CN109472293A - A kind of grid equipment file data error correction method based on machine learning - Google Patents
A kind of grid equipment file data error correction method based on machine learning Download PDFInfo
- Publication number
- CN109472293A CN109472293A CN201811187606.5A CN201811187606A CN109472293A CN 109472293 A CN109472293 A CN 109472293A CN 201811187606 A CN201811187606 A CN 201811187606A CN 109472293 A CN109472293 A CN 109472293A
- Authority
- CN
- China
- Prior art keywords
- data
- training set
- machine learning
- feature vector
- value
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/211—Selection of the most significant subset of features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The present invention provides a kind of grid equipment file data error correction method based on machine learning, existing mass data is handled, excavates wherein hiding rule, automatically generate judgment rule, data are diagnosed automatically based on these rules, work difficulty can be substantially reduced, can be administered for quality of data screening, data rectification, data and a important foundation is provided;The present invention is administered using big data technology Develop Data, to automatic diagnosis abnormal in mass data, and suggestion is provided for the rectification of data, data check work can be reduced to the strong dependency of business personnel, for complete random extractive decentralized data abnormal conditions, it can also be realized and be automatically processed with machine learning, avoid manpower screening bring complex work amount, the present invention carries out classification analysis to data abnormal problem using big data, data generation side is supplied to be rectified and improved, data problem can be reduced from source, provide reference for data source header rectification.
Description
Technical field
The present invention relates to grid equipment field of data correction, specifically a kind of grid equipment archives number based on machine learning
According to error correction method.
Background technique
Power grid production equipment account data are the bases that power grid production work is carried out, current all kinds of production equipment account data
It is stored in equipment (assets) O&M lean management system (PMS2.0) system, total amount of data is more than 60G, is related to 200
More than equipment, such as: bus, overhead transmission line, switchgear, capacitor, transformer, cable etc..
Teams and groups, base personnel are responsible for being updated device data in time maintenance, field device O&M, maintenance, detection, examination
It every production work such as tests to be both needed to based on device data, only guaranteed device data accuracy, related O&M overhaul business
In the accurate registration PMS2.0 system of record ability, weight is provided for equipment state overhauling evaluation and assets whole-life cycle fee
Will foundation and O&M maintenance lean management important embodiment.In addition, equipment scale be people provide staffing delimit the organizational structure, cost accounting
Important evidence, therefore device data accuracy is particularly important.
There are the problems such as imperfect, inaccurate for power grid production equipment account data at present.Device parameter not only influences extremely
The management of equipment archives itself, while the development of O&M service work is directly influenced, such as:
1. equipment account key parameter is imperfect.
2. equipment account data and GIS graph data be not corresponding.
3. equipment account parameters fill in mistake.
4. devices in system account data have differences with scene.
Wherein problem 3,4 can not carry out wrong data screening by refining mistake rule exploitation program, use people at present
Work means are checked, and every 100 data generally requires the workload of 3 man days of investment.Work difficulty is big, and effect is owed
It is good.These data problems directly affect daily O&M service work, are unable to normal registration O&M record of examination, while also influencing battalion
Development with work such as perforation, same period line losses provides staffing and delimits the organizational structure and O&M cost in addition, Unit account of plant data exception will lead to people
Accounting inaccuracy.
Summary of the invention
In view of the above shortcomings of the prior art, the present invention proposes a kind of grid equipment archives number based on machine learning
According to error correction method, existing mass data is handled, excavates wherein hiding rule, judgment rule is automatically generated, is based on
These rules diagnose data automatically, can substantially reduce work difficulty, can be rectified and improved for quality of data screening, data,
Data, which are administered, provides a important foundation.
A kind of grid equipment file data error correction method based on machine learning, includes the following steps:
Step 1: data pick-up, obtains training set: all grid equipment file datas that company is saved are as history number
According to importing in database, using the historical data in database as training set F;
Step 2: carrying out feature extraction to training set F, characteristic set S=is obtained in such a way that character string is split
{s1,s2,s3,...,sn};
Step 3: selecting characteristic value from characteristic set S manually is selected as feature vector S', S'={ s'1,s'2,
s'3…,s'm, wherein
Step 4: being weighted by TF-IDF algorithm to feature vector S', weighting scheme is the characteristic value s' in feature vectorm
The frequency occurred in training set F, is denoted as Nm, every data record includes Feature Words s' in training setmFrequency be N'm, that
The specific word s'mIDF value be
Therefore the weights omega of the specific wordmIt can be expressed as ωm=Nm*IDF(s'm), in this way to feature vector
Each of S' Feature Words carry out calculating weight, obtain weight vectors ω;
Step 5: the weight of the feature vector chosen by third step and the 4th step and obtained feature vector, to original
Data are clustered by distribution K-Means algorithm, finally divide training set F for k class;
Step 6: it is directed to the cluster result that the 5th step obtains, the data in manual verification's cluster, by abnormal data and erroneous judgement
The case where data are selected, and the data erroneous judgement in each class is verified, obtains the accuracy rate of data verification, to accurate in all classes
Rate averaged obtains the accuracy rate of model, and then whether judgment models accuracy rate reaches expected threshold value, if without if
Third step is gone to, characteristic value is reselected, feature vector, determines weight, until accuracy rate reaches expected threshold value;
Step 7: after model determines, in service stage by data clusters, then by every one kind
Abnormal data returns to user, and normal data is recommended user, by user with reference to modification.
Further, including step 8: Modifying model: in user's use process, correcting mould by the feedback of user
The characteristic value and weight of type.
Further, in the 5th step in clustering algorithm the value of k according to the distance in the sample size and all classes of training set
It is determined with the value of k when being minimum, distance is calculated using Euclidean distance calculation formula to two records i, j in training set F
Distance calculated, Euclidean distance calculation formula is
The present invention is administered using big data technology Develop Data, to automatic diagnosis abnormal in mass data, and is data
Rectification suggestion is provided, data check work can be reduced to the strong dependency of business personnel, refined for completely random
Decentralized data abnormal conditions, can also with machine learning realize automatically process, avoid manpower screening bring complex work amount,
The present invention carries out classification analysis to data abnormal problem using big data, is supplied to data generation side and is rectified and improved, can be from source
Data problem is reduced on head, provides reference for data source header rectification.
Detailed description of the invention
Fig. 1 is the flow diagram of the grid equipment file data error correction method the present invention is based on machine learning;
Fig. 2 is the data prediction flow diagram of clustering phase;
Fig. 3 is K-means parallelization operational process schematic diagram;
Fig. 4 is selection fitted figure of the k value from 20 to 1000 when being clustered by distributed K-Means algorithm, cross in figure
Axis indicates k value, and the longitudinal axis indicates loss function value;
Fig. 5 is using cluster result schematic diagram a kind of when distribution K-Means algorithm.
Specific embodiment
Below in conjunction with the attached drawing in the present invention, the technical solution in the present invention is clearly and completely described.
Due to being all that processing is collected in artificial integration before power grid production equipment account data, inevitably can in these data
There are various mistakes and error, these errors and wrong data will greatly affect field device O&M, maintenance, detection, test
Validity.Therefore by the abnormal data in these data be changed to normal data be very it is necessary to.But due to data
Amount is excessively huge, and it is very difficult for only manually carrying out screening error correction, so, the present invention is used based on Spark's
Distributed K-Means algorithm can greatly reduce workload come automatically processing and correcting to abnormal data.
The embodiment of the present invention provides a kind of process signal of grid equipment file data error correction method based on machine learning
Figure, this method mainly carry out abnormal data and diagnose business, and there are two targets: first is that being diagnosed to be abnormal data;Second is that right
Abnormal data provides improving suggestions.The present invention overall includes three big steps: data pick-up, model construction, model are repaired
Just, overall flow is as shown in Figure 1, the method mainly includes the following steps:
Step 1: data pick-up, obtains training set: all grid equipment file datas that company is saved are as history number
According to importing in database, using the historical data in database as training set F;
Step 2: carrying out feature extraction to training set F, characteristic set S=is obtained in such a way that character string is split
{s1,s2,s3,...,sn};
Step 3: selecting characteristic value from characteristic set S manually is selected as feature vector S', S'={ s'1,s'2,
s'3…,s'm, wherein
Step 4: being weighted by TF-IDF algorithm to feature vector S', weighting scheme is the characteristic value s' in feature vectorm
The frequency occurred in training set F, is denoted as Nm, every data record includes Feature Words s' in training setmFrequency be N'm, that
The specific word s'mIDF value be
Therefore the weights omega of the specific wordmIt can be expressed as ωm=Nm*IDF(s'm), in this way to feature vector
Each of S' Feature Words carry out calculating weight, obtain weight vectors ω;
Step 5: the weight of the feature vector chosen by third step and the 4th step and obtained feature vector, to original
Data are clustered by distribution K-Means algorithm, and the purpose of cluster is exactly in order to which by equipment of the same race, (equipment of the same race is not due to
With personnel record, naming method is also different, so needing to select feature) kind attributes value it is poly- and be one kind, conveniently select
Abnormal data out.The k's when value of k is according to the distance in the sample size and all classes of training set and for minimum in clustering algorithm
Value determination, distance is calculated using Euclidean distance calculation formula to two record i in training set F, and the distance of j is calculated,
Euclidean distance calculation formula is
Finally training set F is divided for k class;
Step 6: it is directed to the cluster result that the 5th step obtains, the data in manual verification's cluster, by abnormal data and erroneous judgement
The case where data are selected, and the data erroneous judgement in each class is verified, obtain data verification accuracy rate (after data point good class, due to
Equipment of the same race, the data of same alike result should differ will not be especially big, if it find that data exception attribute in entire classification
The mode and mean value of data, then the data are it is assumed that be abnormal data), to the accuracy rate averaged in all classes,
The accuracy rate of model is obtained, then whether judgment models accuracy rate reaches expected threshold value (such as 90%), if turning without if
It to third step, reselects characteristic value, feature vector, determine weight, until accuracy rate reaches expected threshold value.
Step 7:, in service stage by data clusters, the abnormal data of every one kind is then returned to use after model determines
Family, and normal data is recommended into user, by user with reference to modification;
Step 8: Modifying model
In user's use process, by the feedback of user come the characteristic value and weight of correction model.No matter machine learning
Learning tasks using which kind of algorithm, model evaluation is all a kind of link of end-to-end machine learning assembly line.Monitor mould
Type algorithm shows in a production environment, and objective evaluation model accuracy is related to diagnosis effect, customer experience, user feedback etc. to be referred to
Mark, by adjusting model and parameter and then Optimized model algorithm.
In above-mentioned 5th step, 2 stages can be divided by running K-means algorithm parallel in Spark cluster: data are located in advance
Reason stage and K-means clustering phase.Data preprocessing phase process is as shown in Figure 2.
In clustering phase, after pretreatment stage, data set has met the requirement of cluster, as long as therefore pre- place
Result after reason calculates k cluster centre using K-means algorithm, this K point can gather as k of entire data set
Class center.Clustering phase parallelization process is as shown in Figure 3:
Technical solution of the present invention is described in detail with a specific example below:
Problem description:
In actual electric network work is carried out, same equipment is often in high volume purchased, and is used in batches, in systems should be with
Certain amount grade exists, it is impossible to only occurs once, and the corresponding particular attribute-value of same equipment should be consistent,
In equipment files data, model can judge the equipment specified genus as the unique identification for identifying the equipment according to model
Whether the value of property is correct.
1 experiment sample tables of data of table
Table 2 Sample data table
At present it is mainly following problems in data set:
1. model is filled in and lack of standardization can not be identified
Teams and groups, base personnel are responsible for device data typing, since each teams and groups, base personnel count according to oneself habit
According to filling in, the equipment of same model is caused all to be rendered as difference in " model " field.Such as S11-M-100/10 in upper table,
S11-100/10, S11-100, S11-100KVA are same equipment in fact, but filled in inside database it is different, it is same
Kind model may have tens kinds to fill in mode.
2. same model corresponds to a variety of attribute values
Mistake is filled in often when the personnel's logging data of teams and groups, base, and data user of service does not know the feelings of field device
Condition, this, which directly results in many business, to carry out.Often the rated capacity of the equipment of same model occurs in database at present
A variety of values.
Specific implementation:
1. the extraction of characteristic value
Feature selecting occupies considerable status in machine learning.Feature is extracted from " model " character string text,
Such as " model " " S11-M-100/10 " can be extracted as totally 5 features of S, 11, M, 100,10 by character fractionation.This 5 kinds of spies
It levies and is in altogetherKind combination.In this experiment, by test repeatedly and and true controls, screened from these features
Unrelated out or redundancy feature retains a character subset after removing it.
2. the weight of characteristic value
Due to extracting feature in " model " text, each feature is different to the percentage contribution of classification, therefore is transporting
With being weighted before these features.Weighted value shared by each feature cannot treat different things as the same.It is used herein TF-IDF weighting
Method, TF-IDF is a kind of statistical method, to assess words for wherein one in a file set or a corpus
The significance level of part file.The importance of words, but simultaneously can be with the directly proportional increase of number that it occurs hereof
The frequency that it occurs in corpus is inversely proportional decline.
The selection of 3.k value
Selection for k value, we are based on the common ancon method of industry, a function are defined, with the change of k, it is believed that
Extreme value can be generated in correct k.That is: a suitable class cluster index, such as mean radius or diameter are given, as long as we are false
If the number of class cluster when being at or above the number of true class cluster, which rises can be very slow, and once attempts
To be less than true number class cluster when, which can steeply rise.Such as in this experiment, setting k ∈ (20,1000) is opened from 20
Begin, each k value increases by 98, and loss function and k value relation curve are as shown in Figure 4.
According to cluster result, subsequent processing is carried out to data in conjunction with business, due to " model " can unique identification one kind set
It is standby, and equipment is also equipped with many other attributes, such as " rated capacity ", " voltage class ", " dielectric " etc., it is same to set
Standby attribute value is unique.It is introduced for carrying out error correction to " rated capacity " herein, obtains " correct data ", " exception
Data ", the error correction result of " wrong data ", analyze wrong data therein, abnormal data, provide amending advice, real
Existing intelligent diagnostics.Specific analytic process is as follows:
(1) result after cluster is counted.It is most based on correct data frequency of occurrence in same class data,
Wrong data belongs to a small number of principles, it is believed that the most group of " model " and " rated capacity " frequency of occurrence is combined into the positive exact figures of recommendation
According to.
(2) when " model " and " rated capacity " is inconsistent with the correct data of recommendation, it is judged as wrong data.
(3) when " model " and " rated capacity " is inconsistent with a certain item of correct data of recommendation, it is judged as abnormal data.
It is correct data when " model " and " rated capacity " is completely the same with the data format of recommendation.
The effect experiment figure of cluster is as shown in Figure 5.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Belong to those skilled in the art in the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of, all answers
It is included within the scope of the present invention.Therefore, protection scope of the present invention should be subject to the protection scope in claims.
Claims (3)
1. a kind of grid equipment file data error correction method based on machine learning, it is characterised in that include the following steps:
Step 1: data pick-up, obtains training set: all grid equipment file datas that company saves are led as historical data
Enter in database, using the historical data in database as training set F;
Step 2: carrying out feature extraction to training set F, characteristic set S={ s is obtained in such a way that character string is split1,
s2,s3,...,sn};
Step 3: selecting characteristic value from characteristic set S manually is selected as feature vector S', S'={ s '1,s′2,s
′3…,s′m, wherein
Step 4: being weighted by TF-IDF algorithm to feature vector S', weighting scheme is the characteristic value s' in feature vectormIt is instructing
Practice the frequency occurred in collection F, is denoted as Nm, every data record includes Feature Words s' in training setmFrequency be N'm, then should
Feature Words s'mIDF value be
Therefore the weights omega of the specific wordmIt can be expressed as ωm=Nm*IDF(s'm), in this way in feature vector S'
Each Feature Words carry out calculating weight, obtain weight vectors ω;
Step 5: the weight of the feature vector chosen by third step and the 4th step and obtained feature vector, to initial data
It is clustered by distributed K-Means algorithm, is finally divided training set F for k class;
Step 6: it is directed to the cluster result that the 5th step obtains, the data in manual verification's cluster, by abnormal data and erroneous judgement data
The case where selecting, verifying the data erroneous judgement in each class, obtains the accuracy rate of data verification, asks the accuracy rate in all classes
It is averaged, obtains the accuracy rate of model, then whether judgment models accuracy rate reaches expected threshold value, goes to if not
Third step reselects characteristic value, feature vector, determines weight, until accuracy rate reaches expected threshold value;
Step 7:, in service stage by data clusters, the abnormal data of every one kind is then returned into user after model determines,
And normal data is recommended into user, by user with reference to modification.
2. the grid equipment file data error correction method based on machine learning as described in claim 1, it is characterised in that: also wrap
It includes step 8: Modifying model: in user's use process, by the feedback of user come the characteristic value and weight of correction model.
3. the grid equipment file data error correction method based on machine learning as described in claim 1, it is characterised in that: the 5th
The value of k when the value of k is according to the distance in the sample size and all classes of training set and for minimum in clustering algorithm in step determines
, distance calculate using Euclidean distance calculation formula to two record i in training set F, the distance of j is calculated, it is European away from
It is from calculation formula
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811187606.5A CN109472293A (en) | 2018-10-12 | 2018-10-12 | A kind of grid equipment file data error correction method based on machine learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811187606.5A CN109472293A (en) | 2018-10-12 | 2018-10-12 | A kind of grid equipment file data error correction method based on machine learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109472293A true CN109472293A (en) | 2019-03-15 |
Family
ID=65663800
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811187606.5A Pending CN109472293A (en) | 2018-10-12 | 2018-10-12 | A kind of grid equipment file data error correction method based on machine learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109472293A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110189575A (en) * | 2019-06-27 | 2019-08-30 | 广东电网有限责任公司肇庆供电局 | A kind of distribution O&M simulation training system based on big data |
CN110826605A (en) * | 2019-10-24 | 2020-02-21 | 北京明略软件系统有限公司 | Method and device for identifying user in cross-platform manner |
CN111123039A (en) * | 2019-12-31 | 2020-05-08 | 国网北京市电力公司 | Power distribution network abnormal line loss diagnosis method based on synchronization characteristics and improved K-means clustering |
CN111159169A (en) * | 2019-12-31 | 2020-05-15 | 中国联合网络通信集团有限公司 | Data management method and equipment |
CN111625991A (en) * | 2020-05-20 | 2020-09-04 | 国网河北省电力有限公司电力科学研究院 | Low-voltage distribution network topology verification method |
CN114154561A (en) * | 2021-11-15 | 2022-03-08 | 国家电网有限公司 | Electric power data management method based on natural language processing and random forest |
EP3985569A1 (en) * | 2020-10-16 | 2022-04-20 | Fujitsu Limited | Information processing program, information processing method, and information processing device |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103001825A (en) * | 2012-11-15 | 2013-03-27 | 中国科学院计算机网络信息中心 | Method and system for detecting DNS (domain name system) traffic abnormality |
CN103810264A (en) * | 2014-01-27 | 2014-05-21 | 西安理工大学 | Webpage text classification method based on feature selection |
CN106604267A (en) * | 2017-02-21 | 2017-04-26 | 重庆邮电大学 | Dynamic self-adapting wireless sensor network invasion detection intelligence algorithm |
CN106778259A (en) * | 2016-12-28 | 2017-05-31 | 北京明朝万达科技股份有限公司 | A kind of abnormal behaviour based on big data machine learning finds method and system |
CN107426177A (en) * | 2017-06-13 | 2017-12-01 | 努比亚技术有限公司 | A kind of user behavior clustering method and terminal, computer-readable recording medium |
CN107679734A (en) * | 2017-09-27 | 2018-02-09 | 成都四方伟业软件股份有限公司 | It is a kind of to be used for the method and system without label data classification prediction |
-
2018
- 2018-10-12 CN CN201811187606.5A patent/CN109472293A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103001825A (en) * | 2012-11-15 | 2013-03-27 | 中国科学院计算机网络信息中心 | Method and system for detecting DNS (domain name system) traffic abnormality |
CN103810264A (en) * | 2014-01-27 | 2014-05-21 | 西安理工大学 | Webpage text classification method based on feature selection |
CN106778259A (en) * | 2016-12-28 | 2017-05-31 | 北京明朝万达科技股份有限公司 | A kind of abnormal behaviour based on big data machine learning finds method and system |
CN106604267A (en) * | 2017-02-21 | 2017-04-26 | 重庆邮电大学 | Dynamic self-adapting wireless sensor network invasion detection intelligence algorithm |
CN107426177A (en) * | 2017-06-13 | 2017-12-01 | 努比亚技术有限公司 | A kind of user behavior clustering method and terminal, computer-readable recording medium |
CN107679734A (en) * | 2017-09-27 | 2018-02-09 | 成都四方伟业软件股份有限公司 | It is a kind of to be used for the method and system without label data classification prediction |
Non-Patent Citations (4)
Title |
---|
李杰 等: "《CiteSpace科技文本挖掘及可视化》", 31 January 2016 * |
许鑫: "《基于文本特征计算的信息分析方法》", 30 November 2015 * |
赵杰: "《搜索引擎技术》", 30 November 2007 * |
龙婧 等: "基于机器学习的电网设备档案数据异常诊断研究", 《电力信息与通信技术》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110189575A (en) * | 2019-06-27 | 2019-08-30 | 广东电网有限责任公司肇庆供电局 | A kind of distribution O&M simulation training system based on big data |
CN110189575B (en) * | 2019-06-27 | 2021-01-05 | 广东电网有限责任公司肇庆供电局 | Big data-based distribution network operation and maintenance simulation training system |
CN110826605A (en) * | 2019-10-24 | 2020-02-21 | 北京明略软件系统有限公司 | Method and device for identifying user in cross-platform manner |
CN111123039A (en) * | 2019-12-31 | 2020-05-08 | 国网北京市电力公司 | Power distribution network abnormal line loss diagnosis method based on synchronization characteristics and improved K-means clustering |
CN111159169A (en) * | 2019-12-31 | 2020-05-15 | 中国联合网络通信集团有限公司 | Data management method and equipment |
CN111625991A (en) * | 2020-05-20 | 2020-09-04 | 国网河北省电力有限公司电力科学研究院 | Low-voltage distribution network topology verification method |
EP3985569A1 (en) * | 2020-10-16 | 2022-04-20 | Fujitsu Limited | Information processing program, information processing method, and information processing device |
CN114154561A (en) * | 2021-11-15 | 2022-03-08 | 国家电网有限公司 | Electric power data management method based on natural language processing and random forest |
CN114154561B (en) * | 2021-11-15 | 2024-02-27 | 国家电网有限公司 | Electric power data management method based on natural language processing and random forest |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109472293A (en) | A kind of grid equipment file data error correction method based on machine learning | |
CN107358366B (en) | Distribution transformer fault risk monitoring method and system | |
CN112859822B (en) | Equipment health analysis and fault diagnosis method and system based on artificial intelligence | |
CN105677791B (en) | For analyzing the method and system of the operation data of wind power generating set | |
CN111460167A (en) | Method for positioning pollution discharge object based on knowledge graph and related equipment | |
CN107491381A (en) | A kind of equipment condition monitoring quality of data evaluating system | |
CN106447210B (en) | A kind of distribution net equipment health degree dynamic diagnosis method of meter and trust evaluation | |
CN107368957A (en) | A kind of construction method of equipment condition monitoring quality of data evaluation and test system | |
CN109061341A (en) | Kalman filtering transformer fault prediction technique and system neural network based | |
CN109544399B (en) | Power transmission equipment state evaluation method and device based on multi-source heterogeneous data | |
CN109858886B (en) | Integrated learning-based cost control success rate promotion analysis method | |
CN111199361A (en) | Electric power information system health assessment method and system based on fuzzy reasoning theory | |
CN108647707B (en) | Probabilistic neural network creation method, failure diagnosis method and apparatus, and storage medium | |
CN110297207A (en) | Method for diagnosing faults, system and the electronic device of intelligent electric meter | |
CN110865924B (en) | Health degree diagnosis method and health diagnosis framework for internal server of power information system | |
CN105471647B (en) | A kind of power communication network fault positioning method | |
CN109101483A (en) | A kind of wrong identification method for electric inspection process text | |
CN110826228B (en) | Regional power grid operation quality limit evaluation method | |
CN112464995A (en) | Power grid distribution transformer fault diagnosis method and system based on decision tree algorithm | |
CN109542742A (en) | Database server hardware health evaluating method based on expert model | |
CN112818008A (en) | Intelligent diagnosis method, system, medium and electronic equipment for nuclear power debugging faults | |
CN110968703B (en) | Method and system for constructing abnormal metering point knowledge base based on LSTM end-to-end extraction algorithm | |
CN112560214A (en) | Transformer substation bus balance fault diagnosis method, system, equipment and storage medium | |
CN115587309A (en) | Method, device and equipment for extracting key features of short-circuit resistance of transformer | |
CN111179576B (en) | Power utilization information acquisition fault diagnosis method and system with inductive learning function |
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: 20190315 |