CN106022538A - Photovoltaic power generating predicting method based on K-mean clustering improved generalized weather - Google Patents
Photovoltaic power generating predicting method based on K-mean clustering improved generalized weather Download PDFInfo
- Publication number
- CN106022538A CN106022538A CN201610380692.6A CN201610380692A CN106022538A CN 106022538 A CN106022538 A CN 106022538A CN 201610380692 A CN201610380692 A CN 201610380692A CN 106022538 A CN106022538 A CN 106022538A
- Authority
- CN
- China
- Prior art keywords
- day
- weather
- history
- data
- force data
- 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
- 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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- 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 invention discloses a photovoltaic power generating predicting method based on K-mean clustering improved generalized weather. The photovoltaic power generating predicting method comprises the steps that photovoltaic power station day-by-day historical output data is clustered to be K clusters, and in-cluster data is additionally provided with digital labels; the digital labels and meteorological specialized weathers corresponding to the day-by-day historical output data are counted, and an improved generalized weather mapping is formed, and according to the improved generalized weather mapping, one meteorological specialized weather is corresponding to one digital label or a plurality of digital labels; for the historical output data of two adjacent days, the historical output data of the first day, the digital label corresponding to the historical output data of the first day, and the digital label corresponding to the historical output data of the second day are used as input, and the historical output data of the second day is used as output, and then a photovoltaic power generating BP neural network predicting model is established, and by using the above mentioned model, electricity production is predicted. The photovoltaic power generating predicting method is capable of overcoming a defect of a conventional generalized weather mapping of absolute division of meteorological specialized weathers, and is capable of predicting the photovoltaic electricity production in sunny days, and in addition, the prediction is also accurate in changeable weathers such as cloudy days and rainy days.
Description
Technical field
The present invention relates to a kind of method that the generated energy of photovoltaic plant can be carried out Accurate Prediction, belong to photovoltaic
Technical field of power generation.
Background technology
Photo-voltaic power supply is a kind of intermittent energy source, has by Seasonal Characteristics, day characteristic, weather features and fluctuation
The feature of the factors impacts such as characteristic.Under the influence of numerous factors, the foundation of photovoltaic forecast model
Journey slightly complicated, but in most cases, still use indirect predictions method, i.e. go out force data according to photovoltaic history and build
Vertical mathematical model.This kind of method thinks that data inherently contain and have the information such as region, weather, are carried out data
After statistics and classification process, the algorithm by some with abilities such as self studies sets up photovoltaic generation model.
The foundation of photovoltaic generation model, in terms of input, needs the content of two aspects: one is photovoltaic plant
Go out force data day by day, two is the meteorological specialty weather data corresponding to data every day.Day by day force data one is gone out
As send power as sample point, if day in summer is that morning 6 is to evening six with the illumination every day period integral point moment
Point, totally ten three moment, i.e. each group of history of day in summer go out force data and all include 13 integral point moment
Generated output.Meteorological specialty weather is issued by local meteorological department, shows that photovoltaic plant runs the sky of day vaporous
Condition concept.According to GB/T 22,164 2008 national standard, all meteorological weather marks are divided by China Meteorological Administration
It is 33 kinds of professional weather patterns.Under different meteorological specialty weather, the curve that photovoltaic is exerted oneself is had nothing in common with each other.In order to
Meteorological specialty weather can be changed into the parameter that computer is capable of identify that, the concept of broad sense weather is arisen at the historic moment.
I.e. by the evaluation index (such as sunlight irradiation degree, relative humidity, cloud amount etc.) that certain predefined is good, will
33 kinds of meteorological specialty weather are divided into several set, each set all to comprise multiple meteorological specialty weather the most artificially,
And it is described set one numeral of distribution or letter labels accordingly.Simultaneously as run at photovoltaic plant
During, every day can obtain the meteorological specialty weather conditions on the same day from local meteorological department, in conjunction with
Ready-portioned broad sense collective, just makes day by day history go out force data and the most all brings the label of broad sense weather.
The broad sense weather label that two adjacent groups history day by day goes out force data and correspondence thereof inputs to algorithm, warp
Cross a large amount of training, can obtain exerting oneself data, broad sense weather label, second day on the firstth with photovoltaic on the firstth
Broad sense weather label is independent variable, goes out the matching between the force data data as dependent variable with second day photovoltaic and closes
System.Finally, then by the forecast model trained, exert oneself according to the photovoltaic in prediction day in proxima luce (prox. luc) each moment
Data, the broad sense state of weather label of prediction day proxima luce (prox. luc), the prediction day issued in conjunction with meteorological department is meteorological specially
Industry weather, it is possible to obtain predicting that the photovoltaic in day in each moment goes out force data.
At present, when utilizing the generated energy of broad sense weather forecasting photovoltaic plant, there is the division of broad sense weather excessively
The problem of absolutization, i.e. when building broad sense weather mapping ensemblen, a certain evaluation for dividing broad sense weather
Index once it is determined that, meteorological specialty weather will be ranged qualitatively in the broad sense weather pattern of a certain set,
So that described meteorological specialty weather is with the broad sense weather label corresponding to described set.Owing to photovoltaic is exerted oneself
Easily affected by factors such as wind-force on the same day, cloud amount, even if (the finest under identical meteorological specialty weather
Turn cloudy weather), also tend to the situation that respective generated output curve difference is bigger occurs, if given the most again
History is exerted oneself broad sense state of weather label on data band, describes with a certain broad sense state of weather individually,
Forecast error will be made to increase, particularly under the unsettled such as clear to cloudy, overcast and rainy, photovoltaic power generation quantity
Precision of prediction is lower.
Summary of the invention
Present invention aims to the drawback of prior art, it is provided that a kind of wide based on the improvement of K mean cluster
The photovoltaic generation Forecasting Methodology of justice weather, to improve the precision of prediction of photovoltaic power generation quantity.
Problem of the present invention solves with following technical proposals:
A kind of photovoltaic generation Forecasting Methodology improving broad sense weather based on K mean cluster, described method is by photovoltaic electric
History day by day of standing data clusters of exerting oneself is K bunch, and encloses digital label for data；Add up history day by day to exert oneself
Digital label corresponding to data and meteorology specialty weather, constitute corresponding one or many of a meteorological specialty weather
The improvement broad sense weather of individual digital label maps；History for the most adjacent two days went out force data, with first day
History goes out force data and the digital label of its correspondence, second day history go out the digital label conduct that force data is corresponding
Input, goes out force data as output using second day history, sets up the BP neural network prediction model of photovoltaic generation
And utilize described model prediction generated energy.Predict the outcome and be probably one or more groups.If one group of prediction knot
Really, then described in predict the outcome as finally predicting the outcome；If many groups predict the outcome, calculate often group prediction knot
Fruit and the Euclidean distance of its cluster centre corresponding to digital numbering, Euclidean distance minima person is final prediction
Result.
The above-mentioned photovoltaic generation Forecasting Methodology improving broad sense weather based on K mean cluster, concrete prediction is by following step
Suddenly carry out:
1. cluster day by day history and go out force data
Determine classification status number K the most as required；
B. arbitrarily choose K object history goes out force data day by day as cluster centre from photovoltaic plant, press
Other data are respectively allocated to the most close cluster centre by the shortest principle of distance, form K bunch；
C. using the average of K bunch as K new cluster centre, calculate all objects with in K new cluster
The distance of the heart, is updated cluster centre with the object minimum with each new cluster centre distance, and by away from
Cluster centre after other data are reassigned to renewal by the shortest principle, the K after being updated bunch；
D. step c is repeated, until K cluster centre does not changes；
2. go out force data for history day by day and enclose corresponding digital label
Go out force data with the history day by day in cluster and there are identical digital label, different bunch interior history day by day
Go out force data and there is different digital labels；
3. set up and improve the mapping of broad sense weather
Add up history day by day and go out the digital label corresponding to force data and meteorology specialty weather, constitute a meteorology
The improvement broad sense weather of the corresponding one or more digital labels of specialty weather maps；
4. train and verify BP neural network prediction model
From all history goes out force data day by day, part history day by day is selected to go out force data as training sample,
The most adjacent two days during history goes out force data day by day, go out with exert oneself data, history on the firstth of history on the firstth
Digital label that force data is corresponding, second day history go out digital label corresponding to force data as input, with
History on the two data of exerting oneself, as output, are closed by the nonlinear mapping between BP neutral net matching input and output
System, sets up the BP neural network prediction model of photovoltaic generation；Exert oneself data detection mould by remaining history day by day
The precision of prediction of type；
5. the BP neural network prediction model prediction photovoltaic power generation quantity of photovoltaic generation is utilized
Obtain the meteorological specialty weather of prediction day from meteorological department, map according to improving broad sense weather, find with
One or more digital labels that this meteorological specialty weather is corresponding, by predict day previous calendar history go out force data,
Digital label, one or more digital labels of prediction day that prediction day proxima luce (prox. luc) is corresponding are input to photovoltaic generation
BP neural network prediction model, obtain one group or corresponding with multiple digital labels respectively many groups prediction knot
Really, if one group predicts the outcome, then described in predict the outcome as finally predicting the outcome；If organizing prediction more
As a result, according to Euclidean distance formula:
Calculate often group to predict the outcome and the Euclidean distance of cluster centre corresponding to its numeral numbering, Euclidean distance
Minima person is for finally to predict the outcome.
In formula: d is Euclidean distance value, i=1,2,3 ..., 13, represent that 6:00 AM started to afternoon
13 integral point moment of 18, xiForce value, y is gone out for integral point moment every dayiFor xiThe cluster of affiliated bunch
Each integral point moment value at center, n represents total moment number of exerting oneself on the same day, and middle n exemplified here takes 13.
The above-mentioned photovoltaic generation Forecasting Methodology improving broad sense weather based on K mean cluster, described classification status number
K is 4.
The present invention, by the improvement to existing broad sense weather mapping relations, overcomes what existing broad sense weather mapped
The shortcoming that mapping result is excessively thought in absolute terms, thus improve the precision of prediction of photovoltaic power generation quantity.Described method is not
Only can accurately predict photovoltaic power generation quantity under fair weather, also have under the unsettleds such as cloudy turn to overcast, overcast and rainy
There is good precision of prediction.
Accompanying drawing explanation
Fig. 1 is the photovoltaic generation Forecasting Methodology flow chart of the present invention；
Fig. 2 is the photovoltaic power curve feelings that certain photovoltaic plant belongs under " cloudy turn to overcast " meteorological specialty weather
Condition, in order to illustrate that existing broad sense weather concept divides the most absolute shortcoming to weather；
Fig. 3 is to predict the outcome under thunder shower meteorology specialty weather for 28th；
Predict the outcome under Fig. 4 is that cloudy turn to overcast on 26th meteorological specialty weather；
Fig. 5 is to predict the outcome under light rain meteorology specialty weather for 23rd.
Detailed description of the invention
The invention will be further described below in conjunction with the accompanying drawings.
The shortcoming the most excessively thought in absolute terms for existing broad sense weather, this method is reflected at broad sense weather
Penetrating structure aspect, by K means clustering algorithm history will be gone out force data and gather and be four bunches day by day, each bunch all
A corresponding cluster centre.The upper digital label of band in first making each bunch, then bunch in history go out force data and also bring
Digital label.In statistics bunch, history goes out the meteorological specialty weather corresponding to force data, according to this corresponding relation,
Obtain improving broad sense weather to map.In improving the mapping of broad sense weather, corresponding one of same meteorological specialty weather
Or multiple broad sense weather label, thus it is adapted under same prediction day meteorology specialty weather, it is understood that there may be many
Degeneration feature.
The present invention comprises the following steps:
1. cluster day by day history and go out force data
Cluster flow process is:
A. the classification status number K required for input, it is 4 that the present invention takes K.
B. choosing any 4 objects from history goes out force data day by day as cluster centre, other data are respectively
It is assigned to the most close cluster centre, forms K bunch.
C. take the average of K bunch as 4 new cluster centres, calculate all objects and K new cluster
The distance at center.Take the object producing minimum range as the new cluster centre of the next one, and by step b
Distribution method other data are reassigned to the cluster centre after updating, individual bunch of the K after being updated.
D. step c is repeated, until till 4 cluster centres do not change.
By K means clustering algorithm, the data during history goes out force data day by day with similar power producing characteristics are gathered
In class to same bunch, it is achieved that the data classifying rationally based on data internal relation.Life in 4 bunches
4 cluster centres become, as each bunch of interior benchmark data, establish for the last many groups of screenings predicted the outcome
Fixed basis.
2. it is that 4 bunches of interior history day by day go out force data and enclose digital label
Day by day history go out force data clustered after, be scattered in 4 bunches, enclose digital label for each bunch, 4
Individual bunch of interior history day by day goes out force data just with unified digital label.
3. set up and improve the mapping of broad sense weather
Add up history day by day go out the digital label corresponding to force data and day by day history go out the gas corresponding to force data
As specialty weather is in form, constitute changing of the meteorological specialty corresponding one or more digital labels of weather
Enter broad sense weather to map.
Existing broad sense weather maps and improves broad sense weather mapping pair and compares:
Table 1 is existing broad sense weather mapping relations, and table 2 is for improving broad sense weather mapping relations.
Table 1 broad sense weather pattern correspondence table
Broad sense weather pattern correspondence table after table 2 improvement.
The two discovery of contrast, when the meteorological specialty weather predicting day determines, the broad sense weather class of the former correspondence
Type is fixing, and the same Professional Meteorological weather improving broad sense weather can corresponding multiple numerals be numbered,
Comprising more kinds of prediction in it may." cloudy turn to overcast " in below with prediction day weather for meteorological specialty weather
As a example by weather, analyze the necessity improved:
When processing " cloudy turn to overcast " meteorological specialty weather with existing broad sense weather concept, once refer to through evaluation
Mark is evaluated, and " cloudy turn to overcast " meteorological specialty weather will the most corresponding " B " state.But from shown in Fig. 2
Certain photovoltaic plant belong to the photovoltaic power curve under " cloudy turn to overcast " meteorological specialty weather it can be seen that
Under same meteorological specialty weather, three photovoltaic power curves still differ relatively big, the wherein song on July 8
Line is closer to fine day, but the curve under the curve on July 15 and rainy weather is much like.So, as
Fruit describes with " B " state merely, can make the bigger error that predicts the outcome undoubtedly.And this phenomenon can also solve
Be interpreted as what under fair weather, existing Forecasting Methodology precision the highest (because do not have more changeable factor do
Disturb), and under unsettled, it was predicted that precision will be greatly reduced.
4. train and verify BP neural network prediction model
In all data, part history day by day is selected to go out force data as training sample.The most adjacent two
Day during history goes out force data day by day, with the numeral mark that historical data on the firstth, historical data on the firstth are corresponding
Sign, digital label corresponding to second day data as input, using second day historical data as output, pass through
Nonlinear mapping relation between BP neutral net matching input and output, sets up the BP neutral net of photovoltaic generation
Forecast model.Exert oneself by remaining history day by day the precision of prediction of data detection model.
5. the BP neural network prediction model prediction photovoltaic power generation quantity of photovoltaic generation is utilized
According to the prediction day meteorology specialty weather obtained from meteorological department, map in conjunction with improving broad sense weather, look for
To one or more digital labels that this meteorological specialty weather is corresponding.Day previous calendar history data, pre-will be predicted
Digital label, one or more digital labels of prediction day that survey day proxima luce (prox. luc) is corresponding are pre-as BP neutral net
Survey mode input, obtain one or more groups and predict the outcome.If one group of situation about predicting the outcome, then it is
Predict the outcome eventually；If many groups of situations about predicting the outcome, then by Euclidean distance formula, calculate often group pre-
Survey result and the Euclidean distance of its cluster centre corresponding to numeral numbering, obtained Euclidean distance minima person,
For finally predicting the outcome.
Fig. 1 is the photovoltaic generation Forecasting Methodology flow process improving broad sense weather based on K mean cluster of the present invention
Figure, the technical scheme implemented the present invention below in conjunction with Fig. 1 carries out in detail, describes exactly.
With first three ten days certain photovoltaic plant in July, 2015 generating data instance, improve broad sense based on K mean cluster
The photovoltaic generation Forecasting Methodology flow process of weather is as follows:
Certain photovoltaic plant July of table 3 in first three each moment on the ten generating data
Step 1: exert oneself number according to K means clustering process in Fig. 1 flow chart and above-mentioned cluster history day by day
According to step, MATLAB realizes the K mean cluster to 30 groups of historical datas.Wherein, K value definition
Being 4, i.e. 30 groups data, after clustering, will be distributed in 4 bunches.Cluster result and each bunch
Cluster centre is as shown in table 4, table 5:
The 4 30 groups of data of table result after K mean cluster
The cluster centre that table 5 is each bunch
Wherein, it is believed that bunch " 1 " possesses sensu lato " fine day " characteristic, bunch " 2 " possess sensu lato
" cloudy " characteristic, bunch " 3 " possess sensu lato " shower " characteristic, bunch " 4 " possess sensu lato " big
Rain " characteristic.
Step 2: same bunch of interior data are with identical digital label, and different bunches with different numerals
Label, 30 groups of data are after cluster, all with digital label.
Step 3: according to step described in step 1 and step 2, to photovoltaic generation data annual every day in 2015
Cluster, according to digital label and the corresponding situation of Professional Meteorological weather, after classified statistic is concluded, obtain
Broad sense weather pattern correspondence table after improvement as shown in Table 2 above, it is achieved to 33 kinds of meteorological specialty weather
Reclassify.
Step 4: the most adjacent two days during history goes out force data day by day, with historical data on the firstth, first
Digital label corresponding to digital label corresponding to day historical data, second day historical data is as input, with the
Historical data was as output on 2nd, circulated this process, until till prediction day proxima luce (prox. luc).Neural by BP
Nonlinear mapping relation between network matching input and output, sets up the BP neural network prediction model of photovoltaic generation.
Step 5: the neural network model trained by step 4, to predict photovoltaic on July 28
As a example by going out force value, it was predicted that each moment photovoltaic under this day Professional Meteorological weather goes out force value.Meteorological department issues
The same day, meteorology specialty weather was " thunder shower ", and according to the broad sense weather correspondence table after improving in table 2, it is corresponding
" 1,2,3,4 " these four digital label.So, according to the neural network model trained, can obtain
Predict the outcome to four groups.In order to from four groups predict the outcome in filter out and finally predict the outcome, can be by European
Range formula, calculates four groups and predicts the outcome and the Euclidean distance of corresponding cluster centre, Euclidean distance reckling
For final predictive value.Result of calculation is as shown in table 6.
6 four groups of Euclidean distances predicted the outcome with its cluster centre of table
As shown in Table 6, when weather was defined as digital label " 1 " on 28th, its predictive value and its correspondence
Cluster centre is pressed close to the most, so the predictive value of digital label " 1 " correspondence is final predictive value.
Predict July 28 " thunder shower " weather, it was predicted that result and essence with improving both front and back broad sense weather method simultaneously
Degree is as shown in table 7.
Table 7 improves both front and back Forecasting Methodology result and accuracy comparison
From table 7 root-mean-square error it can be seen that broad sense weather prediction method is compared to non-improved method after Gai Jining,
It is greatly improved on precision of prediction.Fig. 3, Fig. 4, Fig. 5 are " thunder shower ", " cloudy turn to overcast " respectively
Predicting the outcome under " light rain " meteorological specialty weather.From figure it can also be seen that herein improve after wide
Justice weather photovoltaic power generation output forecasting method has higher precision of prediction.
Claims (3)
1. improve a photovoltaic generation Forecasting Methodology for broad sense weather based on K mean cluster, it is characterized in that, institute
Method of stating data clusters of photovoltaic plant history day by day being exerted oneself is K bunch, and for bunch in data enclose digital label；
Add up history day by day and go out the digital label corresponding to force data and meteorology specialty weather, constitute a meteorological specialty
The improvement broad sense weather of the corresponding one or more digital labels of weather maps；History for the most adjacent two days goes out
Force data, exerts oneself data with history on the firstth and the digital label of its correspondence, second day history goes out force data pair
The digital label answered as input, goes out force data as output using second day history, sets up the BP of photovoltaic generation
Neural network prediction model also utilizes described model prediction generated energy.
A kind of photovoltaic generation based on K mean cluster improvement broad sense weather the most according to claim 1 is pre-
Survey method, is characterized in that, said method comprising the steps of:
1. cluster day by day history and go out force data
Determine classification status number K the most as required；
B. arbitrarily choose K object history goes out force data day by day as cluster centre from photovoltaic plant, press
Other data are respectively allocated to the most close cluster centre by the shortest principle of distance, form K bunch；
C. using the average of K bunch as K new cluster centre, calculate all objects with in K new cluster
The distance of the heart, is updated cluster centre with the object minimum with each new cluster centre distance, and by away from
Cluster centre after other data are reassigned to renewal by the shortest principle, the K after being updated bunch；
D. step c is repeated, until K cluster centre does not changes；
2. go out force data for history day by day and enclose corresponding digital label
Go out force data with the history day by day in cluster and there are identical digital label, different bunch interior history day by day
Go out force data and there is different digital labels；
3. set up and improve the mapping of broad sense weather
Add up history day by day and go out the digital label corresponding to force data and meteorology specialty weather, constitute a meteorology
The improvement broad sense weather of the corresponding one or more digital labels of specialty weather maps；
4. train and verify BP neural network prediction model
From all history goes out force data day by day, part history day by day is selected to go out force data as training sample,
The most adjacent two days during history goes out force data day by day, go out with exert oneself data, history on the firstth of history on the firstth
Digital label that force data is corresponding, second day history go out digital label corresponding to force data as input, with
History on the two data of exerting oneself, as output, are closed by the nonlinear mapping between BP neutral net matching input and output
System, sets up the BP neural network prediction model of photovoltaic generation；Exert oneself data detection mould by remaining history day by day
The precision of prediction of type；
5. the BP neural network prediction model prediction photovoltaic power generation quantity of photovoltaic generation is utilized
Obtain the meteorological specialty weather of prediction day from meteorological department, map according to improving broad sense weather, find with
One or more digital labels that this meteorological specialty weather is corresponding, by predict day previous calendar history go out force data,
Digital label, one or more digital labels of prediction day that prediction day proxima luce (prox. luc) is corresponding are input to photovoltaic generation
BP neural network prediction model, obtain one group or corresponding with multiple digital labels respectively many groups prediction knot
Really, if one group predicts the outcome, then described in predict the outcome as finally predicting the outcome；If organizing prediction more
As a result, calculate often group and predict the outcome and the Euclidean distance of cluster centre corresponding to its numeral numbering, European away from
From minima person for finally to predict the outcome.
A kind of photovoltaic generation prediction improving broad sense weather based on K mean cluster the most according to claim 2
Method, is characterized in that, described classification status number K is 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610380692.6A CN106022538B (en) | 2016-05-31 | 2016-05-31 | Photovoltaic power generation prediction method based on K-means clustering improved generalized weather |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610380692.6A CN106022538B (en) | 2016-05-31 | 2016-05-31 | Photovoltaic power generation prediction method based on K-means clustering improved generalized weather |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106022538A true CN106022538A (en) | 2016-10-12 |
| CN106022538B CN106022538B (en) | 2020-04-07 |
Family
ID=57093180
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610380692.6A Active CN106022538B (en) | 2016-05-31 | 2016-05-31 | Photovoltaic power generation prediction method based on K-means clustering improved generalized weather |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106022538B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109325631A (en) * | 2018-10-15 | 2019-02-12 | 华中科技大学 | Electric car charging load forecasting method and system based on data mining |
| CN109858667A (en) * | 2018-12-21 | 2019-06-07 | 国网江苏省电力有限公司苏州供电分公司 | It is a kind of based on thunder and lightning weather to the short term clustering method of loading effects |
| CN110009385A (en) * | 2019-01-11 | 2019-07-12 | 国网浙江杭州市临安区供电有限公司 | A kind of photovoltaic power generation user group division methods based on multifactor mapping |
| CN110008982A (en) * | 2019-01-11 | 2019-07-12 | 国网浙江省电力有限公司 | A kind of weather monitoring point selection method based on photovoltaic power generation power output cluster |
| CN110573883A (en) * | 2017-04-13 | 2019-12-13 | 美国西门子医学诊断股份有限公司 | Method and apparatus for determining tag count during sample characterization |
| CN111680939A (en) * | 2020-08-13 | 2020-09-18 | 国网浙江省电力有限公司 | Enterprise re-work and re-production degree monitoring method based on artificial intelligence |
| CN112365081A (en) * | 2020-11-24 | 2021-02-12 | 国家电网有限公司 | Photovoltaic power station power generation capacity prediction method and device |
| CN115577857A (en) * | 2022-11-15 | 2023-01-06 | 南方电网数字电网研究院有限公司 | Energy system output data prediction method and device and computer equipment |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110066401A1 (en) * | 2009-09-11 | 2011-03-17 | Wattminder, Inc. | System for and method of monitoring and diagnosing the performance of photovoltaic or other renewable power plants |
| CN102298707A (en) * | 2011-08-24 | 2011-12-28 | 辽宁力迅风电控制系统有限公司 | Wind power prediction method based on continuous time slice clustering and support vector machine (SVM) modeling |
| CN102999786A (en) * | 2012-10-17 | 2013-03-27 | 浙江埃菲生能源科技有限公司 | Method for short-term predicting of photovoltaic generation power on the basis of similar day feature classification and extreme learning machine |
| CN103390200A (en) * | 2013-07-18 | 2013-11-13 | 国家电网公司 | Photovoltaic power station electricity generation output power forecasting method based on similar days |
| CN103390199A (en) * | 2013-07-18 | 2013-11-13 | 国家电网公司 | Photovoltaic power generation capacity/power prediction device |
| CN103390202A (en) * | 2013-07-18 | 2013-11-13 | 华北电力大学 | Output power prediction method based on similarity data selection for photovoltaic plant |
| CN104299044A (en) * | 2014-07-01 | 2015-01-21 | 沈阳工程学院 | Clustering-analysis-based wind power short-term prediction system and prediction method |
| CN104463356A (en) * | 2014-11-27 | 2015-03-25 | 国网浙江省电力公司嘉兴供电公司 | Photovoltaic power generation power prediction method based on multi-dimension information artificial neural network algorithm |
| CN104573879A (en) * | 2015-01-30 | 2015-04-29 | 河海大学 | Photovoltaic power station output predicting method based on optimal similar day set |
| CN105005825A (en) * | 2015-07-03 | 2015-10-28 | 深圳供电局有限公司 | Method and system for predicting photovoltaic power based on dynamic neural network |
-
2016
- 2016-05-31 CN CN201610380692.6A patent/CN106022538B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110066401A1 (en) * | 2009-09-11 | 2011-03-17 | Wattminder, Inc. | System for and method of monitoring and diagnosing the performance of photovoltaic or other renewable power plants |
| CN102298707A (en) * | 2011-08-24 | 2011-12-28 | 辽宁力迅风电控制系统有限公司 | Wind power prediction method based on continuous time slice clustering and support vector machine (SVM) modeling |
| CN102999786A (en) * | 2012-10-17 | 2013-03-27 | 浙江埃菲生能源科技有限公司 | Method for short-term predicting of photovoltaic generation power on the basis of similar day feature classification and extreme learning machine |
| CN103390200A (en) * | 2013-07-18 | 2013-11-13 | 国家电网公司 | Photovoltaic power station electricity generation output power forecasting method based on similar days |
| CN103390199A (en) * | 2013-07-18 | 2013-11-13 | 国家电网公司 | Photovoltaic power generation capacity/power prediction device |
| CN103390202A (en) * | 2013-07-18 | 2013-11-13 | 华北电力大学 | Output power prediction method based on similarity data selection for photovoltaic plant |
| CN104299044A (en) * | 2014-07-01 | 2015-01-21 | 沈阳工程学院 | Clustering-analysis-based wind power short-term prediction system and prediction method |
| CN104463356A (en) * | 2014-11-27 | 2015-03-25 | 国网浙江省电力公司嘉兴供电公司 | Photovoltaic power generation power prediction method based on multi-dimension information artificial neural network algorithm |
| CN104573879A (en) * | 2015-01-30 | 2015-04-29 | 河海大学 | Photovoltaic power station output predicting method based on optimal similar day set |
| CN105005825A (en) * | 2015-07-03 | 2015-10-28 | 深圳供电局有限公司 | Method and system for predicting photovoltaic power based on dynamic neural network |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110573883A (en) * | 2017-04-13 | 2019-12-13 | 美国西门子医学诊断股份有限公司 | Method and apparatus for determining tag count during sample characterization |
| CN109325631A (en) * | 2018-10-15 | 2019-02-12 | 华中科技大学 | Electric car charging load forecasting method and system based on data mining |
| CN109858667A (en) * | 2018-12-21 | 2019-06-07 | 国网江苏省电力有限公司苏州供电分公司 | It is a kind of based on thunder and lightning weather to the short term clustering method of loading effects |
| CN110009385A (en) * | 2019-01-11 | 2019-07-12 | 国网浙江杭州市临安区供电有限公司 | A kind of photovoltaic power generation user group division methods based on multifactor mapping |
| CN110008982A (en) * | 2019-01-11 | 2019-07-12 | 国网浙江省电力有限公司 | A kind of weather monitoring point selection method based on photovoltaic power generation power output cluster |
| CN110008982B (en) * | 2019-01-11 | 2023-11-03 | 国网浙江省电力有限公司 | Meteorological monitoring point selection method based on photovoltaic power generation output clustering |
| CN110009385B (en) * | 2019-01-11 | 2023-12-12 | 国网浙江杭州市临安区供电有限公司 | Multi-factor mapping-based photovoltaic power generation user group division method |
| CN111680939A (en) * | 2020-08-13 | 2020-09-18 | 国网浙江省电力有限公司 | Enterprise re-work and re-production degree monitoring method based on artificial intelligence |
| CN112365081A (en) * | 2020-11-24 | 2021-02-12 | 国家电网有限公司 | Photovoltaic power station power generation capacity prediction method and device |
| CN115577857A (en) * | 2022-11-15 | 2023-01-06 | 南方电网数字电网研究院有限公司 | Energy system output data prediction method and device and computer equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106022538B (en) | 2020-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106022538A (en) | Photovoltaic power generating predicting method based on K-mean clustering improved generalized weather | |
| Liu et al. | Random forest solar power forecast based on classification optimization | |
| CN105631483B (en) | A kind of short-term electro-load forecast method and device | |
| CN106485262A (en) | A kind of bus load Forecasting Methodology | |
| CN104573879B (en) | Photovoltaic plant based on optimal similar day collection goes out force prediction method | |
| CN106203867A (en) | Grid division methods based on power distribution network assessment indicator system and cluster analysis | |
| CN106022528A (en) | Short-term power prediction method of photovoltaic power plant based on density peak hierarchical clustering | |
| CN103390200A (en) | Photovoltaic power station electricity generation output power forecasting method based on similar days | |
| CN105069521A (en) | Photovoltaic power plant output power prediction method based on weighted FCM clustering algorithm | |
| CN107730059A (en) | The method of transformer station's electricity trend prediction analysis based on machine learning | |
| CN106485093B (en) | Based on the solar irradiance time series synthetic method for improving Markov chain | |
| CN109934395B (en) | Multi-hydropower-region short-term power load prediction method based on time-sharing and regional meteorological data | |
| CN108304978A (en) | A kind of mid-term Electric Power Load Forecast method based on data clusters theory | |
| CN114429053B (en) | Basin scale WEFE system adaptability simulation optimization method | |
| CN108074004A (en) | A kind of GIS-Geographic Information System short-term load forecasting method based on gridding method | |
| CN104408562A (en) | Photovoltaic system generating efficiency comprehensive evaluation method based on BP (back propagation) neural network | |
| CN102147839A (en) | Method for forecasting photovoltaic power generation quantity | |
| WO2024051524A1 (en) | Joint prediction method and apparatus for hydraulic, wind and photovoltaic generation power | |
| CN103793887A (en) | Short-term electrical load on-line predicting method based on self-adaptation enhancing algorithm | |
| CN109034500A (en) | A kind of mid-term electric load forecasting method of multiple timings collaboration | |
| CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
| CN106600060A (en) | Method for predicting amount of solar radiation based on similar day sunny coefficient correction | |
| CN111191966A (en) | Time-space characteristic-based power distribution network voltage unqualified time period identification method | |
| CN112149890A (en) | Comprehensive energy load prediction method and system based on user energy label | |
| CN110070228A (en) | BP neural network wind speed prediction method for neuron branch evolution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |