CN104063480A - Load curve parallel clustering method based on big data of electric power - Google Patents
Load curve parallel clustering method based on big data of electric power Download PDFInfo
- Publication number
- CN104063480A CN104063480A CN201410312530.XA CN201410312530A CN104063480A CN 104063480 A CN104063480 A CN 104063480A CN 201410312530 A CN201410312530 A CN 201410312530A CN 104063480 A CN104063480 A CN 104063480A
- Authority
- CN
- China
- Prior art keywords
- load
- load curve
- curve
- cluster
- electric power
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a load curve parallel clustering method based on big data of electric power. Wavelet denoising is performed on all load curves through a dbN wavelet system in order to reduce effect of small fluctuation in the curves on clustering results, a K means clustering algorithm based on the multi-core parallel technology is adopted to perform load curve clustering, the clustering results with obvious features are screened out, and load curve classification is obtained through confluence analysis. By means of the load curve parallel clustering method based on big data of electric power, the parallel clustering algorithm of a great number of the load curves is achieved, clustering speed of the load curves is increased effectively, and a foundation is laid for load and electricity quantity predication.
Description
Technical field
The present invention relates to a kind of load curve parallel clustering method based on the large data of electric power, belong to power marketing intelligent use technical field.
Background technology
The infosystems such as power marketing, production, scheduling have produced the power information data of magnanimity, only Jiangsu extraction system, every 96 power consumption data that gather day by day the whole province's more than 3,000 ten thousand resident's daily load electric weight and more than 20 ten thousand negative control large users, amount to GB more than 30, since two thousand six the power information data of accumulation reach more than 39TB.Electric power large data age in Jiangsu arrives already, but how to control the data message of magnanimity like this, therefrom obtains Useful Information, excavates potential value, is challenge and opportunity that Jiangsu electric power faces.
Power system load modeling is the important foundation that electric system simulation is analyzed, and the accuracy of load modeling is directly connected to confidence level and the accuracy of simulation calculation.Load modeling need to be based upon on the basis that Characteristics of Electric Load is fully analyzed, and in the face of the load data of magnanimity in Jiangsu Province's electricity consumption acquisition system, impossible to each user's part throttle characteristics analysis, therefore be necessary user to carry out load characteristics clustering, analyze different user group's part throttle characteristics according to cluster result.
K-means clustering algorithm is simple, easy and simple to handle, is one of current the most widely used clustering method.Its first selected one group of initial cluster center, makes to keep between class to keep closely in independent, class by iteration, during iteration using the average of all data samples in cluster subset as Lei center.The fields such as current K-means clustering algorithm is cut apart at figure, client segmentation, load characteristics clustering are widely used.
But under the background of the large data of power information, number need to carry out cluster analysis with trillion daily load curve, traditional K-means clustering algorithm calculates and is difficult to deal with.In the desktop computer of main flow, notebook computer cpu, all comprise multiple cores at present, therefore can make full use of existing hardware resource, the K-means clustering algorithm of exploitation based on multi-core parallel concurrent technology, to accelerate the speed of load characteristics clustering, improves the efficiency of load characteristics clustering.
Summary of the invention
Technical matters to be solved by this invention is for the Load Characteristic Analysis with under TV university data background, provides a kind of K-means clustering algorithm based on multi-core parallel concurrent technology, for improving the speed of load characteristics clustering.
The present invention for achieving the above object, adopts following technical scheme:
A load curve parallel clustering method based on the large data of electric power, comprises the steps:
1) collect the load curve row filter of going forward side by side;
2) load curve is normalized;
3) load curve is carried out to denoising;
4) load is carried out to cluster analysis;
5) to described step 4) the K type load cluster result that obtains of K mean cluster carries out signature analysis, win with the obvious cluster result of electrical feature, calculate the cluster centre of all kinds of curves of gained, the cluster centre of all kinds of curves is carried out to secondary classification, obtain several typical load characteristics clustering results.
Aforesaid step 1) in, Data Collection refers to screening the 96 point load curves that extract corporate client from province electricity consumption acquisition system storehouse, delete the curve that wherein load data is incomplete and load capacity is 0, obtain the normal big customer's load curve of data integrity and situation.
Aforesaid step 2) in, load curve normalization refers to described step 1) every load curve travel through, find out the maximum point of loading in 96, as reference capacity, carry out curve normalization, normalization formula is as follows:
P′
i,j＝P
i,j/P
i,max
Wherein: subscript i represents load curve numbering, j represents that the load in every load curve gathers sequence number, P
i,jrepresent the load value of i user j collection point, P
i, maxrepresent the load maximal value in 96 of i user's daily loads.
Aforesaid step 3) in, load denoising comprises the following steps:
3-1) select wavelet function and determine the number of plies of decomposing, then carrying out wavelet decomposition;
3-2) select a threshold value to carry out soft-threshold quantification treatment to the high frequency coefficient under each decomposition scale;
3-3) carry out the wavelet reconstruction of load curve according to each layer of frequency coefficient of wavelet decomposition, obtain the load curve after denoising.
Aforesaid step 3-1) in, ' dN4 ' in the selected little wave system of Daubechies of wavelet function, decomposing the number of plies is 3 layers.
Aforesaid step 4) in, adopt the K means clustering algorithm based on multi-core parallel concurrent technology to carry out the cluster analysis of load curve, comprise the following steps:
The core that 4-1) detects current computer CPU is counted n, activates all cores, prepares to calculate;
4-2) to described step 3) denoising after load curve carry out distinctiveness ratio analysis, find out the K bar load curve of distinctiveness ratio maximum, as the initial cluster center of parallel K-means clustering algorithm;
4-3) all load curves are divided into n part, give n core of current computer, carry out similarity calculating with K cluster centre respectively, and load curve is included in that class of the load curve as cluster centre that similarity is the highest;
4-4) after all load curve all classifications complete, calculate and upgrade current cluster centre of all categories, check whether the cluster centre difference that current all cluster centres and last iteration obtain is all less than pre-set threshold value, if, enter step 4-5), otherwise, proceed to step 4-3);
4-5) cluster finishes, and closes parallel computation process, and releasing memory shows cluster result.
Aforesaid step 5) in, typical load characteristics clustering result comprises: occur in short-term high load capacity; Whole day load balancing; Daytime, load was higher; Daytime, load was higher, but slightly fell noon; Night load is higher.
The present invention is without setting up distributed computing system, only utilize the multi-core CPU of existing desktop computer or notebook computer, can realize the parallel computation of load curve cluster, and this algorithm can obtain the core cpu number of current computer, and automatically open all cores and participate in clusters, take full advantage of the calculated performance of current computer, greatly improved the speed of magnanimity load curve cluster.The present invention, compared to traditional K means clustering method, takes full advantage of the multi-core CPU of computing machine, has effectively improved the load characteristics clustering efficiency under large data background, is therefore worthy of promotion and application.
Brief description of the drawings
Fig. 1 is the main process figure of the inventive method;
Fig. 2 is the process flow diagram of the K means clustering method based on multi-core parallel concurrent technology;
Fig. 3 is that Jiangsu electricity consumption big customer daily load curve small echo is soft, relatively schematic diagram of hard-threshold denoising effect;
Fig. 4 is the curve cluster result that Jiangsu part electricity consumption big customer daily load curve K mean cluster obtains;
Fig. 5 is the curve number distribution situation obtaining after Fig. 4 cluster.
Embodiment
Describe the present invention in detail below in conjunction with the drawings and specific embodiments.
As shown in Figure 1, the load curve parallel clustering method based on the large data of electric power of the present invention comprises the following steps:
One, collect the load curve row filter of going forward side by side
Concrete grammar is: the 96 point load curves that extract corporate client from province electricity consumption acquisition system storehouse, corporate client refers to the enterprise customer that load control terminal is installed, delete the curve that wherein load data is incomplete and load capacity is 0, obtain the normal big customer's load curve of data integrity and situation.96 point load curves refer to that user gathers 1 point for general 15 minutes, one day 96 point, 1 client's of formation 96 point load curves.
Two, load curve is normalized
In electric system, each user's power load, load capacity vary, the high average daily power consumption of power consumption is up to hundreds thousand of kilowatt hours, load capacity is up to ten tens thousand of kilowatts, and the low average daily power consumption of power consumption is low to moderate tens kilowatt hours, only several kilowatts of load capacities.Therefore need all load curves to be normalized.Normalization refers to every load curve in the first step traveled through, and finds out the maximum point of loading in 96, as reference capacity, carries out curve normalization, and normalization formula is as follows:
P′
i,j＝P
i,j/P
i,max
Wherein: subscript i represents load curve numbering, j represents that the load in every load curve gathers sequence number, general 15 minutes 1 points, one day 96 point, P
i,jrepresent the load value of i user j collection point, P
i, maxrepresent the load maximal value in 96 of i user's daily loads.
Three, load curve is carried out to denoising
Because current most curve clustering algorithms all carry out cluster by the numerical value similarity of curve, and under large data clusters background, the minor swing existing in curve can not representative of consumer electricity consumption trend, and can cause that cluster result is undesirable, the noncommittal problem of Clustering Tendency.Therefore first the present invention utilizes the little wave system of Daubechies to carry out wavelet decomposition to every load curve, by self-defined soft-threshold Wavelet Denoising Method, load curve is carried out to smoothing processing, then carries out wavelet reconstruction and obtains the load curve after denoising.The concrete steps of carrying out denoising are as follows:
Step1 selects wavelet function and determines the number of plies of decomposing, and then carries out wavelet decomposition, ' dN4 ' in the selected little wave system of Daubechies of wavelet function of the present invention, and decomposing the number of plies is 3 layers;
Step2 selects a threshold value to carry out soft-threshold quantification treatment to the high frequency coefficient under each decomposition scale;
Step3 carries out the wavelet reconstruction of load curve according to each layer of frequency coefficient of wavelet decomposition, obtain the load curve after denoising.
' dN4 ' in the selected little wave system of Daubechies of the present invention, as wavelet function, decomposing the number of plies is 3 layers.The tendency of the pressure of Matlab and default threshold denoising meeting change of load curve, therefore this method adopts self-defined soft-threshold denoising.
Four, load is carried out to cluster analysis
With under TV university data background; the quantity of customer charge curve is very huge; when traditional K means clustering method carries out cluster, speed is very slow, and the present invention adopts the K means clustering algorithm based on multi-core parallel concurrent technology to carry out the cluster analysis of load curve, to improve the speed of cluster analysis.Referring to Fig. 2, concrete steps are as follows:
The core that 4-1) detects current computer CPU is counted n, activates all cores, prepares to calculate;
4-2) load curve after the 3rd step denoising is carried out to distinctiveness ratio analysis, find out the K bar load curve of distinctiveness ratio maximum, as the initial cluster center of parallel K-means clustering algorithm;
4-3) all load curves are divided into n part, give n core of current computer, carry out similarity calculating with K cluster centre respectively, and load curve is included in that class of the load curve as cluster centre that similarity is the highest;
4-4) after all load curve all classifications complete, all group of curves under of all categories are averaged, and upgrade current cluster centre of all categories with this mean value curve, check whether the cluster centre difference that current all cluster centres and last iteration obtain is all less than pre-set threshold value, if, enter step 4-5), otherwise, proceed to step 4-3);
4-5) cluster finishes, and closes parallel computation process, and releasing memory shows cluster result.
Five, the K type load cluster result the 4th step K mean cluster being obtained carries out signature analysis, win with the obvious cluster result of electrical feature, as there is higher, the higher feature of load in evening of whole day load balancing, daytime load, calculate the cluster centre of all kinds of curves of gained, the feature of all kinds of curves of identification, the cluster centre of all kinds of curves is carried out to secondary classification, obtain several typical load characteristics clustering results.Typical load characteristics clustering result comprises and occurs in short-term high load capacity; Whole day load balancing; Daytime, load was higher; Daytime, load was higher, but slightly fell noon; Night load is higher.
Describe the present invention in detail below by a specific embodiment.
From Jiangsu Province's electricity consumption acquisition system, extract 45,000 corporate clients, the 96 point load curves of a certain day, delete the curve that wherein load data is incomplete and load capacity is 0, obtain normal 41487 load curves of data integrity and situation.For reducing the calculated amount of cluster, 96 point curves are equivalent to 24 hours daily load curves.
According to the present invention, be normalized with denoising after soft, the hard-threshold denoising effect of certain electricity consumption big customer daily load curve small echo more as shown in Figure 3.
After visible hard-threshold wavelet de-noising, curve becomes unimodally from bimodal, has lost the characteristic of load reduction at noon; And after self-defined threshold value wavelet de-noising, it is more level and smooth that curve becomes, and fundamental characteristics is all retained.
In order to improve the speed of the large data clusters of electricity consumption curve, the improvement K means clustering method of employing based on concurrent technique is by 40, it is 20 classes that more than 000 load curve gathers, 41487 load curves are carried out to distinctiveness ratio analysis, find out 20 load curves of distinctiveness ratio maximum, as the initial cluster center of parallel K-means clustering algorithm.Be Intel Corei7-2600K at processor, core number is 4, on the computer of internal memory 4GB, carry out cluster, be about to 41487 load curves and be divided into 4 parts, give 4 cores and carry out similarity calculating with 20 cluster centres respectively, and curve is included into the classification at the load curve place as cluster centre that similarity is the highest.
The computing time of multi-core parallel concurrent algorithm and traditional algorithm is more as shown in table 1.
The parallel K mean cluster of table 1 computing time
| Parallel Kernel calculation | 1 | 2 | 4 |
| Computing time/s | 20.61 | 11.90 | 7.25 |
As can be seen from Table 1, when 4 core parallel computation, the time shortens to originally 35.18%, and this is very favorable to the cluster analysis under large data background.
After all load curve all classifications complete, calculate and upgrade current cluster centre of all categories, checking whether the cluster centre difference that current all cluster centres and last iteration obtain is all less than pre-set threshold value, if so, cluster finishes, Output rusults, otherwise, re-start cluster calculation.
After cluster finishes, obtain 20 class curves from cluster, choose with electrical feature obviously the 16 class curves on (for example whole day load balancing, daytime load higher, the higher feature of load in evening) (amount to 35204, account for sum 84.86%), as shown in Figure 4, in figure, white curve represents the cluster centre of such curve.The number of 16 class curves distributes as shown in Figure 5.
Above-mentioned 16 class curves can incorporate into as five large classes as shown in table 2 with electrical feature according to it:
The large type load curve of table 2 five
| Large class | Contained classification | Curve number | Use electrical feature |
| I | 1,2,3,4 | 7896 | There is in short-term high load capacity |
| II | 5,6,7 | 9842 | Whole day load balancing |
| III | 8,9,10,11 | 9892 | Daytime, load was higher |
| IV | 12,13 | 4911 | Load on daytime higher (lunch break is slightly fallen) |
| V | 14,15,16 | 2663 | Night load is higher |
Classification results shows, parallel clustering algorithm effect of the present invention is remarkable, practical, is a kind of effective way that solves load curve cluster under large data background.
Below announce the present invention as above with preferred embodiment, so it is not intended to limiting the invention, and all technical schemes of taking the mode that is equal to replacement or equivalent transformation to obtain, all drop in protection scope of the present invention.
Claims (7)
1. the load curve parallel clustering method based on the large data of electric power, is characterized in that, comprises the steps:
1) collect the load curve row filter of going forward side by side;
2) load curve is normalized;
3) load curve is carried out to denoising;
4) load is carried out to cluster analysis;
5) to described step 4) the K type load cluster result that obtains of K mean cluster carries out signature analysis, win with the obvious cluster result of electrical feature, calculate the cluster centre of all kinds of curves of gained, the cluster centre of all kinds of curves is carried out to secondary classification, obtain several typical load characteristics clustering results.
2. a kind of load curve parallel clustering method based on the large data of electric power according to claim 1, it is characterized in that, described step 1) in, Data Collection refers to screening the 96 point load curves that extract corporate client from province electricity consumption acquisition system storehouse, delete the curve that wherein load data is incomplete and load capacity is 0, obtain the normal big customer's load curve of data integrity and situation.
3. a kind of load curve parallel clustering method based on the large data of electric power according to claim 1, it is characterized in that, described step 2) in, load curve normalization refers to described step 1) every load curve travel through, find out the maximum point of loading in 96, as reference capacity, carry out curve normalization, normalization formula is as follows:
P′
i,j＝P
i,j/P
i,max
Wherein: subscript i represents load curve numbering, j represents that the load in every load curve gathers sequence number, P
i,jrepresent the load value of i user j collection point, P
i, maxrepresent the load maximal value in 96 of i user's daily loads.
4. a kind of load curve parallel clustering method based on the large data of electric power according to claim 1, is characterized in that described step 3) in, load denoising comprises the following steps:
3-1) select wavelet function and determine the number of plies of decomposing, then carrying out wavelet decomposition;
3-2) select a threshold value to carry out soft-threshold quantification treatment to the high frequency coefficient under each decomposition scale;
3-3) carry out the wavelet reconstruction of load curve according to each layer of frequency coefficient of wavelet decomposition, obtain the load curve after denoising.
5. a kind of load curve parallel clustering method based on the large data of electric power according to claim 4, is characterized in that described step 3-1) in, ' dN4 ' in the selected little wave system of Daubechies of wavelet function, decomposing the number of plies is 3 layers.
6. a kind of load curve parallel clustering method based on the large data of electric power according to claim 1, it is characterized in that, described step 4) in, adopt the K means clustering algorithm based on multi-core parallel concurrent technology to carry out the cluster analysis of load curve, comprise the following steps:
The core that 4-1) detects current computer CPU is counted n, activates all cores, prepares to calculate;
4-2) to described step 3) denoising after load curve carry out distinctiveness ratio analysis, find out the K bar load curve of distinctiveness ratio maximum, as the initial cluster center of parallel K-means clustering algorithm;
4-3) all load curves are divided into n part, give n core of current computer, carry out similarity calculating with K cluster centre respectively, and load curve is included in that class of the load curve as cluster centre that similarity is the highest;
4-4) after all load curve all classifications complete, calculate and upgrade current cluster centre of all categories, check whether the cluster centre difference that current all cluster centres and last iteration obtain is all less than pre-set threshold value, if, enter step 4-5), otherwise, proceed to step 4-3);
4-5) cluster finishes, and closes parallel computation process, and releasing memory shows cluster result.
7. a kind of load curve parallel clustering method based on the large data of electric power according to claim 1, is characterized in that described step 5) in, typical load characteristics clustering result comprises: occur in short-term high load capacity; Whole day load balancing; Daytime, load was higher; Daytime, load was higher, but slightly fell noon; Night load is higher.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410312530.XA CN104063480B (en) | 2014-07-02 | 2014-07-02 | A kind of load curve parallel clustering method of electrically-based big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410312530.XA CN104063480B (en) | 2014-07-02 | 2014-07-02 | A kind of load curve parallel clustering method of electrically-based big data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104063480A true CN104063480A (en) | 2014-09-24 |
| CN104063480B CN104063480B (en) | 2016-09-21 |
Family
ID=51551194
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410312530.XA Active CN104063480B (en) | 2014-07-02 | 2014-07-02 | A kind of load curve parallel clustering method of electrically-based big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104063480B (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104376402A (en) * | 2014-10-31 | 2015-02-25 | 国家电网公司 | Load characteristic classification and synthesis method based on frequency domain indexes |
| CN104680261A (en) * | 2015-03-16 | 2015-06-03 | 朗新科技股份有限公司 | Power load operation control method based on load curve clustering of major clients |
| CN106022509A (en) * | 2016-05-07 | 2016-10-12 | 国网浙江省电力公司经济技术研究院 | Power distribution network space load prediction method taking region and load property dual differences into consideration |
| CN106354803A (en) * | 2016-08-26 | 2017-01-25 | 清华大学 | Bad load data detection algorithm for power transmission and transformation equipment based on index of characteristic |
| CN106408008A (en) * | 2016-09-08 | 2017-02-15 | 国网江西省电力公司赣州供电分公司 | Load curve distance and shape-based load classification method |
| CN106446967A (en) * | 2016-10-28 | 2017-02-22 | 国网福建省电力有限公司 | Novel power system load curve clustering method |
| CN107145986A (en) * | 2017-05-24 | 2017-09-08 | 北京中电普华信息技术有限公司 | A kind of charge capacity Forecasting Methodology and device |
| CN107194600A (en) * | 2017-06-01 | 2017-09-22 | 国网山东省电力公司济南市历城区供电公司 | A kind of electric load Seasonal Characteristics sorting technique |
| CN107403247A (en) * | 2016-05-18 | 2017-11-28 | 中国电力科学研究院 | Based on the adaptive load classification polymerization analysis method for finding cluster core algorithm |
| CN108205721A (en) * | 2016-12-16 | 2018-06-26 | 国家电网公司 | Spline interpolation typical day load curve selecting device based on cluster |
| CN108898154A (en) * | 2018-09-29 | 2018-11-27 | 华北电力大学 | A kind of electric load SOM-FCM Hierarchical clustering methods |
| CN109636101A (en) * | 2018-11-02 | 2019-04-16 | 国网辽宁省电力有限公司朝阳供电公司 | Large user's electricity consumption behavior analysis method under opening sale of electricity environment based on big data |
| CN109829916A (en) * | 2019-03-04 | 2019-05-31 | 国网内蒙古东部电力有限公司 | A kind of Diagnosis Method of Transformer Faults based on CNN |
| CN110069467A (en) * | 2019-04-16 | 2019-07-30 | 沈阳工业大学 | System peak load based on Pearson's coefficient and MapReduce parallel computation clusters extraction method |
| CN110659699A (en) * | 2019-09-30 | 2020-01-07 | 重庆大学 | Distributed double-layer clustering analysis method based on feature index dimension reduction |
| CN111600304A (en) * | 2020-06-17 | 2020-08-28 | 广东工业大学 | Building power scheduling method, device and equipment |
| CN111950620A (en) * | 2020-08-07 | 2020-11-17 | 国网能源研究院有限公司 | User screening method based on DBSCAN and K-means algorithm |
| CN112330051A (en) * | 2020-11-20 | 2021-02-05 | 国网辽宁省电力有限公司营口供电公司 | Short-term load prediction method based on Kmeans and FR-DBN |
| CN113052938A (en) * | 2019-12-26 | 2021-06-29 | 新奥数能科技有限公司 | Method and device for constructing boiler energy efficiency curve |
| CN113298300A (en) * | 2021-05-14 | 2021-08-24 | 上海电机学院 | Load sequence clustering method based on HP filter and re-standard range method |
| CN113761700A (en) * | 2020-06-05 | 2021-12-07 | 国家电网有限公司华东分部 | Load modeling and online correction method and system based on dynamic clustering |
| US11977925B2 (en) * | 2020-08-04 | 2024-05-07 | Smart Software, Inc. | Clustering and visualizing demand profiles of resources |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101706778A (en) * | 2008-08-27 | 2010-05-12 | 王阳 | Method for predicting power load |
| CN102999791A (en) * | 2012-11-23 | 2013-03-27 | 广东电网公司电力科学研究院 | Power load forecasting method based on customer segmentation in power industry |
| CN103049651A (en) * | 2012-12-13 | 2013-04-17 | 航天科工深圳（集团）有限公司 | Method and device used for power load aggregation |
| CN103577883A (en) * | 2013-11-18 | 2014-02-12 | 国家电网公司 | Grid-load intelligent interaction method and device |
-
2014
- 2014-07-02 CN CN201410312530.XA patent/CN104063480B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101706778A (en) * | 2008-08-27 | 2010-05-12 | 王阳 | Method for predicting power load |
| CN102999791A (en) * | 2012-11-23 | 2013-03-27 | 广东电网公司电力科学研究院 | Power load forecasting method based on customer segmentation in power industry |
| CN103049651A (en) * | 2012-12-13 | 2013-04-17 | 航天科工深圳（集团）有限公司 | Method and device used for power load aggregation |
| CN103577883A (en) * | 2013-11-18 | 2014-02-12 | 国家电网公司 | Grid-load intelligent interaction method and device |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104376402B (en) * | 2014-10-31 | 2017-06-13 | 国家电网公司 | A kind of load classification and integrated approach based on frequency-domain index |
| CN104376402A (en) * | 2014-10-31 | 2015-02-25 | 国家电网公司 | Load characteristic classification and synthesis method based on frequency domain indexes |
| CN104680261A (en) * | 2015-03-16 | 2015-06-03 | 朗新科技股份有限公司 | Power load operation control method based on load curve clustering of major clients |
| CN106022509A (en) * | 2016-05-07 | 2016-10-12 | 国网浙江省电力公司经济技术研究院 | Power distribution network space load prediction method taking region and load property dual differences into consideration |
| CN106022509B (en) * | 2016-05-07 | 2019-11-26 | 国网浙江省电力公司经济技术研究院 | Consider the Spatial Load Forecasting For Distribution method of region and load character double differences |
| CN107403247A (en) * | 2016-05-18 | 2017-11-28 | 中国电力科学研究院 | Based on the adaptive load classification polymerization analysis method for finding cluster core algorithm |
| CN106354803A (en) * | 2016-08-26 | 2017-01-25 | 清华大学 | Bad load data detection algorithm for power transmission and transformation equipment based on index of characteristic |
| CN106354803B (en) * | 2016-08-26 | 2020-04-10 | 清华大学 | Method for detecting bad data of electric power transmission and transformation equipment load based on characteristic indexes |
| CN106408008A (en) * | 2016-09-08 | 2017-02-15 | 国网江西省电力公司赣州供电分公司 | Load curve distance and shape-based load classification method |
| CN106446967A (en) * | 2016-10-28 | 2017-02-22 | 国网福建省电力有限公司 | Novel power system load curve clustering method |
| CN108205721A (en) * | 2016-12-16 | 2018-06-26 | 国家电网公司 | Spline interpolation typical day load curve selecting device based on cluster |
| CN107145986A (en) * | 2017-05-24 | 2017-09-08 | 北京中电普华信息技术有限公司 | A kind of charge capacity Forecasting Methodology and device |
| CN107194600A (en) * | 2017-06-01 | 2017-09-22 | 国网山东省电力公司济南市历城区供电公司 | A kind of electric load Seasonal Characteristics sorting technique |
| CN108898154A (en) * | 2018-09-29 | 2018-11-27 | 华北电力大学 | A kind of electric load SOM-FCM Hierarchical clustering methods |
| CN109636101A (en) * | 2018-11-02 | 2019-04-16 | 国网辽宁省电力有限公司朝阳供电公司 | Large user's electricity consumption behavior analysis method under opening sale of electricity environment based on big data |
| CN109829916A (en) * | 2019-03-04 | 2019-05-31 | 国网内蒙古东部电力有限公司 | A kind of Diagnosis Method of Transformer Faults based on CNN |
| CN110069467A (en) * | 2019-04-16 | 2019-07-30 | 沈阳工业大学 | System peak load based on Pearson's coefficient and MapReduce parallel computation clusters extraction method |
| CN110659699A (en) * | 2019-09-30 | 2020-01-07 | 重庆大学 | Distributed double-layer clustering analysis method based on feature index dimension reduction |
| CN113052938A (en) * | 2019-12-26 | 2021-06-29 | 新奥数能科技有限公司 | Method and device for constructing boiler energy efficiency curve |
| CN113761700A (en) * | 2020-06-05 | 2021-12-07 | 国家电网有限公司华东分部 | Load modeling and online correction method and system based on dynamic clustering |
| CN111600304A (en) * | 2020-06-17 | 2020-08-28 | 广东工业大学 | Building power scheduling method, device and equipment |
| US11977925B2 (en) * | 2020-08-04 | 2024-05-07 | Smart Software, Inc. | Clustering and visualizing demand profiles of resources |
| CN111950620A (en) * | 2020-08-07 | 2020-11-17 | 国网能源研究院有限公司 | User screening method based on DBSCAN and K-means algorithm |
| CN112330051A (en) * | 2020-11-20 | 2021-02-05 | 国网辽宁省电力有限公司营口供电公司 | Short-term load prediction method based on Kmeans and FR-DBN |
| CN113298300A (en) * | 2021-05-14 | 2021-08-24 | 上海电机学院 | Load sequence clustering method based on HP filter and re-standard range method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104063480B (en) | 2016-09-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104063480B (en) | A kind of load curve parallel clustering method of electrically-based big data | |
| CN108805213B (en) | Power load curve double-layer spectral clustering method considering wavelet entropy dimensionality reduction | |
| CN109871860A (en) | A kind of daily load curve dimensionality reduction clustering method based on core principle component analysis | |
| CN110263873A (en) | A kind of power distribution network platform area classification method merging sparse noise reduction autoencoder network dimensionality reduction and cluster | |
| CN110826618A (en) | Personal credit risk assessment method based on random forest | |
| CN107578149B (en) | Power grid enterprise key data analysis method | |
| CN103440539B (en) | A kind of user power utilization data processing method | |
| CN106778826A (en) | Based on the hereditary Hybrid Clustering Algorithm with preferred Fuzzy C average of self adaptation cellular | |
| CN109117872A (en) | A kind of user power utilization behavior analysis method based on automatic Optimal Clustering | |
| CN109002926A (en) | The photovoltaic power generation quantity prediction model and its construction method of a kind of high accuracy and application | |
| CN112819299A (en) | Differential K-means load clustering method based on center optimization | |
| CN109034370A (en) | A kind of convolutional neural networks compressing method based on Feature Mapping beta pruning | |
| CN109325631A (en) | Electric car charging load forecasting method and system based on data mining | |
| CN109544029A (en) | Analysis method, analytical equipment and the terminal of a kind of area's line loss | |
| CN114444539A (en) | Power load identification method, apparatus, device, medium, and program product | |
| Liu et al. | An unsupervised feature selection algorithm: Laplacian score combined with distance-based entropy measure | |
| CN116307059A (en) | Power distribution network region fault prediction model construction method and device and electronic equipment | |
| CN111612319A (en) | Load curve depth embedding clustering method based on one-dimensional convolution self-encoder | |
| CN110059938B (en) | Power distribution network planning method based on association rule driving | |
| CN117743870A (en) | Water conservancy data management system based on big data | |
| CN117035837B (en) | Method for predicting electricity purchasing demand of power consumer and customizing retail contract | |
| CN111309718B (en) | Distribution network voltage data missing filling method and device | |
| CN106340874A (en) | Identification decision-making method and system for power load decomposition | |
| CN115511297A (en) | Virtual power plant application benefit evaluation method, equipment and storage medium | |
| CN109670526A (en) | A kind of interference source type discrimination method, device, equipment and the storage medium of power grid |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |