CN103745417A - Power grid load characteristic curve analysis method - Google Patents
Power grid load characteristic curve analysis method Download PDFInfo
- Publication number
- CN103745417A CN103745417A CN201410037326.1A CN201410037326A CN103745417A CN 103745417 A CN103745417 A CN 103745417A CN 201410037326 A CN201410037326 A CN 201410037326A CN 103745417 A CN103745417 A CN 103745417A
- Authority
- CN
- China
- Prior art keywords
- data
- user
- characteristic curve
- threshold value
- class
- 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
Images
Abstract
The invention relates to a big data-based power grid load characteristic curve analysis method. The method comprises the steps: average daily 80,000 data are subjected to elimination, repair and smoothing processing; taking public transformer data as an example, by comparing data of 96 collection points everyday, constructing a function model, and calculating the correlation of electrical loads of different users; carrying out classifying according to a calculated correlation coefficient; by a preset threshold value rho[0], classifying the users with correlation higher than rho[0] as one category; sequencing the obtained categories from big to small; according to a preset threshold value m, classifying the categories of which the element number is larger than m again, wherein the adopted method is same as the abovementioned method, and thus the daily characteristic curves of the public transformer users are concluded; carrying out correlation measurement on the characteristic curves, and classifying the characteristic curves with correlation higher than rho[0] as one category, and finally obtaining the annual load characteristic curves of the public transformer users. In a similar way, annual load characteristic curves of private transformer users and individual line (scheduling) users can be obtained.
Description
Technical field
The present invention relates to a kind of network load feature curve analysis method, particularly a kind of analytical approach of the network load characteristic curve based on large data.
Background technology
In electric system, network load specificity analysis is as the basic work of research, analysis electricity market, being related to the construction of electrical network, the supply programme of electric power, is also the foundation of Economic Dispatch operation, formulation measures to regulate rush-hour traffic, alleviation power supply shortage situation.The analysis of load characteristic curve is the important content of network load specificity analysis.Along with developing rapidly of strong intelligent grid, infotech is just with unprecedented range, the degree of depth and electrical production, business administration rapid fusion, and infosystem has become " nervous centralis " of intelligent grid, supports electrical network production and administration development of new generation.At present, State Grid Corporation of China has tentatively built up domestically leading, world-class information integrated platform, and electrical network business datum all begins to take shape from total amount and kind, for Load Characteristic Analysis provides data basis.
How analyzing and utilizing these data is current a great problems, existing network load characteristic curve is mostly studied according to industry, but research shows, different enterprises in same industry are with often also having larger difference, so the load characteristic curve based on trade classification can not reflect consumption habit and feature that enterprise customer is real exactly on electrical feature.In order to solve an above-mentioned difficult problem, the invention provides a kind of analytical approach of the network load characteristic curve based on large data.
The network load feature curve analysis method based on data mining and analysis under large data background that the present invention proposes, based on magnanimity Power system load data maintenance data, excavate with the method for analyzing and draw public change, specially become, the load characteristic curve of individual line subscriber, calculated load characteristic curve is with the related coefficient of each user's load curve again, and this user is included in the load characteristic curve classification that associated coefficient is the highest, thereby can be targetedly analyzing by electrical characteristics user, for user's rational utilization of electricity, the proposition of the related economic of economy operation of power grid and technical measures and policy provides foundation.
Summary of the invention
The object of the invention is to provide according to the weak point of existing network load characteristic curve a kind of analytical approach of the network load characteristic curve based on large data.
The present invention includes following steps:
(1) collect the daily load data of monitor user ' in electrical network user power utilization information acquisition system and dispatching system, according to class of subscriber, be divided into and publicly become user data, specially become user data, special line (scheduling) user data;
(2) rejecting abnormalities data;
(3) adopt data recovery technique that 48 load datas that gather time point are extended for to 96 collection time point load datas;
(4) adopt data smoothing technology to carry out smoothing processing to data;
(5) public affairs are become to user data by comparison 96 collection time point data every day, build public similarity function model, statistical test and the related coefficient that becomes customer charge curve and differentiate;
(6) choose wherein one and calculate remaining pieces of data and its related coefficient and set a threshold value for benchmark, related coefficient is greater than to the class that is classified as of threshold value;
(7) adopt similar method repeating step (6), total data is sorted out;
(8) for all classes, according to the size order of class, sort, and set the threshold value of class, choose higher than the class of threshold value and carry out data smoothing processing;
(9) for the category feature curve of the threshold value higher than class, the method for repeating step (6) is sorted out, and obtains public affairs and becomes user's characteristic curve of every day;
(10) for public affairs, become the characteristic curve of user every day and analyze, the characteristic curve of choosing some day is benchmark, calculates remaining characteristic curve and its related coefficient, gets the class that is classified as that related coefficient is greater than threshold value, repeats above-mentioned steps until sort out end;
(11) reject the data that do not meet day characteristic curve, obtain public affairs and become user's load characteristic curve of the whole year;
(12) for special change user, special line (scheduling) user, adopt to use the same method and analyze, repeating step (5) ~ (11), obtain specially becoming user, special line (scheduling) the user load characteristic curve of the whole year;
(13) set up mathematical model, use Fourier to convert the linear combination that certain function table that meets certain condition is shown as to trigonometric function or their integration, obtain the function expression of annual load characteristic curve, express the undulatory property of user power utilization.
Described data smoothing disposal route is centered by this data collection point, before and after it, respectively gets two collection points, with the mean value of these five collection point data, replaces former collection point data.
Further, the described threshold value of above-mentioned steps (6) is 0.8 or 0.9.
Further, the threshold value of the class described in above-mentioned steps (8) is 50.
Compared with prior art, the present invention has following beneficial effect: the present invention is the analytical approach that a kind of network load characteristic curve based on large data is provided according to the weak point of existing network load characteristic curve, thereby can be targetedly to the analyzing by electrical characteristics of user, for the related economic of user's rational utilization of electricity, economy operation of power grid and the proposition of technical measures and policy provide foundation.
Below in conjunction with the drawings and specific embodiments, the present invention will be further described in detail.
Accompanying drawing explanation
Figure 1 shows that the process flow diagram of the analytical approach of the network load characteristic curve based on large data of the present invention.
Figure 2 shows that the analytical approach of the network load characteristic curve based on large data of the present invention sets up the process flow diagram of mathematical model.
Embodiment
Embodiment mono-: a kind of network load feature curve analysis method, described method comprises the steps:
(1) collect the daily load data of monitor user ' in electrical network user power utilization information acquisition system and dispatching system, according to class of subscriber, be divided into and publicly become user data, specially become user data, special line (scheduling) user data;
(2) rejecting abnormalities data;
(3) adopt data recovery technique that 48 load datas that gather time point are extended for to 96 collection time point load datas;
(4) adopt data smoothing technology to carry out smoothing processing to data;
(5) public affairs are become to user data by comparison 96 collection time point data every day, build public similarity function model, statistical test and the related coefficient that becomes customer charge curve and differentiate;
(6) choose wherein one and calculate remaining pieces of data and its related coefficient and set a threshold value for benchmark, related coefficient is greater than to the class that is classified as of threshold value;
(7) adopt similar method repeating step (6), total data is sorted out;
(8) for all classes, according to the size order of class, sort, and set the threshold value of class, choose higher than the class of threshold value and carry out data smoothing processing;
(9) for the category feature curve of the threshold value higher than class, the method for repeating step (6) is sorted out, and obtains public affairs and becomes user's characteristic curve of every day;
(10) for public affairs, become the characteristic curve of user every day and analyze, the characteristic curve of choosing some day is benchmark, calculates remaining characteristic curve and its related coefficient, gets the class that is classified as that related coefficient is greater than threshold value, repeats above-mentioned steps until sort out end;
(11) reject the data that do not meet day characteristic curve, obtain public affairs and become user's load characteristic curve of the whole year;
(12) for special change user, special line (scheduling) user, adopt to use the same method and analyze, repeating step (5)-(11), obtain specially becoming user, special line (scheduling) the user load characteristic curve of the whole year;
(13) set up mathematical model, use Fourier to convert the linear combination that certain function table that meets certain condition is shown as to trigonometric function or their integration, obtain the function expression of annual load characteristic curve, express the undulatory property of user power utilization.
In conjunction with Fig. 1 and Fig. 2, utilize mathematic(al) representation to be described in detail as follows:
(1) obtain in electrical network user power utilization information acquisition system and dispatching system can monitor user ' daily load data, according to class of subscriber, be divided into and publicly become user data, specially become user data, special line (scheduling) user data;
(2), for average daily nearly 80000 data (wherein public affairs are altered an agreement 42000, specially alter an agreement 36000, approximately 300, special line), some abnormal datas (as being the data of " 0 " because of cause whole that have a power failure) are rejected;
(3) adopt data recovery technique that 48 load datas that gather time point are filled to 96 collection time point load datas;
(4) adopt data smoothing technology to carry out smoothly some abnormal datas, (numerical value that might as well establish this collection point is centered by can this data collection point
), before and after it, respectively get two collection points and (be respectively
with
), with the mean value of these five collection point data
replace former collection point data, wherein:
;
(5) for above nearly 80000 data, with public affairs, become data instance, by comparison 96 collection time point data every day, reflect the similarity of load curve, and build following function model:
And build statistical test and related coefficient and differentiate:
,
, wherein,
for the residual sum of squares (RSS) of fitting of a polynomial,
for residual sum of square.
Choose wherein one and calculate remaining pieces of data and its related coefficient for benchmark, get related coefficient and be greater than
, they are classified as to a class, wherein
for default threshold value, generally get
or
;
(6) adopt similar method repeating step (5), total data is sorted out.For all classes, according to the size order of class, sort, and before choosing
class is carried out data smoothing processing (because the element that class below comprises very little, generally not considering), wherein
for default threshold value, generally get
, concrete smoothing technique is shown in step (4);
(7) for front
category feature curve, the method for repeating step (5) is sorted out, and so can summarize public affairs and become user's characteristic curve of every day;
(8) for public affairs, become the characteristic curve of user every day and analyze, the characteristic curve of choosing some day is benchmark, calculates remaining characteristic curve and its related coefficient, gets related coefficient and is greater than
be classified as a class, repeat above-mentioned steps until sort out and finish, as for group (as element number in class is less than 30), owing to can not reflecting annual part throttle characteristics, should give rejecting, obtain thus annual load characteristic curve;
(9) for special change user, adopting uses the same method analyzes, and repeating step (5) ~ (8) so far obtain public become, specially becoming and the load characteristic curve of special line (scheduling) user whole year;
Wherein, set up the concrete steps following (referring to Fig. 2) of mathematical model:
(10) for public affairs, become first kind user, by analysis before, first draw out its annual load characteristic curve, then use the instruments such as Matlab to carry out higher order polynomial matching, function expression is as follows:
(11) for superior function, carry out fourier conversion, result is as follows:
。
The foregoing is only preferred embodiment of the present invention, all equalizations of doing according to the present patent application the scope of the claims change and modify, and all should belong to covering scope of the present invention.
Claims (4)
1. a network load feature curve analysis method, is characterized in that: described method comprises the steps:
(1) collect the daily load data of monitor user ' in electrical network user power utilization information acquisition system and dispatching system, according to class of subscriber, be divided into and publicly become user data, specially become user data, special line (scheduling) user data;
(2) rejecting abnormalities data;
(3) adopt data recovery technique that 48 load datas that gather time point are extended for to 96 collection time point load datas;
(4) adopt data smoothing technology to carry out smoothing processing to data;
(5) public affairs are become to user data by comparison 96 collection time point data every day, build public similarity function model, statistical test and the related coefficient that becomes customer charge curve and differentiate;
(6) choose wherein one and calculate remaining pieces of data and its related coefficient and set a threshold value for benchmark, related coefficient is greater than to the class that is classified as of threshold value;
(7) adopt similar method repeating step (6), total data is sorted out;
(8) for all classes, according to the size order of class, sort, and set the threshold value of class, choose higher than the class of threshold value and carry out data smoothing processing;
(9) for the category feature curve of the threshold value higher than class, the method for repeating step (6) is sorted out, and obtains public affairs and becomes user's characteristic curve of every day;
(10) for public affairs, become the characteristic curve of user every day and analyze, the characteristic curve of choosing some day is benchmark, calculates remaining characteristic curve and its related coefficient, gets the class that is classified as that related coefficient is greater than threshold value, repeats above-mentioned steps until sort out end;
(11) reject the data that do not meet day characteristic curve, obtain public affairs and become user's load characteristic curve of the whole year;
(12) for special change user, special line (scheduling) user, adopt to use the same method and analyze, repeating step (5) ~ (11), obtain specially becoming user, special line (scheduling) the user load characteristic curve of the whole year;
(13) set up mathematical model, use Fourier to convert the linear combination that certain function table that meets certain condition is shown as to trigonometric function or their integration, obtain the function expression of annual load characteristic curve, express the undulatory property of user power utilization.
2. network load feature curve analysis method according to claim 1, it is characterized in that: described data smoothing disposal route is for centered by this data collection point, before and after it, respectively get two collection points, with the mean value of these five collection point data, replace former collection point data.
3. network load feature curve analysis method according to claim 1, is characterized in that: the described threshold value of step (6) is 0.8 or 0.9.
4. network load feature curve analysis method according to claim 1, is characterized in that: the threshold value of the class described in step (8) is 50.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410037326.1A CN103745417A (en) | 2014-01-27 | 2014-01-27 | Power grid load characteristic curve analysis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410037326.1A CN103745417A (en) | 2014-01-27 | 2014-01-27 | Power grid load characteristic curve analysis method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103745417A true CN103745417A (en) | 2014-04-23 |
Family
ID=50502432
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410037326.1A Pending CN103745417A (en) | 2014-01-27 | 2014-01-27 | Power grid load characteristic curve analysis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103745417A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104834978A (en) * | 2015-05-21 | 2015-08-12 | 国家电网公司 | Load restoration and prediction method |
| CN107870893A (en) * | 2017-10-24 | 2018-04-03 | 顺特电气设备有限公司 | A kind of daily load similitude quantitative analysis method of intelligent transformer |
| CN111667135A (en) * | 2020-03-25 | 2020-09-15 | 国网天津市电力公司 | Load structure analysis method based on typical feature extraction |
| CN113033870A (en) * | 2021-02-25 | 2021-06-25 | 国网河北省电力有限公司营销服务中心 | Flexible load scheduling method for power special transformer customer and terminal equipment |
-
2014
- 2014-01-27 CN CN201410037326.1A patent/CN103745417A/en active Pending
Non-Patent Citations (2)
| Title |
|---|
| 刘莉 等: "k-means聚类算法在负荷曲线分类中的应用", 《电力系统保护与控制》 * |
| 蒋雯倩: "面向负荷建模的变电站日负荷曲线在线分类方法及应用", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104834978A (en) * | 2015-05-21 | 2015-08-12 | 国家电网公司 | Load restoration and prediction method |
| CN107870893A (en) * | 2017-10-24 | 2018-04-03 | 顺特电气设备有限公司 | A kind of daily load similitude quantitative analysis method of intelligent transformer |
| CN111667135A (en) * | 2020-03-25 | 2020-09-15 | 国网天津市电力公司 | Load structure analysis method based on typical feature extraction |
| CN111667135B (en) * | 2020-03-25 | 2023-07-28 | 国网天津市电力公司 | Load structure analysis method based on typical feature extraction |
| CN113033870A (en) * | 2021-02-25 | 2021-06-25 | 国网河北省电力有限公司营销服务中心 | Flexible load scheduling method for power special transformer customer and terminal equipment |
| CN113033870B (en) * | 2021-02-25 | 2022-04-26 | 国网河北省电力有限公司营销服务中心 | Flexible load scheduling method for power special transformer customer and terminal equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106022509B (en) | Consider the Spatial Load Forecasting For Distribution method of region and load character double differences | |
| CN106505593B (en) | A kind of analysis of distribution transforming three-phase imbalance and the method for load adjustment based on big data | |
| CN110991786B (en) | 10kV static load model parameter identification method based on similar daily load curve | |
| WO2018082523A1 (en) | Load cycle mode identification method | |
| CN107038167A (en) | Big data excavating analysis system and its analysis method based on model evaluation | |
| CN105389636A (en) | Low-voltage area KFCM-SVR reasonable line loss prediction method | |
| CN109118067A (en) | A kind of Renewable Energy Development potential evaluation method | |
| CN103049651A (en) | Method and device used for power load aggregation | |
| CN106372747B (en) | Random forest-based reasonable line loss rate estimation method for transformer area | |
| CN110111024A (en) | Scientific and technological achievement market valuation method based on AHP model of fuzzy synthetic evaluation | |
| CN103793788B (en) | A kind of ordered electric management method | |
| CN103942606A (en) | Residential electricity consumption customer segmentation method based on fruit fly intelligent optimization algorithm | |
| CN104809658A (en) | Method for rapidly analyzing low-voltage distributing network area line loss | |
| Quintana et al. | Islands of misfit buildings: Detecting uncharacteristic electricity use behavior using load shape clustering | |
| CN111242161B (en) | Non-invasive non-resident user load identification method based on intelligent learning | |
| CN103745417A (en) | Power grid load characteristic curve analysis method | |
| CN103632306A (en) | Distribution network power supply area division method based on clustering analysis | |
| CN107832927A (en) | 10kV circuits line based on gray relative analysis method becomes relation evaluation method | |
| CN105809573A (en) | Big data analysis based load nature authentication method | |
| CN110188221B (en) | Shape distance-based load curve hierarchical clustering method | |
| CN109003010A (en) | The evaluation method of power grid project investment benefit | |
| CN104484412A (en) | Big data analysis system based on multiform processing | |
| CN106526425A (en) | Low-voltage diagnosis method based on real collected data of intelligent electric meter | |
| CN108122173A (en) | A kind of conglomerate load forecasting method based on depth belief network | |
| CN109034241A (en) | Load cluster control method and system based on support vector machines |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140423 |
|
| RJ01 | Rejection of invention patent application after publication |