CN110261811A - Intelligent electric meter batch method for early warning and system - Google Patents
Intelligent electric meter batch method for early warning and system Download PDFInfo
- Publication number
- CN110261811A CN110261811A CN201910602064.1A CN201910602064A CN110261811A CN 110261811 A CN110261811 A CN 110261811A CN 201910602064 A CN201910602064 A CN 201910602064A CN 110261811 A CN110261811 A CN 110261811A
- Authority
- CN
- China
- Prior art keywords
- data
- intelligent electric
- electric meter
- batch
- early warning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R35/00—Testing or calibrating of apparatus covered by the other groups of this subclass
- G01R35/04—Testing or calibrating of apparatus covered by the other groups of this subclass of instruments for measuring time integral of power or current
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a kind of intelligent electric meter batch method for early warning and system, this method comprises: obtaining intelligent electric meter actual production data；Setting model batch；The data are grouped according to the model batch of setting；Weibull Distribution is carried out to the survival functions of the data after each grouping, and obtains using Maximum Likelihood Estimation the parameter Estimation of the Weibull distribution of each groupingWithAccording to parameter EstimationWith default survival rate threshold value, the intelligent electric meter life expectancy node of each grouping is obtained；The life expectancy node is greater than to operation duration in analysis group or issues early warning apart from the batch that the life expectancy node time is less than or equal to preset time threshold.The intelligent electric meter batch method for early warning of the application is based on Right censored data, the problem of solving method in the prior art and be based on historical failure data, may not apply to all intelligent electric meter manufacturers run at present and phenotype.
Description
Technical field
This application involves intelligent electric meter reliability assessment technical field more particularly to a kind of intelligent electric meter batch method for early warning
And system.
Background technique
The relatively previous common electric energy meter of intelligent electric meter, in addition to having basic function of measuring, be additionally provided with clock, communication,
The functions such as storage.Since intelligent electric energy meter production technology is simpler than stem-winder, the producer of production is more.Due to different productions
The used production technology of producer is different, not only causes intelligent electric energy meter pattern various, and the intelligent electric energy meter matter of different batches
It measures and service life also disunity.
To avoid causing because of quality or service life reason the same batch intelligent electric meter short time that batch failure, Yi Leichang occurs
Method is to obtain Accelerating running data by accelerated life test, and then estimate batch expected life, is arrived in expected life
Terminated its operation before phase；Another kind of common method is using big data method, usage history fault data training corresponding model, so
Afterwards according to the model prediction batch electric meter fault probability after training, decided whether to terminate its operation according to probability.
Above two method is based on historical failure data, but since historical failure data is still less, and a large amount of new
Manufacturer, new phenotype lack the historical failure data that can use scale, so, above two method may not apply to the institute run at present
There are intelligent electric meter manufacturer and phenotype.
Summary of the invention
The embodiment of the present application discloses a kind of intelligent electric meter batch method for early warning and system, existing to avoid because of matter to solve
The method that amount or service life reason cause the same batch intelligent electric meter short time that batch failure occurs is based on historical failure data,
But since historical failure data is still less, and a large amount of new manufacturers, new phenotype lack the historical failure data that can use scale, institute
With the problem of above two method may not apply to all intelligent electric meter manufacturers and phenotype run at present.
The application's in a first aspect, disclosing a kind of intelligent electric meter batch method for early warning, comprising:
Obtain intelligent electric meter actual production data, wherein the data include Right censored data, the Right censored data packet
Include the intelligent electric meter data of non-mass failure dismounting and still in the intelligent electric meter data of normal operation；
Setting model batch, any one of the selected bid batch of the model batch, receipt lot, manufacturer, phenotype or
It is several；
The data are grouped according to the model batch of setting, wherein the mould having the same of the data in same grouping
Type batch；
Weibull Distribution is carried out to the survival function of the data after each grouping, and is obtained using Maximum Likelihood Estimation
Obtain the parameter Estimation of the Weibull distribution of each groupingWithWherein,WithThe respectively scale parameter and shape of Weibull distribution
Parameter；
According to parameter EstimationWith default survival rate threshold value, the intelligent electric meter life expectancy node of each grouping is obtained；
The life expectancy node is greater than to operation duration in analysis group or is less than apart from the life expectancy node time
Batch equal to preset time threshold issues early warning, the data of same analysis group model batch having the same and arrival batch
It is secondary.
Further, the parameter Estimation of the Weibull distribution that each grouping is obtained using Maximum Likelihood EstimationWithInclude:
Obtain the operation duration T of every piece of intelligent electric meteriδ is marked with observationi, wherein i is the positive integer more than or equal to 1；
According to the T of every piece of intelligent electric meteri、δiAnd survival function, setting include the very big of the Weibull distribution of Right censored data
Likelihood function；
Convert the maximum likelihood function to the maximization of log-likelihood function；
The parameter Estimation of the minimum for negative log-likelihood function of sening as an envoy to is calculated using limited memory pseudo-Newtonian algorithmWith
Further, the operation duration T for obtaining every piece of intelligent electric meteri, comprising:
If the intelligent electric meter has been removed, the operation duration TiFor the difference for removing date and installed date；
If the intelligent electric meter is still being run, the operation duration TiTo delete the difference for losing date and installed date.
Further, the observation of every piece of intelligent electric meter of the determination marks δi, comprising:
Item determines the observation label δ of every piece of intelligent electric meter according to the observationiIf observation thing occurs for the intelligent electric meter
, then δi=1；If the intelligent electric meter does not occur to observe item, δi=0.
Further, further includes:
The String data type of the data is mapped as integer data type；
The data are read in into memory in a manner of integer data type.
Further, the operation duration in same analysis group is to delete the difference for the median for losing date and installed date.
Further, the default survival rate threshold value is 0.5.
Further, the preset time threshold is 600 days.
In the second aspect of the application, a kind of intelligent electric meter batch early warning system is disclosed, comprising: first obtains module, uses
In acquisition intelligent electric meter actual production data, wherein the data include Right censored data, and the Right censored data includes non-matter
Measure the intelligent electric meter data of failure dismounting and still in the intelligent electric meter data of normal operation；
Setting module is used for setting model batch, the selected bid batch, receipt lot, manufacturer, phenotype of the model batch
Any one or more of；
Grouping module, for being grouped according to the model batch of setting to the data, wherein the number in same grouping
According to model batch having the same；
Fitting module carries out Weibull Distribution for the survival function to the data after each grouping, and using very big
Likelihood estimation obtains the parameter Estimation of the Weibull distribution of each groupingWithWherein,WithRespectively Weibull distribution
Scale parameter and form parameter；
Second obtains module, for according to parameter EstimationWith default survival rate threshold value, the intelligence electricity of each grouping is obtained
Table life expectancy node；
Warning module, for being greater than the life expectancy node to operation duration in analysis group or apart from the life expectancy
The batch that node time is less than or equal to preset time threshold issues early warning, the data model having the same of same described point of grouping
Batch and receipt lot.
Further, the system also includes:
Third obtains module, for obtaining the operation duration T of every piece of intelligent electric meteriδ is marked with observationi；
Setup module, for the T according to every piece of intelligent electric meteri、δiAnd survival function, setting include the prestige of Right censored data
The maximum likelihood function of boolean's distribution；
Conversion module, for converting the maximum likelihood function to the maximization of log-likelihood function；
Computing module, the ginseng of the minimum for calculating negative log-likelihood function of sening as an envoy to using limited memory pseudo-Newtonian algorithm
Number estimationWith
In the third aspect of the application, a kind of computer readable storage medium, the computer readable storage medium are disclosed
On be stored with computer program, the method for the application first aspect is realized when described program is executed by processor.
Scheme disclosed in the embodiment of the present application, first, by using intelligent electric meter actual production data rather than experimental data,
On the one hand the cost of experiment is eliminated, another aspect actual production data can more reflect life of the electric energy meter under real operating environments
Deposit situation；Second, the intelligent electric meter batch method for early warning of the application is removed based on Right censored data, that is, non-mass failure
Intelligent electric meter data and still in the intelligent electric meter data of normal operation, solve method in the prior art be based on history therefore
Hinder data, but since historical failure data is still less, and a large amount of new manufacturers, new phenotype lack the historical failure that can use scale
Data, so, the problem of existing method may not apply to all intelligent electric meter manufacturers and phenotype run at present.
Detailed description of the invention
In order to illustrate more clearly of the technical solution of the application, letter will be made to attached drawing needed in the embodiment below
Singly introduce, it should be apparent that, for those of ordinary skills, without any creative labor,
It is also possible to obtain other drawings based on these drawings.
Fig. 1 is a kind of workflow schematic diagram of intelligent electric meter batch method for early warning disclosed in the embodiment of the present application；
Fig. 2 is the workflow schematic diagram of another intelligent electric meter batch method for early warning disclosed in the embodiment of the present application；
Fig. 3 is the workflow schematic diagram of another intelligent electric meter batch method for early warning disclosed in the embodiment of the present application；
Fig. 4 is a kind of structural block diagram of intelligent electric meter batch early warning system disclosed in the embodiment of the present application；
Fig. 5 is the structural block diagram of another intelligent electric meter batch early warning system disclosed in the embodiment of the present application.
Specific embodiment
Technical solution in order to enable those skilled in the art to better understand the present invention, below in conjunction with of the invention real
The attached drawing in example is applied, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described implementation
Example is only a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, this field is common
Technical staff's every other embodiment obtained without making creative work, all should belong to protection of the present invention
Range.
In order to solve existing to be to avoid that the same batch intelligent electric meter short time is caused to occur because of quality or service life reason
The method of batch failure is based on historical failure data, but since historical failure data is still less, and a large amount of new manufacturer, new
Phenotype lacks the historical failure data that can use scale, so, existing method may not apply to all intelligence electricity run at present
Watch Factory quotient and the problem of phenotype, the application disclose a kind of intelligent electric meter batch method for early warning and system by following embodiment.
The application first embodiment discloses a kind of intelligent electric meter batch method for early warning.Workflow shown in Figure 1 is shown
Be intended to, intelligent electric meter batch method for early warning disclosed in the embodiment of the present application the following steps are included:
Step S11, intelligent electric meter actual production data are obtained, wherein the data include Right censored data, and the right side is deleted
Lose data include non-mass failure remove intelligent electric meter data and still in the intelligent electric meter data of normal operation.
System periodically can be generated from national grid provincial company correlation in above-mentioned steps obtain intelligence electricity across the entire province
Table actual production data, wherein data include file data and operation data, file data include device identification, bid batch,
The information that receipt lot, manufacturer, phenotype, chip type, bearing type, guarantee period etc. do not change over time, operation data include
The data that installed date, dismounting date, fault type etc. can change over time.File data and operation data are distributed in different
In data source.
Random censorship refers to that within the observation period, that is, data start to be collected between final collection time, does not observe sight
Examine the data of item, wherein observation item refers to determining the item that sample terminates operation or death, the observation item of the application
Refer to that quality fault occurs for intelligent electric meter.
Left Random censorship refers to that the item time started is unknown, and Right censored data refers to that the item end time is unknown.This
Right censored data in application refers to the intelligent electric meter data that non-mass failure is removed and still in the intelligent electric meter data of normal operation,
That is the quality fault time of intelligent electric energy meter is unknown.
Step S12, setting model batch, the selected bid batch of the model batch, receipt lot, manufacturer, in phenotype
It is any one or several.
In an example, selected bid batch is set as model batch, in another example, can select simultaneously manufacturer with
Phenotype is set as model batch.
Step S13, the data are grouped according to the model batch of setting, wherein the data in same grouping have
Identical model batch.
The data obtained in step S11 are grouped according to the model batch of setting, such as selected bid batch is set as
Model batch, then according to bid batch by data grouping.
Step S14, Weibull Distribution is carried out to the survival function of the data after each grouping, and is estimated using maximum likelihood
Meter method obtains the parameter Estimation of the Weibull distribution of each groupingWithWherein,WithThe respectively scale ginseng of Weibull distribution
Several and form parameter.
Step S14 mainly establishes Weibull model, using Maximum Likelihood Estimation to Weibull Distribution, obtains
The parameter Estimation of Weibull distributionWith
The survival function of intelligent electric meter obeys Weibull distribution.Survival function S (t) expression is still survived general in t moment
Rate refers to intelligent electric meter in t moment still in the probability of operation, specific formula are as follows:
S (t)=Pr (T > t)
T indicates the operation duration of intelligent electric meter in formula.
Estimation of the Weibull distribution parameters estimation technique to survival functionAre as follows:
λ and k is respectively form parameter and scale parameter in formula.
Then, its estimation is obtained by Maximum Likelihood EstimationWith
According to the grouping of step S13, Weibull Distribution is carried out to each grouping by group, and use Maximum-likelihood estimation side
Method obtains the parameter Estimation of the Weibull distribution of each groupingWith
Step S15, according to parameter EstimationWith default survival rate threshold value, the intelligent electric meter life expectancy of each grouping is obtained
Node.
Because the Weibull distribution after fitting is dull function, it is possible to by presetting survival rate threshold value Sth, obtain
The unique life expectancy node T being respectively groupedth, specific formula is as follows:
Tth=-λ (lnSth)-k,
Wherein λ and k is respectively the scale parameter and form parameter of Weibull distribution.
Calculate unique life expectancy node T of each groupingthWhen, wherein λ and k bring into respectively obtained in step S14 it is each
The parameter Estimation of the Weibull distribution of groupingWithDefault survival rate threshold value Sth0.5 can be preferably arranged to self-setting.
Step S16, the life expectancy node is greater than to operation duration in analysis group or apart from the life expectancy node
The batch that time is less than or equal to preset time threshold issues early warning, the data of same analysis group model batch having the same
And receipt lot.
The data after step S13 grouping are further grouped according to receipt lot, the data of same grouping are having the same
Model batch and receipt lot, and the grouping is defined as analysis group.The installed date phase of the intelligent electric meter of same receipt lot
Closely (usually in 6 months), therefore same receipt lot has similar operation duration, thus in the step according to receipt lot into
The grouping of one step.
Issuing early warning to batch can be understood as carrying out early warning as unit of analysis group to the intelligent electric meter being currently running,
The operation duration of intelligent electric meter for being currently running is to delete the difference for losing date and installed date, wherein deletes and loses what the date referred to
It is the date terminated the observation period, generally deadline of data collection.
Further, in same analysis group, that is, with a batch of intelligent electric meter operation duration be delete lose the date with
The difference of the median of installed date.
In this step, it first obtains in same analysis group in the median and installed date for deleting the mistake date of intelligent electric meter
Then intelligent electric meter in same analysis group is calculated with the median that the median for deleting the mistake date subtracts installed date in digit
Operation duration.
Life expectancy node is calculated by step S15, the operation duration and life expectancy node of more each analysis group,
The life expectancy node is greater than to operation duration in each analysis group or is less than or equal to apart from the life expectancy node time pre-
If the batch of time threshold issues early warning.Preset time threshold is preferably arranged to 600 days.Namely also apart from life expectancy node
When there are 600 days, early warning is issued, batch failure occurs to avoid the batch intelligent electric meter short time.
Wherein, it can be understood as operation duration less than or equal to preset time threshold apart from the life expectancy node time to connect
Close still to arrive life expectancy node not yet, proximity values are preset time threshold, and preset time threshold refers specifically to life expectancy section
The point time subtracts the difference of operation duration.
Scheme disclosed in the embodiment of the present application, first, by using intelligent electric meter actual production data rather than experimental data,
On the one hand the cost of experiment is eliminated, another aspect actual production data can more reflect life of the electric energy meter under real operating environments
Deposit situation；Second, the intelligent electric meter batch method for early warning of the application is removed based on Right censored data, that is, non-mass failure
Intelligent electric meter data and still in the intelligent electric meter data of normal operation, solve method in the prior art be based on history therefore
Hinder data, but since historical failure data is still less, and a large amount of new manufacturers, new phenotype lack the historical failure that can use scale
Data, so, the problem of existing method may not apply to all intelligent electric meter manufacturers and phenotype run at present.
Referring to Fig. 2, in the Wei Bu for obtaining each grouping using Maximum Likelihood Estimation provided in an embodiment of the present invention
The parameter Estimation of your distributionWithSpecifically includes the following steps:
Step S21, the operation duration T of every piece of intelligent electric meter is obtainediδ is marked with observationi, wherein i is just more than or equal to 1
Integer.
Setting, which is deleted, loses date and observation item, wherein deleting the date for losing that the date refers to that the observation period terminates, generally data
The deadline of collection.Observation item is usually that intelligent electric meter fault type is quality fault.Obtain the fortune of every piece of intelligent electric meter
Row duration TiIf the intelligent electric meter has been removed, the operation duration TiFor the difference for removing date and installed date；
If the intelligent electric meter is still being run, the operation duration TiTo delete the difference for losing date and installed date.
Item determines the observation label δ of every piece of intelligent electric meter i according to the observationiIf observation thing occurs for the intelligent electric meter
, then δi=1；If the intelligent electric meter does not occur to observe item, δi=0.
Step S22, according to the T of every piece of intelligent electric meteri、δiAnd survival function, setting include the Weibull point of Right censored data
The maximum likelihood function of cloth.
Estimation of the Weibull distribution parameters estimation technique to survival function S (t)Are as follows:
Consider the maximum likelihood function setting of the Weibull distribution of Right censored data are as follows:
Wherein, f (t)=d (F (t))/dt, f (t) are the probability density function of F (t), and F (t)=1-S (t), F (t) are tired
Product failure function, wherein f (Ti) and S (Ti) it is respectively S (t) and f (t) in t=TiValue.
The estimation of the failure density function of Weibull are as follows:
Step S23, for convenience of calculating, the maximization of L can be converted into the maximization of log-likelihood function:
Formula (1) (2) are brought into obtain:
Further abbreviation obtains log-likelihood function are as follows:
Step S24, the parameter that the minimum for negative log-likelihood function of sening as an envoy to is calculated using limited memory pseudo-Newtonian algorithm is estimated
MeterWith
Due to can not directly acquire parameter lambda and the explicit expression of k by way of seeking local derviation, limited memory is further used
Pseudo-Newtonian algorithm finds out the parameter Estimation for making the minimum of negative log-likelihood function (i.e. (3) formula takes negative)With
It is with BFGS matrix that limited memory, which intends newton (Broyden-Fletcher-Goldfarb-Shanno, BFGS) method,
Method as the symmetric positive definite Iterative Matrix in quasi-Newton method.Limited memory pseudo-Newtonian algorithm does not store Iterative Matrix, but
The m vector pair and the current negative gradient direction generation direction of search that BFGS iteration generates is used continuously.So limited memory
The complexity of pseudo-Newtonian algorithm is unrelated with data volume.
Step S21 to step S24 is executed to the data in being respectively grouped in step S13 by group, obtains the Weibull ginseng of each group
Number estimationWith
Referring to Fig. 3, in intelligent electric meter batch method for early warning provided in an embodiment of the present invention, further includes:
Step S31, the String data type of the data is mapped as integer data type.
Step S32, the data are read in into memory in a manner of integer data type.
Since the data volume of intelligent electric meter actual production data generally reaches hundred million grades, if directly by data with character string shape
Formula, which is read in memory post-processing and will be configured to computer hardware, proposes higher requirement, for this purpose, the present invention is by by character string number
It is integer data type according to Type mapping, and then datarams is accounted for into middle reduction by 93%.
Due to that would generally include containing empty data in actual production data, format error data, logic colliding data, peel off
The abnormal datas such as data, abnormal data will cause fitting result and deviation, shake etc. occur, and it is clear to carry out data to data thus
It washes, finds and correct identifiable mistake in data file, including check data consistency, handle invalid value and missing values etc..
The concrete operations of data cleansing include: to carry out data cross verification according to multiple data sources, are patrolled according to related service
Volume accordingly cleaned, while being that data is avoided to be tampered, can by the longitudinal comparison of different time nodes same data source,
Intelligent electric meter file data is cleaned.
It is understood that the data of processing of the step S12 into step S16 can be the number after carrying out data cleansing
According to.
Pass through to verify the feasibility of intelligent electric meter batch method for early warning of the invention due to artificial and time cost
Tear back the more difficult implementation of method of early warning intelligent electric meter verifying batch method for early warning accuracy open, therefore, the present invention will be by will really count
According to rollback certain time, on the one hand compare the mistake of prediction survival rate and true survival rate based on the corresponding operation duration of rollback data
Difference, to differentiate survival rate forecasting accuracy；On the other hand by bug list data before and after comparison life expectancy node, analysis is in batches
The feasibility of the method for secondary early warning.
Specific verification method includes: that acquisition cut-off all tender out date of certain Provincial Power Grid Corporation on December 25 in 2018 exist
The actual production data of intelligent electric meter in 2009.
Firstly, pre-processing to data, observation label δ is obtainediWith operation duration Ti, taken 2018 wherein deleting the mistake date
December 25.
Secondly, being grouped according to bid Mission Number, supplier number and phenotype to data, 23 groups of data are obtained, by group
Survival function is fitted using the Kaplan-Meier product method of limits and Weibull distribution, while obtaining and estimating a practical life
Deposit rate data.
Then, by data-backoff 600 days, i.e., will delete lose the date be set as retract 600 days after date, mistake is deleted according to new
Date obtains each group operation duration Ti, and the data for deleting the mistake date will be later than the dismounting date and be set as Right censored data.
Finally, be fitted using Weibull distribution to rollback data, and obtain prediction 200 days, 400 days, life in 600 days
Rate and its standard deviation are deposited, specific data are as shown in table 1, and calculate the absolute value error of each prediction survival rate and actual value, specifically
Data are as shown in table 2.
The survival rate predicted value and its standard deviation that table 1 is fitted based on Weibull
The practical survival rate of table 2 and its error
Data by observing Tables 1 and 2 can obtain, and the prediction survival rate absolute value error of each group is whole within 0.1
Forecasting accuracy is higher.Further, comparison is by the Kaplan-Meier product method of limits and Weibull distribution to survival function
The prediction survival rate for the two ways being fitted and the absolute value error of actual value, using Weibull distribution to survival function into
The prediction survival rate of the mode of row fitting and the absolute value error of actual value are much smaller than and use the Kaplan-Meier product method of limits
The prediction survival rate for the mode that survival function is fitted and the absolute value error of actual value, therefore, using Weibull distribution
The mode forecasting accuracy being fitted to survival function is higher.
To the compliance test result of early warning in batches:
By the distribution of number of faults before and after analysis and early warning, the accuracy of early warning is differentiated, if the intelligence of Warning years internal fault
Can ammeter quantity account for the major part of the quantity of all intelligent electric meters after early warning, illustrate that early warning is effective；If Warning years internal fault
Intelligent electric meter quantity accounts for the fraction of the quantity of all intelligent electric meters after early warning, then illustrates that expected life is too small, early warning is invalid.
Unification to maximum operation duration 5 years, is then intended the data-backoff in Tables 1 and 2 using Weibull distribution
It closes, the survival function of each group is obtained, for enlarged sample amount, it is assumed that one group of correspondence, one batch, while preset time threshold is
600 days, it can get the early warning of 2 batches, it is specific as shown in table 3.
The distribution of number of faults before and after 3 early warning of table
As shown in table 3, the intelligent electric meter quantity of Warning years internal fault includes intelligent electric meter number of faults and early warning before early warning
Phase intelligent electric meter number of faults two parts, intelligent electric meter failure total quantity in quantity, that is, table 3 of all intelligent electric meters after early warning, in advance
The quantity of all intelligent electric meters includes intelligent electric meter number of faults, Warning years intelligent electric meter number of faults and estimating before early warning after police
Intelligent electric meter number of faults three parts after service life node.
It will be apparent that the intelligent electric meter quantity of Warning years internal fault accounts for the major part of the quantity of all intelligent electric meters after early warning,
So early warning is effective.
Correspondingly, in an alternative embodiment of the invention, a kind of intelligent electric meter batch early warning system is also disclosed referring to Fig. 4, wrap
It includes:
First obtains module 110, for obtaining intelligent electric meter actual production data, wherein the data include right censorship
Data, the Right censored data include non-mass failure remove intelligent electric meter data and still in the intelligent electric meter number of normal operation
According to；
Setting module 120, is used for setting model batch, the selected bid batch of the model batch, receipt lot, manufacturer,
Any one or more of phenotype；
Grouping module 130, for being grouped according to the model batch of setting to the data, wherein in same grouping
Data model batch having the same；
Fitting module 140 carries out Weibull Distribution for the survival function to the data after each grouping, and uses pole
Maximum-likelihood estimation method obtains the parameter Estimation of the Weibull distribution of each groupingWithWherein,WithRespectively Weibull distribution
Scale parameter and form parameter；
Second obtains module 150, for according to parameter EstimationWith default survival rate threshold value, the intelligence of each grouping is obtained
Ammeter life expectancy node；
Warning module 160, for being greater than the life expectancy node to operation duration in analysis group or apart from the expection
The batch that service life node time is less than or equal to preset time threshold issues early warning, and the data of same described point of grouping are having the same
Model batch and receipt lot.
Referring to Fig. 5, in intelligent electric meter batch early warning system disclosed by the embodiments of the present invention, the system also includes:
Third obtains module 210, for obtaining the operation duration T of every piece of intelligent electric meteriδ is marked with observationi；
Setup module 220, for the T according to every piece of intelligent electric meteri、δiAnd survival function, setting include Right censored data
The maximum likelihood function of Weibull distribution；
Conversion module 230, for converting the maximum likelihood function to the maximization of log-likelihood function；
Computing module 240, for calculating the minimum for negative log-likelihood function of sening as an envoy to using limited memory pseudo-Newtonian algorithm
Parameter EstimationWith
Same and similar part may refer to each other between each embodiment in this specification.Especially for the reality of system
For applying example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method
In explanation.
Combine detailed description and exemplary example that the application is described in detail above, but these explanations are simultaneously
It should not be understood as the limitation to the application.It will be appreciated by those skilled in the art that without departing from the application spirit and scope,
A variety of equivalent substitution, modification or improvements can be carried out to technical scheme and embodiments thereof, these each fall within the application
In the range of.The protection scope of the application is determined by the appended claims.
In the specific implementation, the embodiment of the present application also provides a kind of computer readable storage medium, wherein this is computer-readable
Storage medium can be stored with program, which may include the implementation method provided by the present application suitable for multi-service scene when executing
Each embodiment in some or all of step.The storage medium can be magnetic disk, CD, read-only memory (read-
Only memory, ROM) or random access memory (random access memory, RAM) etc..
It is required that those skilled in the art can be understood that the technology in the embodiment of the present application can add by software
The mode of general hardware platform realize.Based on this understanding, the technical solution in the embodiment of the present application substantially or
Say that the part that contributes to existing technology can be embodied in the form of software products, which can deposit
Storage is in storage medium, such as ROM/RAM, magnetic disk, CD, including some instructions are used so that computer equipment (can be with
It is personal computer, server or the network equipment etc.) execute certain part institutes of each embodiment of the application or embodiment
The method stated.
Above-described the application embodiment does not constitute the restriction to the application protection scope.
Claims (10)
1. a kind of intelligent electric meter batch method for early warning characterized by comprising
Obtain intelligent electric meter actual production data, wherein the data include Right censored data, and the Right censored data includes non-
Quality fault remove intelligent electric meter data and still in the intelligent electric meter data of normal operation；
Setting model batch, any one or more of the selected bid batch of the model batch, receipt lot, manufacturer, phenotype；
The data are grouped according to the model batch of setting, wherein the model having the same of the data in same grouping batch
It is secondary；
Weibull Distribution is carried out to the survival function of the data after each grouping, and is obtained respectively using Maximum Likelihood Estimation
The parameter Estimation of the Weibull distribution of groupingWithWherein,WithThe respectively scale parameter of Weibull distribution and shape ginseng
Number；
According to parameter EstimationWith default survival rate threshold value, the intelligent electric meter life expectancy node of each grouping is obtained；
The life expectancy node is greater than to operation duration in analysis group or is less than or equal to apart from the life expectancy node time
The batch of preset time threshold issues early warning, the data of same analysis group model batch having the same and receipt lot.
2. intelligent electric meter batch method for early warning according to claim 1, which is characterized in that described to use Maximum-likelihood estimation
Method obtains the parameter Estimation of the Weibull distribution of each groupingWithInclude:
Obtain the operation duration T of every piece of intelligent electric meteriδ is marked with observationi, wherein i is the positive integer more than or equal to 1；
According to the T of every piece of intelligent electric meteri、δiAnd survival function, setting include the maximum likelihood of the Weibull distribution of Right censored data
Function；
Convert the maximum likelihood function to the maximization of log-likelihood function；
The parameter Estimation of the minimum for negative log-likelihood function of sening as an envoy to is calculated using limited memory pseudo-Newtonian algorithmWith
3. intelligent electric meter batch method for early warning according to claim 2, which is characterized in that every piece of intelligent electric meter of the acquisition
Operation duration Ti, comprising:
If the intelligent electric meter has been removed, the operation duration TiFor the difference for removing date and installed date；
If the intelligent electric meter is still being run, the operation duration TiTo delete the difference for losing date and installed date.
4. intelligent electric meter batch method for early warning according to claim 2, which is characterized in that every piece of intelligent electric meter of the determination
Observation mark δi, comprising:
Item determines the observation label δ of every piece of intelligent electric meter according to the observationiIf observation item, δ occur for the intelligent electric meteri
=1；If the intelligent electric meter does not occur to observe item, δi=0.
5. intelligent electric meter batch method for early warning according to claim 1, which is characterized in that further include:
The String data type of the data is mapped as integer data type；
The data are read in into memory in a manner of integer data type.
6. intelligent electric meter batch method for early warning according to claim 1, which is characterized in that
Operation duration in same analysis group is to delete the difference for the median for losing date and installed date.
7. intelligent electric meter batch method for early warning according to claim 1, which is characterized in that the default survival rate threshold value is
0.5。
8. intelligent electric meter batch method for early warning according to claim 1, which is characterized in that the preset time threshold is
600 days.
9. a kind of intelligent electric meter batch early warning system characterized by comprising
First obtains module, for obtaining intelligent electric meter actual production data, wherein the data include Right censored data, institute
State Right censored data include non-mass failure remove intelligent electric meter data and still in the intelligent electric meter data of normal operation；
Setting module, is used for setting model batch, the selected bid batch of the model batch, receipt lot, manufacturer, in phenotype
It is any one or several；
Grouping module, for being grouped according to the model batch of setting to the data, wherein the data in same grouping have
There is identical model batch；
Fitting module carries out Weibull Distribution for the survival function to the data after each grouping, and uses maximum likelihood
Estimation method obtains the parameter Estimation of the Weibull distribution of each groupingWithWherein,WithThe respectively scale of Weibull distribution
Parameter and form parameter；
Second obtains module, for according to parameter EstimationWith default survival rate threshold value, the intelligent electric meter for obtaining each grouping is pre-
Service life phase node；
Warning module, for being greater than the life expectancy node to operation duration in analysis group or apart from the life expectancy node
The batch that time is less than or equal to preset time threshold issues early warning, the data model batch having the same of same described point of grouping
And receipt lot.
10. a kind of computer readable storage medium, which is characterized in that be stored with computer on the computer readable storage medium
Program realizes such as method of any of claims 1-8 when described program is executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910602064.1A CN110261811A (en) | 2019-07-05 | 2019-07-05 | Intelligent electric meter batch method for early warning and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910602064.1A CN110261811A (en) | 2019-07-05 | 2019-07-05 | Intelligent electric meter batch method for early warning and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110261811A true CN110261811A (en) | 2019-09-20 |
Family
ID=67924505
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910602064.1A Pending CN110261811A (en) | 2019-07-05 | 2019-07-05 | Intelligent electric meter batch method for early warning and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110261811A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111998972A (en) * | 2020-09-17 | 2020-11-27 | 南方电网科学研究院有限责任公司 | Terminal block temperature drastic change alarm function detection method, device, terminal and medium |
| CN112381258A (en) * | 2019-11-26 | 2021-02-19 | 国家电网公司 | Meter operation life cycle prediction method and device |
| CN112380759A (en) * | 2019-11-27 | 2021-02-19 | 国家电网公司 | Smart meter service life prediction method based on deep learning and CoxPH model |
| CN114252794A (en) * | 2021-11-24 | 2022-03-29 | 国电南瑞科技股份有限公司 | Method and device for predicting residual life of disassembled intelligent electric energy meter |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009072204A1 (en) * | 2007-12-06 | 2009-06-11 | Japan Electric Meters Inspection Corporation | Life estimating method and device of electronic weighting instrument |
| CN102708306A (en) * | 2012-06-19 | 2012-10-03 | 华北电网有限公司计量中心 | Prediction method for q-precentile life of intelligent meter |
| CN105302941A (en) * | 2015-09-17 | 2016-02-03 | 广州供电局有限公司 | High-voltage single conductor cable service life evaluating method and system |
| CN105320797A (en) * | 2014-08-05 | 2016-02-10 | 南京理工大学 | Method for predicting service life of key system of rail transit vehicles |
| CN107273688A (en) * | 2017-06-16 | 2017-10-20 | 山东万腾电子科技有限公司 | Equipment life and health degree monitoring method and system based on primary fault data |
| CN109492192A (en) * | 2018-09-26 | 2019-03-19 | 西安电子科技大学 | Based on the signature vector method for parameter estimation for deleting mistake lifetime of system data bilateral to II type |
| CN109598353A (en) * | 2018-12-06 | 2019-04-09 | 国网浙江省电力有限公司电力科学研究院 | A kind of recent life-span prediction method of batch electric energy meter |
| CN109598052A (en) * | 2018-11-29 | 2019-04-09 | 武汉大学 | Intelligent electric meter life cycle prediction technique and device based on correlation analysis |
| CN109697525A (en) * | 2018-12-06 | 2019-04-30 | 国网浙江省电力有限公司电力科学研究院 | A kind of recent life-span prediction method of batch electric energy meter multiple faults mode |
| CN110146840A (en) * | 2019-05-23 | 2019-08-20 | 国网浙江省电力有限公司电力科学研究院 | A kind of recent life-span prediction method of batch electric energy meter based on more stress influences |
| CN110991945A (en) * | 2018-09-30 | 2020-04-10 | 北京国双科技有限公司 | Information processing method and device for equipment spare part, storage medium and processor |
-
2019
- 2019-07-05 CN CN201910602064.1A patent/CN110261811A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009072204A1 (en) * | 2007-12-06 | 2009-06-11 | Japan Electric Meters Inspection Corporation | Life estimating method and device of electronic weighting instrument |
| CN102708306A (en) * | 2012-06-19 | 2012-10-03 | 华北电网有限公司计量中心 | Prediction method for q-precentile life of intelligent meter |
| CN105320797A (en) * | 2014-08-05 | 2016-02-10 | 南京理工大学 | Method for predicting service life of key system of rail transit vehicles |
| CN105302941A (en) * | 2015-09-17 | 2016-02-03 | 广州供电局有限公司 | High-voltage single conductor cable service life evaluating method and system |
| CN107273688A (en) * | 2017-06-16 | 2017-10-20 | 山东万腾电子科技有限公司 | Equipment life and health degree monitoring method and system based on primary fault data |
| CN109492192A (en) * | 2018-09-26 | 2019-03-19 | 西安电子科技大学 | Based on the signature vector method for parameter estimation for deleting mistake lifetime of system data bilateral to II type |
| CN110991945A (en) * | 2018-09-30 | 2020-04-10 | 北京国双科技有限公司 | Information processing method and device for equipment spare part, storage medium and processor |
| CN109598052A (en) * | 2018-11-29 | 2019-04-09 | 武汉大学 | Intelligent electric meter life cycle prediction technique and device based on correlation analysis |
| CN109598353A (en) * | 2018-12-06 | 2019-04-09 | 国网浙江省电力有限公司电力科学研究院 | A kind of recent life-span prediction method of batch electric energy meter |
| CN109697525A (en) * | 2018-12-06 | 2019-04-30 | 国网浙江省电力有限公司电力科学研究院 | A kind of recent life-span prediction method of batch electric energy meter multiple faults mode |
| CN110146840A (en) * | 2019-05-23 | 2019-08-20 | 国网浙江省电力有限公司电力科学研究院 | A kind of recent life-span prediction method of batch electric energy meter based on more stress influences |
Non-Patent Citations (3)
| Title |
|---|
| 史道济，张玉环主编: "《应用数理统计》", 30 April 2008 * |
| 张景元等: "基于多应力退化模型的智能电表可靠寿命预估", 《北京航空航天大学学报》 * |
| 段浩，李经源主编: "《金融信息分析》", 30 April 2015, 国防工业出版社 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112381258A (en) * | 2019-11-26 | 2021-02-19 | 国家电网公司 | Meter operation life cycle prediction method and device |
| CN112381258B (en) * | 2019-11-26 | 2024-03-19 | 国家电网公司 | Meter operation life cycle prediction method and device |
| CN112380759A (en) * | 2019-11-27 | 2021-02-19 | 国家电网公司 | Smart meter service life prediction method based on deep learning and CoxPH model |
| CN112380759B (en) * | 2019-11-27 | 2024-01-09 | 国家电网公司 | Smart electric meter service life prediction method based on deep learning and CoxPH model |
| CN111998972A (en) * | 2020-09-17 | 2020-11-27 | 南方电网科学研究院有限责任公司 | Terminal block temperature drastic change alarm function detection method, device, terminal and medium |
| CN111998972B (en) * | 2020-09-17 | 2021-07-16 | 南方电网科学研究院有限责任公司 | Terminal block temperature drastic change alarm function detection method, device, terminal and medium |
| CN114252794A (en) * | 2021-11-24 | 2022-03-29 | 国电南瑞科技股份有限公司 | Method and device for predicting residual life of disassembled intelligent electric energy meter |
| CN114252794B (en) * | 2021-11-24 | 2024-04-09 | 国电南瑞科技股份有限公司 | Method and device for predicting residual life of disassembled intelligent ammeter |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110261811A (en) | Intelligent electric meter batch method for early warning and system | |
| CN103502942B (en) | Virtual machine administration device and virtual machine management method | |
| London et al. | Analysis of measurement-set qualitative characteristics for state-estimation purposes | |
| CN105989155B (en) | Identify the method and device of risk behavior | |
| CN109986994A (en) | Manage the System and method for of interchangeable energy storage device battery switching station | |
| CN106505557B (en) | Remote measurement error identification method and device | |
| CN107688591B (en) | A kind of actuarial treating method and apparatus | |
| CN108154244A (en) | The O＆M methods, devices and systems of real estate power equipment | |
| CN109558400A (en) | Data processing method, device, equipment and storage medium | |
| CN104134999A (en) | Power-distribution-network measurement effectiveness analysis practical calculation method based on multiple data sources | |
| CN107750443A (en) | Use the parallelizing network Business Stream availability modeling of random process and traffic engineered algorithm | |
| CN108710905A (en) | One kind being based on the united spare part quantitative forecasting technique of multi-model and system | |
| CN110347407A (en) | A kind of method, apparatus, computer equipment and medium obtaining EMS memory occupation amount | |
| CN110991871A (en) | Risk monitoring method, device, equipment and computer readable storage medium | |
| CN109752629A (en) | A kind of power grid measurement problem intelligent diagnosing method and system | |
| CN109684320A (en) | The method and apparatus of monitoring data on-line cleaning | |
| CN109670637A (en) | Building energy consumption prediction technique, storage medium, device and system | |
| Bartolini et al. | The robust traveling salesman problem with time windows under knapsack-constrained travel time uncertainty | |
| CN115508770A (en) | KL-NB algorithm-based electric energy meter operation state online evaluation method | |
| CN112700131A (en) | AB test method and device based on artificial intelligence, computer equipment and medium | |
| CN109739684A (en) | The copy restorative procedure and device of distributed key value database based on vector clock | |
| CN110413722A (en) | Address choice method, apparatus and non-transient storage medium | |
| CN109636338A (en) | A kind of broad band photoelectrical operation management platform system | |
| Wang et al. | A Bayesian combination forecasting model for retail supply chain coordination | |
| CN117217505A (en) | Resource management system based on book field |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190920 |