A CN 201811373125A CN 109858663 A CN109858663 A CN 109858663A
- Authority
- CN
- China
- Prior art keywords
- failure
- follows
- fault outage
- distribution
- calculation method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The distribution network failure power failure INFLUENCING FACTORS analysis method based on BP neural network algorithm that the present invention relates to a kind of, step 1: relevant to the investment distribution network failure power failure influence factor set of combing, the incidence relation of formation investment-influence factor-fault outage index per family；Step 2: using each influence factor as the input of BP neural network algorithm, using fault outage index per family as the output of BP neural network algorithm, establishing fault outage prediction model；Step 3: using each influence factor of Sensitivity Analysis Method analyzing influence fault outage index per family；Step 4: by the absolute value of each influence factor sensitivity and in magnitude order, it obtains to the maximum several influence factors of fault outage Index Influence per family, wherein the absolute value sequence of influence factor sensitivity is more forward, it is then bigger to fault outage Index Influence per family, illustrate that the influence factor is weaker, the upward guide of provider thus can be invested to power distribution network.
Description
Technical field
The invention belongs to field of power systems, and in particular to a kind of distribution network failure power failure based on BP neural network algorithm
INFLUENCING FACTORS analysis method.
Background technique
In recent years, with " action plan (2015-the year two thousand twenty) is transformed in distribution network construction " of National Energy Board's publication in 2015
For mark, China's Modern power distribution net construction retrofit process promoted rapidly, promoted power supply reliability by be construction retrofit core mesh
One of mark.How scientific and reasonable identification power distribution network weak link, it is reliable that power supply is more targetedly promoted with limited fund
Property be one it is urgently to be resolved with research the problem of.
There is part research to carry out diagnostic analysis by the practical power failure data of power distribution network to somewhere in the past, stops according to history
Safety at power cut and power failure location lookup in electric data restrict the weak link of electric network reliability, are directed to multiple malfunctions region
Property is transformed.But such method is only capable of solving the subproblem of distribution network reliability to a certain extent, not deep layer
The secondary root problem for excavating electric network fault and having a power failure.There are also part researchs by carrying out reliability to the power distribution network in regional scope
Assessment, by assess software assessment result find distribution network reliability weak spot, but this method acquired results although it is contemplated that
The influence that grid structure has a power failure to electric network fault, essence still only knows as a result, unknown cause, can not from influencing power distribution network therefore
Hinder and finds weak link on the practical source having a power failure.
Sensitivity analysis is that the state of one system (or model) of research and analysis or output changed to system parameter or week
The method of the sensitivity of foxing part variation.It is larger can also to determine which parameter has system or model by sensitivity analysis
Influence.Sensitivity Analysis Method is introduced into the analysis of distribution network reliability weak link by the present invention, and science identification power distribution network is thin
Weak link provides support for power distribution network investment decision.
Summary of the invention
Goal of the invention: the long-term existence due to weighing the light distribution investment tactics mode of major network in the past, and major network construction and distribution
Construction mode and having differences property of purpose, cause the investment tactics for power distribution network weak link it is more difficult have mature method according to
It follows.Therefore, for the poor problem of purpose is invested in current power distribution network investment process, the present invention concludes combed and throwing first
Provide relevant distribution network failure power failure influence factor set.Then, using influence factor set as the input of BP neural network, with
The fault outage time as output, realizes the forecast analysis to fault outage reliability index to power distribution network per family.Finally, using
Sensitivity Analysis Method carries out sensitivity analysis to each input data of BP neural network, and finding out, which influences distribution network failure power failure, refers to
Target weak link, convenient for power grid construction, administrative department is targetedly invested.
It is above-mentioned in the prior art because lacking scientific and reasonable power distribution network method for analyzing weak link in order to overcome the problems, such as, this
Invention provides a kind of distribution network failure power failure INFLUENCING FACTORS analysis method based on BP neural network algorithm.
To achieve the above objectives, the technical solution adopted by the present invention is that:
A kind of distribution network failure power failure INFLUENCING FACTORS analysis method based on BP neural network algorithm, including it is following
Step:
Step 1: combing distribution network failure power failure influence factor set relevant to investment forms investment-influence factor-family
The incidence relation of equal fault outage index；
Step 2: refreshing using fault outage index per family as BP using each influence factor as the input of BP neural network algorithm
Fault outage prediction model is established in output through network algorithm；
Step 3: using each influence factor of Sensitivity Analysis Method analyzing influence fault outage index per family；
Step 4: by the absolute value of each influence factor sensitivity and in magnitude order, obtaining to fault outage index per family
Maximum several influence factors are influenced, the absolute value sequence of wherein influence factor sensitivity is more forward, then to fault outage per family
Index Influence is bigger, illustrates that the influence factor is weaker, and the upward guide of provider thus can be invested to power distribution network.
On the basis of above scheme, distribution network failure power failure influence factor set relevant to investment includes: rack knot
Structure factor, equipment factor, repairing horizontal factor and depreciation factor.
On the basis of above scheme, the grid structure factor includes: the hung load proportion of averagely every section of route, route
Average length, can turn averagely to install for feeder line ratio, the average radius of electricity supply of single substation, medium-voltage line average segments
Capacity, per family capacity.
On the basis of above scheme, the calculation method of average the hung load proportion of every section of route are as follows: be to the moon
The only transformer number of units of average every section of route mounting；The calculation method of the route average length are as follows: in transport line until the moon
Road average length；The calculation method of the average segments are as follows: average every line sectionalizing number；It is described to turn for feeder line ratio
Calculation method are as follows: possess interconnection switch feeder line accounting；The calculation method of the average radius of electricity supply of the single substation are as follows: about etc.
There is number of switches in line length/substation；Medium pressure route is averaged the calculation method of installed capacity are as follows: distribution transformer
Total capacity (public+specially)/distribution wire travel permit number；The calculation method of the capacity per family are as follows: the distribution transforming total capacity/total family of this month low pressure
Number.
On the basis of above scheme, the equipment factor includes: overhead transmission line insulation rate, cable rate, matches
Become rated capacity distribution, panel switches arc extinguishing property accounting, the distribution transforming distribution of moon limit, the switch moon that puts into operation that puts into operation limit.
On the basis of above scheme, the calculation method of the overhead transmission line insulation rate are as follows: overhead transmission line insulation in this month
Line total length/overhead transmission line total length；The calculation method of the cable rate are as follows: this month cable run total length/route overall length
Degree；The calculation method of the distribution transforming rated capacity distribution are as follows: each rated capacity distribution transforming accounting situation statistics, accounting is bigger, equipment
It is horizontal higher；The calculation method of the panel switches arc extinguishing property accounting are as follows: in oil circuit breaker/master switch of fortune until this month
Number, until this month the vacuum switch of fortune/master switch number, until this month in sulphur hexafloride circuit breaker number/master switch number of fortune；
The distribution transforming puts into operation the calculation method of moon limit distribution are as follows: the time distribution transforming number of units/total in 3 years that puts into operation until this month in fortune becomes
Depressor number of units, the time of putting into operation until this month in fortune are transporting in 3-5 distribution transforming number of units/total transformer number of units, until the moon
Time of putting into operation in 5 years or more distribution transforming number of units/total transformer number of units；The calculation method for switching the moon limit that puts into operation are as follows: to the moon
Until in the time of putting into operation of fortune number of units/master switch number of units, the time of putting into operation until this month in fortune are switched in 3 years in 3-5
Number of units/master switch number of units, the time of putting into operation until this month in fortune are switched in 5 years or more switch number of units/master switch number of units.
On the basis of above scheme, the depreciation factor is the variation of equipment failure rate over time；It is described
Repairing horizontal factor includes: mean failure rate recovery time, medium-and-large-sized failure mean failure rate recovery time, mini fault mean failure rate
Under recovery time, medium-and-large-sized fault occurrence frequency, livewire work number, the distribution of fault outage range, exceedingly odious weather condition
Failure Mean Time To Recovery.
On the basis of above scheme, the calculation method of the mean failure rate recovery time are as follows: when fault outage was total in this month
Length/fault outage number；The medium-and-large-sized failure mean failure rate recovery time is more than or equal to 4 hours, calculation method are as follows: the middle of the month
Large-scale fault outage total duration/medium-and-large-sized fault outage number；The mini fault mean failure rate recovery time less than 4 hours,
Calculation method are as follows: this month mini fault power failure total duration/mini fault frequency of power cut；The meter of the medium-and-large-sized fault occurrence frequency
Calculation method is this month medium-and-large-sized failure whole month cumulative number/whole month to add up the number of stoppages；The calculating side of the livewire work number
Method are as follows: this month livewire work total degree；The calculation method of the fault outage range distribution are as follows: this month 1 family of fault outage influence,
The family 2-5, the family 6-10, the family 11-20, all kinds of accountings of power-off event more than 20 families；Failure is flat under the exceedingly odious weather condition
The calculation method of equal recovery time are as follows: the moon extreme bad weather power failure total duration/exceedingly odious weather frequency of power cut.
On the basis of above scheme, the detailed process of step 2 are as follows:
Step 21) will affect factor and be divided into training set, test set and verifying collection according to a certain percentage, with training set and test
Collect the training data as fault outage prediction model；
The expression formula of the fault outage prediction model is as follows:
In formula: i, j, l respectively represent input layer, hidden layer, output layer；xiIndicate the influence factor of input；wijAnd wjlPoint
It Biao Shi not input layer and hidden layer, hidden layer and the weight for exporting interlayer, wijAnd wjlInitial value be respectively [- 1,1] on
Machine value；θjAnd θlRespectively indicate the threshold value of hidden layer and output layer, θjAnd θlInitial value be respectively [- 1,1] on random value；f
The excitation function of hidden layer and output layer is respectively indicated with g；Y ' is to predict the fault outage time per family；
Weight w of the step 22) to formula (1)ij、wjlWith threshold θj、θlIt is modified, makes error function E under gradient direction
Drop, specific as follows shown:
In formula: y is that fault outage time, y ' are to predict the fault outage time per family to reality per family；
wij、wjl、θjAnd θlMore new formula it is as follows:
In formula,It respectively indicates before not updated in input layer and hidden layer, hidden layer and output layer iterative process
Weight,Respectively indicate the threshold value before not updating in hidden layer and output layer iterative process；Table respectively
Show the weight updated in input layer and hidden layer, hidden layer and output layer iterative process,Respectively indicate hidden layer with
The threshold value updated in output layer iterative process；η indicates that learning efficiency, value are between 0 to 1；
When the error function of all samples is met the requirements, then deconditioning；
Verifying is collected the fault outage prediction model for substituting into training and completing by step 23), verifies the fault outage that training is completed
The accuracy of prediction model；If accuracy is higher, the training process of termination failure power failure prediction model；If accuracy is lower,
It then replaces data set to repeat the above steps, until accuracy rate is satisfied.
On the basis of above scheme, the influence factor acquired recently is substituted into the fault outage prediction model that training is completed
In, obtain the predicted value y of fault outage time per familyend′；The value of the influence factor acquired recently is all increased to 10% size,
As the input of fault outage prediction model, the predicted value y of the new time of fault outage per family is obtainedk', wherein k=1,2,
3 ..., 20, then calculating each influence factor influences the susceptibility of fault outage index per family, and calculation formula is as follows:
△yk=yk'-yend' k=1,2,3 ..., 20 (7)
In formula, △ ykSize be susceptibility of each influence factor to fault outage index per family.
Invention effect
(1) the invention proposes distribution network reliability influence factor set relevant to investment, can be by investment and power distribution network
Fault outage is associated, and breaches the limitation of convectional reliability appraisal procedure, can be horizontal by electric network composition and operation management
Historical data realize distribution network failure have a power failure forecast analysis.
(2) concept for introducing sensitivity, can extract each influence factor of certain regional effect fault outage according to model built
Susceptibility, sorted by susceptibility, can effectively identify power distribution network weak link, for power distribution network investment assistant decision making support is provided
Support.
Detailed description of the invention
The present invention has following attached drawing:
Distribution network failure power failure INFLUENCING FACTORS analysis method flow chart of the Fig. 1 based on BP neural network algorithm.
Specific embodiment
Below in conjunction with attached drawing, invention is further described in detail.
Step 1: combing distribution network failure power failure influence factor set relevant to investment forms investment-influence factor-family
The incidence relation of equal fault outage index；
The major influence factors for influencing distribution network failure power failure reliability index relevant to investment can be summarized as rack knot
Structure factor, equipment factor, repairing horizontal (operation management is horizontal) factor, depreciation factor.Grid structure factor therein is again
Including；Average every section of route hung load proportion, average segments, can turn for feeder line ratio, single power transformation route average length
The average radius of electricity supply stood, medium-voltage line are averaged installed capacity, per family capacity；Equipment factor includes overhead transmission line insulating
The put into operation distribution of moon limit, switch of rate, cable rate, the distribution of distribution transforming rated capacity, panel switches arc extinguishing property accounting, distribution transforming puts into operation the moon
Limit；Depreciation factor is the variation of equipment failure rate over time；Operation management horizontal factor includes that mean failure rate restores
Time, medium-and-large-sized failure mean failure rate recovery time, mini fault mean failure rate recovery time, medium-and-large-sized fault occurrence frequency,
Failure Mean Time To Recovery under livewire work number, the distribution of fault outage range, exceedingly odious weather condition.Be specifically related to because
Plain, each factor calculation formula as shown in the table below:
Table 1. is invested relevant distribution network failure power failure influence factor and is summarized
Using the moon as granularity, when collecting the historical data and the fault outage per family in corresponding month of each influence factor in somewhere
Between, history of forming data acquisition system.
Step 2: using each influence factor data of the history of collection as the input of BP neural network algorithm, being stopped with failure per family
Electric index is the output of BP neural network algorithm, establishes fault outage prediction model；
Historical data is divided into training set, test set, verifying collection in the ratio of 7:2:1 first, with training set and test set
Model training data as BP neural network algorithm.
BP neural network is a kind of multilayer feedforward neural network, is made of input layer, hidden layer and output layer, is had reversed
The ability of study exports expression formula are as follows:
In formula, i, j, l respectively represent input layer, hidden layer, output layer；xiIndicate input；wijAnd wjlRespectively indicate input
The weight of layer and hidden layer, hidden layer and output interlayer；
θjAnd θlRespectively indicate the threshold value of hidden layer and output layer；F and g respectively indicates the excitation letter of hidden layer and output layer
Number；The prediction of y ' expression BP neural network exports.
By above-mentioned algorithm and historical data, to the network weight w of formula (1)ij、wjlWith threshold θj、θlIt is modified, makes to miss
Difference function E declines along gradient direction:
In formula, y is that fault outage time, y ' are to predict the fault outage time per family to reality per family.
Neural network uses error feedback learning algorithm in learning process, wijAnd wjl、θjAnd θlBe initialized to [- 1,
1] random value on.The learning process of entire model is made of the forward-propagating of input data and the backpropagation of error.Just
During propagation, input signal is successively handled since the input layer of model, and is transmitted to output layer.When error is unsatisfactory for wanting
When asking, then error back propagation is carried out, by modifying wijAnd wjl、θjAnd θl, error is made to become smaller.
wij、wjl、θjAnd θlMore new formula it is as follows:
In formula,It respectively indicates before not updated in input layer and hidden layer, hidden layer and output layer iterative process
Weight,Respectively indicate the threshold value before not updating in hidden layer and output layer iterative process；Table respectively
Show the weight updated in input layer and hidden layer, hidden layer and output layer iterative process,Respectively indicate hidden layer with
The threshold value updated in output layer iterative process；η indicates that learning efficiency, value are between 0 to 1；
When the error function of all samples is met the requirements, then deconditioning.
Verifying is collected to the BP neural network model for substituting into training and completing, tests the accuracy of training pattern.If accuracy compared with
Height then terminates the training process of model；If accuracy is lower, replaces data set and repeat the above process, until accuracy rate is satisfied
Until.
Step 3: using each influence factor of Sensitivity Analysis Method analyzing influence fault outage index per family；
A nearest historical data of time point is substituted into the model that above-mentioned training is completed, obtains the fault outage time per family
Predicted value yend′。
The size that the value of each factor of historical data of time point recently is increased to 10%, gradually instead of not changing originally
Become larger small data, and as the input of BP neural network algorithm, the predicted value y of the new time of fault outage per family can be obtainedk′(k
=1,2,3 ..., 20).Thus each influence factor, which can be calculated, influences the susceptibility of fault outage index per family.Calculation formula is such as
Under:
△yk=yk'-yend' k=1,2,3 ..., 20 (3)
In formula, △ ykSize be susceptibility of each influence factor to fault outage time index per family.
Step 4: in magnitude order by the absolute value of each influence factor sensitivity, sorting more forward, influence to get on having a power failure
Greatly, illustrate that this index value of this area is weaker.Thus the upward guide of provider can be invested to power distribution network.
Key point of the invention and point to be protected:
1. in the past do not had research power distribution network investment and fault outage relationship aiming at the problem that, the invention proposes with invest phase
The distribution network reliability influence factor set of pass.
2. the present invention is according to history power failure thing aiming at the problem that power distribution network operation management horizontal (repairing is horizontal) is difficult to quantify
The retrievable index such as different types of faults service restoration time, livewire work number indirectly reflects management level in part, expands
The influence factor of fault outage reliability index is opened up.
3. the invention proposes one kind based on BP mind aiming at the problem that current power distribution network weak link still lacks maturation method
Sensitivity Analysis Method through network is applied to power distribution network weak link identification process, can provide for power distribution network investment auxiliary
Decision support is helped to act on.
The content being not described in detail in this specification belongs to the prior art well known to professional and technical personnel in the field.
Claims (10)
1. a kind of distribution network failure power failure INFLUENCING FACTORS analysis method based on BP neural network algorithm, feature exist
In, comprising the following steps:
Step 1: combing distribution network failure power failure influence factor set relevant to investment forms the event per family of investment-influence factor-
Hinder the incidence relation of power failure index；
Step 2: using each influence factor as the input of BP neural network algorithm, using fault outage index per family as BP nerve net
Fault outage prediction model is established in the output of network algorithm；
Step 3: using each influence factor of Sensitivity Analysis Method analyzing influence fault outage index per family；
Step 4: by the absolute value of each influence factor sensitivity and in magnitude order, obtaining to fault outage Index Influence per family
The absolute value sequence of maximum several influence factors, wherein influence factor sensitivity is more forward, then to fault outage index per family
Influence is bigger, illustrates that the influence factor is weaker, and the upward guide of provider thus can be invested to power distribution network.
2. the distribution network failure power failure INFLUENCING FACTORS analysis side based on BP neural network algorithm as described in claim 1
Method, which is characterized in that distribution network failure power failure influence factor set relevant to investment includes: grid structure factor, equipment water
Flat factor, repairing horizontal factor and depreciation factor.
3. the distribution network failure power failure INFLUENCING FACTORS analysis side based on BP neural network algorithm as claimed in claim 2
Method, which is characterized in that the grid structure factor includes: the hung load proportion of averagely every section of route, route average length, is averaged
Segments can turn be averaged for feeder line ratio, the average radius of electricity supply of single substation, medium-voltage line installed capacity, per family capacity.
4. the distribution network failure power failure INFLUENCING FACTORS analysis side based on BP neural network algorithm as claimed in claim 3
Method, which is characterized in that the calculation method of average the hung load proportion of every section of route are as follows: average every section of route until the moon
The transformer number of units of mounting；The calculation method of the route average length are as follows: in transport line road average length until the moon；It is described
The calculation method of average segments are as follows: average every line sectionalizing number；The calculation method turned for feeder line ratio are as follows: possess
Interconnection switch feeder line accounting；The calculation method of the average radius of electricity supply of the single substation are as follows: be equal to line length/substation
There is number of switches；Medium pressure route is averaged the calculation method of installed capacity are as follows: distribution transformer total capacity/distribution wire travel permit
Number；The calculation method of the capacity per family are as follows: the distribution transforming total capacity/total amount of this month low pressure.
5. the distribution network failure power failure INFLUENCING FACTORS analysis side based on BP neural network algorithm as claimed in claim 2
Method, which is characterized in that the equipment factor includes: overhead transmission line insulation rate, cable rate, distribution transforming rated capacity point
The put into operation distribution of moon limit, switch of cloth, panel switches arc extinguishing property accounting, distribution transforming puts into operation moon limit.
6. the distribution network failure power failure INFLUENCING FACTORS analysis side based on BP neural network algorithm as claimed in claim 5
Method, which is characterized in that the calculation method of the overhead transmission line insulation rate are as follows: this month overhead transmission line insulated wire total length/aerial
Total line length；The calculation method of the cable rate are as follows: this month cable run total length/total line length；The distribution transforming volume
The calculation method of constant volume distribution are as follows: each rated capacity distribution transforming accounting situation statistics, accounting is bigger, and equipment is higher；It is described
The calculation method of panel switches arc extinguishing property accounting are as follows: until this month the oil circuit breaker of fortune/master switch number, until the moon
The vacuum switch of fortune/master switch number, until this month in sulphur hexafloride circuit breaker number/master switch number of fortune；The distribution transforming puts into operation the moon
Limit the calculation method of distribution are as follows: until this month fortune put into operation the time in 3 years distribution transforming number of units/total transformer number of units, to should
Time of putting into operation until month in fortune is in 3-5 distribution transforming number of units/total transformer number of units, the time of putting into operation until this month in fortune 5
Year or more distribution transforming number of units/total transformer number of units；The calculation method for switching the moon limit that puts into operation are as follows: the putting into operation in fortune until this month
Time switchs number of units/master switch number of units in 3 years, the time of putting into operation until this month in fortune switchs number of units/master switch in 3-5
Number of units, the time of putting into operation until this month in fortune were in 5 years or more switch number of units/master switch number of units.
7. the distribution network failure power failure INFLUENCING FACTORS analysis side based on BP neural network algorithm as claimed in claim 2
Method, which is characterized in that the depreciation factor is the variation of equipment failure rate over time；The repairing horizontal factor packet
Include: mean failure rate recovery time, medium-and-large-sized failure mean failure rate recovery time, mini fault mean failure rate recovery time, in it is big
When failure is averagely restored under type fault occurrence frequency, livewire work number, the distribution of fault outage range, exceedingly odious weather condition
Between.
8. the distribution network failure power failure INFLUENCING FACTORS analysis side based on BP neural network algorithm as claimed in claim 7
Method, which is characterized in that the calculation method of the mean failure rate recovery time are as follows: this month fault outage total duration/fault outage
Number；The medium-and-large-sized failure mean failure rate recovery time is more than or equal to 4 hours, calculation method are as follows: this month medium-and-large-sized fault outage
Total duration/medium-and-large-sized fault outage number；The mini fault mean failure rate recovery time less than 4 hours, calculation method are as follows: should
Moon mini fault power failure total duration/mini fault frequency of power cut；The calculation method of the medium-and-large-sized fault occurrence frequency is the moon
Medium-and-large-sized failure whole month cumulative number/whole month adds up the number of stoppages；The calculation method of the livewire work number are as follows: moon electrification
Operation total degree；The calculation method of fault outage range distribution are as follows: this month fault outage influence 1 family, the family 2-5, the family 6-10,
All kinds of accountings of the power-off event more than family 11-20,20 families；The meter of failure Mean Time To Recovery under the exceedingly odious weather condition
Calculation method are as follows: the moon extreme bad weather power failure total duration/exceedingly odious weather frequency of power cut.
9. the distribution network failure based on BP neural network algorithm as described in claim 1-8 any claim have a power failure influence because
Plain sensitivity analysis method, which is characterized in that the detailed process of step 2 are as follows:
Step 21) will affect factor and be divided into training set, test set and verifying collection according to a certain percentage, be made with training set and test set
For the training data of fault outage prediction model；
The expression formula of the fault outage prediction model is as follows:
In formula: i, j, l respectively represent input layer, hidden layer, output layer；xiIndicate the influence factor of input；wijAnd wjlTable respectively
Show the weight of input layer and hidden layer, hidden layer and output interlayer, wijAnd wjlInitial value be respectively [- 1,1] on it is random
Value；θjAnd θlRespectively indicate the threshold value of hidden layer and output layer, θjAnd θlInitial value be respectively [- 1,1] on random value；F and
G respectively indicates the excitation function of hidden layer and output layer；Y ' is to predict the fault outage time per family；
Weight w of the step 22) to formula (1)ij、wjlWith threshold θj、θlIt is modified, declines error function E along gradient direction, tool
Body is as follows:
In formula: y is that fault outage time, y ' are to predict the fault outage time per family to reality per family；
wij、wjl、θjAnd θlMore new formula it is as follows:
In formula,Respectively indicate the power before not updating in input layer and hidden layer, hidden layer and output layer iterative process
Value,Respectively indicate the threshold value before not updating in hidden layer and output layer iterative process；It respectively indicates defeated
Enter the weight updated in layer and hidden layer, hidden layer and output layer iterative process,Respectively indicate hidden layer and output
The threshold value updated during stacking generation；η indicates that learning efficiency, value are between 0 to 1；
When the error function of all samples is met the requirements, then deconditioning；
Verifying is collected the fault outage prediction model for substituting into training and completing by step 23), verifies the fault outage prediction that training is completed
The accuracy of model；If accuracy is higher, the training process of termination failure power failure prediction model；If accuracy is lower, more
It changes data set to repeat the above steps, until accuracy rate is satisfied.
10. the distribution network failure power failure INFLUENCING FACTORS analysis based on BP neural network algorithm as claimed in claim 9
Method, which is characterized in that the influence factor acquired recently is substituted into the fault outage prediction model that training is completed, obtained per family
The predicted value y of fault outage timeend′；The size that the value of the influence factor acquired recently is all increased to 10%, stops as failure
The input of electric prediction model obtains the predicted value y of the new time of fault outage per familyk', wherein k=1,2,3 ..., 20, then
Calculating each influence factor influences the susceptibility of fault outage index per family, and calculation formula is as follows:
△yk=yk'-yend' k=1,2,3 ..., 20 (7)
In formula, △ ykSize be susceptibility of each influence factor to fault outage index per family.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811373125.3A CN109858663A (en) | 2018-11-19 | 2018-11-19 | Distribution network failure power failure INFLUENCING FACTORS analysis based on BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811373125.3A CN109858663A (en) | 2018-11-19 | 2018-11-19 | Distribution network failure power failure INFLUENCING FACTORS analysis based on BP neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109858663A true CN109858663A (en) | 2019-06-07 |
Family
ID=66890115
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811373125.3A Pending CN109858663A (en) | 2018-11-19 | 2018-11-19 | Distribution network failure power failure INFLUENCING FACTORS analysis based on BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109858663A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111523718A (en) * | 2020-04-16 | 2020-08-11 | 广东电网有限责任公司 | Line fault prediction method, system and storage medium based on artificial intelligence |
| CN111681128A (en) * | 2020-05-14 | 2020-09-18 | 国网河北能源技术服务有限公司 | Power failure sensitivity analysis method based on neural network and clustering |
| CN111864732A (en) * | 2020-01-09 | 2020-10-30 | 南方电网科学研究院有限责任公司 | Power supply unit optimizing system of distribution network |
| CN111967180A (en) * | 2020-07-02 | 2020-11-20 | 华南理工大学 | Quantitative analysis method for influence degree of papermaking fiber characteristics on mechanical properties of finished paper |
| CN112036682A (en) * | 2020-07-10 | 2020-12-04 | 广西电网有限责任公司 | Early warning method and device for frequent power failure |
| CN112149903A (en) * | 2020-09-22 | 2020-12-29 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Primary frequency modulation analysis and optimization method of thermal power generating unit based on BP neural network algorithm |
| CN112348384A (en) * | 2020-11-13 | 2021-02-09 | 国网河南省电力公司经济技术研究院 | Data-driven power distribution network planning scheme reliability rapid evaluation method |
| CN112525438A (en) * | 2020-10-15 | 2021-03-19 | 国网浙江省电力有限公司杭州供电公司 | SF (sulfur hexafluoride)6Air leakage monitoring method and system for density relay |
| CN112734115A (en) * | 2021-01-13 | 2021-04-30 | 国网山东省电力公司日照供电公司 | Big data-based client side trip risk pre-control method and system |
| CN112906251A (en) * | 2021-04-16 | 2021-06-04 | 云南电网有限责任公司 | Analysis method and system for reliability influence factors of power distribution network |
| CN113988655A (en) * | 2021-10-30 | 2022-01-28 | 国网河南省电力公司电力科学研究院 | Power transmission line running state evaluation method considering multiple meteorological factors |
| CN114548601A (en) * | 2022-03-16 | 2022-05-27 | 国网河南省电力公司电力科学研究院 | Power distribution network power failure prediction method and system under extreme disasters based on BP neural network |
| CN115513934A (en) * | 2022-08-22 | 2022-12-23 | 国网江苏省电力有限公司南京供电分公司 | Client-centered distribution network disposal and fault self-healing strategy optimization method |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100070227A1 (en) * | 2008-09-18 | 2010-03-18 | Searete Llc, A Limited Liability Corporation Of The State Of Delaware | System and method for identifying appliances by electrical characteristics |
| CN102722759A (en) * | 2012-05-17 | 2012-10-10 | 河海大学 | Method for predicting power supply reliability of power grid based on BP neural network |
| CN104318318A (en) * | 2014-10-10 | 2015-01-28 | 国家电网公司 | Power grid loss influence quantization model by charge, network and source coordination control-based factors |
| CN105375461A (en) * | 2015-07-10 | 2016-03-02 | 华北电力大学（保定） | Active power distribution network power supply capacity real-time assessment method based on prediction technology |
| CN105427195A (en) * | 2015-12-23 | 2016-03-23 | 国家电网公司 | Calculation method of reliability index of power transmission and distribution integration |
| CN106845711A (en) * | 2017-01-22 | 2017-06-13 | 南方电网科学研究院有限责任公司 | Processing method and processing device for power supply reliability data |
| CN106980905A (en) * | 2017-03-15 | 2017-07-25 | 南方电网科学研究院有限责任公司 | Power distribution network power supply reliability prediction method and system |
| CN107067337A (en) * | 2016-12-08 | 2017-08-18 | 国家电网公司 | A kind of power distribution network high reliability powering mode determines method |
-
2018
- 2018-11-19 CN CN201811373125.3A patent/CN109858663A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100070227A1 (en) * | 2008-09-18 | 2010-03-18 | Searete Llc, A Limited Liability Corporation Of The State Of Delaware | System and method for identifying appliances by electrical characteristics |
| CN102722759A (en) * | 2012-05-17 | 2012-10-10 | 河海大学 | Method for predicting power supply reliability of power grid based on BP neural network |
| CN104318318A (en) * | 2014-10-10 | 2015-01-28 | 国家电网公司 | Power grid loss influence quantization model by charge, network and source coordination control-based factors |
| CN105375461A (en) * | 2015-07-10 | 2016-03-02 | 华北电力大学（保定） | Active power distribution network power supply capacity real-time assessment method based on prediction technology |
| CN105427195A (en) * | 2015-12-23 | 2016-03-23 | 国家电网公司 | Calculation method of reliability index of power transmission and distribution integration |
| CN107067337A (en) * | 2016-12-08 | 2017-08-18 | 国家电网公司 | A kind of power distribution network high reliability powering mode determines method |
| CN106845711A (en) * | 2017-01-22 | 2017-06-13 | 南方电网科学研究院有限责任公司 | Processing method and processing device for power supply reliability data |
| CN106980905A (en) * | 2017-03-15 | 2017-07-25 | 南方电网科学研究院有限责任公司 | Power distribution network power supply reliability prediction method and system |
Non-Patent Citations (2)
| Title |
|---|
| 刘亚南，等: "基于PCA和RVM的电网供电可靠性预测模型研究", 《电力系统保护与控制》 * |
| 宋云亭，等: "南方电网"十一五"城市供电可靠性规划", 《电网技术》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111864732A (en) * | 2020-01-09 | 2020-10-30 | 南方电网科学研究院有限责任公司 | Power supply unit optimizing system of distribution network |
| CN111523718A (en) * | 2020-04-16 | 2020-08-11 | 广东电网有限责任公司 | Line fault prediction method, system and storage medium based on artificial intelligence |
| CN111681128A (en) * | 2020-05-14 | 2020-09-18 | 国网河北能源技术服务有限公司 | Power failure sensitivity analysis method based on neural network and clustering |
| CN111967180B (en) * | 2020-07-02 | 2022-07-26 | 华南理工大学 | Quantitative analysis method for influence degree of papermaking fiber characteristics on mechanical properties of finished paper |
| CN111967180A (en) * | 2020-07-02 | 2020-11-20 | 华南理工大学 | Quantitative analysis method for influence degree of papermaking fiber characteristics on mechanical properties of finished paper |
| CN112036682A (en) * | 2020-07-10 | 2020-12-04 | 广西电网有限责任公司 | Early warning method and device for frequent power failure |
| CN112149903A (en) * | 2020-09-22 | 2020-12-29 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Primary frequency modulation analysis and optimization method of thermal power generating unit based on BP neural network algorithm |
| CN112525438A (en) * | 2020-10-15 | 2021-03-19 | 国网浙江省电力有限公司杭州供电公司 | SF (sulfur hexafluoride)6Air leakage monitoring method and system for density relay |
| CN112348384A (en) * | 2020-11-13 | 2021-02-09 | 国网河南省电力公司经济技术研究院 | Data-driven power distribution network planning scheme reliability rapid evaluation method |
| CN112734115A (en) * | 2021-01-13 | 2021-04-30 | 国网山东省电力公司日照供电公司 | Big data-based client side trip risk pre-control method and system |
| CN112906251A (en) * | 2021-04-16 | 2021-06-04 | 云南电网有限责任公司 | Analysis method and system for reliability influence factors of power distribution network |
| CN113988655A (en) * | 2021-10-30 | 2022-01-28 | 国网河南省电力公司电力科学研究院 | Power transmission line running state evaluation method considering multiple meteorological factors |
| CN114548601A (en) * | 2022-03-16 | 2022-05-27 | 国网河南省电力公司电力科学研究院 | Power distribution network power failure prediction method and system under extreme disasters based on BP neural network |
| CN115513934A (en) * | 2022-08-22 | 2022-12-23 | 国网江苏省电力有限公司南京供电分公司 | Client-centered distribution network disposal and fault self-healing strategy optimization method |
| CN115513934B (en) * | 2022-08-22 | 2024-10-18 | 国网江苏省电力有限公司南京供电分公司 | Customer-centric distribution network treatment and fault self-healing strategy optimization method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109858663A (en) | Distribution network failure power failure INFLUENCING FACTORS analysis based on BP neural network | |
| CN109146124B (en) | Power distribution terminal transformation decision method based on time-varying failure rate | |
| CN103647276A (en) | Electric energy quality early warning system and method thereof | |
| CN110570016B (en) | Disaster-resistant backbone net rack optimization method considering multi-stage performance | |
| CN103559654A (en) | Comprehensive analysis system and method of power distribution network | |
| CN110222889B (en) | Power distribution network feeder automation terminal configuration method based on multiple intelligent algorithms | |
| CN110210095A (en) | A kind of distribution network reliability index calculating method based on mixed integer linear programming | |
| CN103810328A (en) | Transformer maintenance decision method based on hybrid model | |
| CN105281344A (en) | Smart distribution network self-restoration optimization method considering power quality and uncertainty constraint thereof | |
| CN103679294A (en) | Differentiation based power transmission network anti-disaster ability planning method | |
| CN115587649A (en) | Power transmission line maintenance sequence optimization decision method under extreme ice and snow disasters | |
| CN107453354B (en) | A kind of weak link recognition methods of power distribution network | |
| CN107563620A (en) | A kind of integrated evaluating method based on equipment life-cycle information | |
| CN109687410A (en) | A kind of cascading failure collection construction method increased based on maximum entropy | |
| Heidari et al. | Reliability evaluation in power distribution system planning studies | |
| CN105140973A (en) | High-voltage direct-current power transmission system reliability assessment method considering alternating-current side fault | |
| CN107527123A (en) | A kind of disturbance event Forecasting Methodology and device based on distributed association rules | |
| CN111369388A (en) | Method and system for evaluating anti-vulnerability of urban power grid | |
| CN107563641B (en) | Disaster-resistant multi-scene differentiation planning method for power distribution network considering disaster preference | |
| CN107292759A (en) | Distribution network planning based on power supply reliability calculates analysis system | |
| Roland et al. | Reliability prediction of Port Harcourt electricity distribution network using NEPLAN | |
| CN114814402A (en) | Abnormal electricity utilization analysis method based on integrated electricity quantity and line loss system big data | |
| CN110492488B (en) | Automatic generation method for low-frequency low-voltage load reduction sequence of regional power grid | |
| CN107294205A (en) | A kind of transformer station's state monitoring method based on Prudential Master data | |
| CN103390249B (en) | Based on the distribution scheduling aid decision method of various dimensions |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190607 |
|
| RJ01 | Rejection of invention patent application after publication |