CN105719029A - Combined wind power prediction method based on wind speed fluctuation characteristic extraction - Google Patents
Combined wind power prediction method based on wind speed fluctuation characteristic extraction Download PDFInfo
- Publication number
- CN105719029A CN105719029A CN201610140115.XA CN201610140115A CN105719029A CN 105719029 A CN105719029 A CN 105719029A CN 201610140115 A CN201610140115 A CN 201610140115A CN 105719029 A CN105719029 A CN 105719029A
- Authority
- CN
- China
- Prior art keywords
- wind speed
- time window
- wind
- spectrum
- method based
- 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
- 238000000605 extraction Methods 0.000 title claims abstract description 24
- 238000000034 method Methods 0.000 title claims abstract description 16
- 238000001228 spectrum Methods 0.000 claims abstract description 54
- 230000006870 function Effects 0.000 claims abstract description 16
- 238000012549 training Methods 0.000 claims abstract description 14
- 238000012706 support-vector machine Methods 0.000 claims abstract description 4
- 238000012360 testing method Methods 0.000 claims abstract description 3
- 238000004458 analytical method Methods 0.000 claims description 21
- 238000013277 forecasting method Methods 0.000 claims description 15
- 241001269238 Data Species 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 230000007423 decrease Effects 0.000 claims description 4
- 239000000178 monomer Substances 0.000 claims description 4
- 101100379079 Emericella variicolor andA gene Proteins 0.000 claims description 3
- 241000949477 Toona ciliata Species 0.000 claims description 3
- 238000000205 computational method Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000013145 classification model Methods 0.000 claims description 2
- 238000010276 construction Methods 0.000 claims description 2
- 230000003247 decreasing effect Effects 0.000 claims description 2
- 238000005457 optimization Methods 0.000 abstract 1
- 238000010183 spectrum analysis Methods 0.000 abstract 1
- 230000005611 electricity Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000001154 acute effect Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 238000000053 physical method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
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
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/24765—Rule-based classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Economics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Medical Informatics (AREA)
- Operations Research (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Game Theory and Decision Science (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Control Of Eletrric Generators (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Wind Motors (AREA)
Abstract
The invention discloses a combined wind power prediction method based on wind speed fluctuation characteristic extraction. The combined wind power prediction method includes the following steps that wind speed data acquired by training samples are normalized; time windows are established for the normalized wind speeds, and multifractal spectrum analysis is performed in the time windows; the widths omega of singular index alpha value taking intervals of the time windows and symmetry parameters S of peak value differences Deltaf (alpha) and f (alpha) of a singular spectrum function f (alpha) are analyzed and compared. The wind speeds are classified according to the parameters [omega, Delta f (alpha), S], and the sizes of the time windows are further adjusted. Divided categories are sequentially trained by using an extreme learning machine, a support vector machine and an optimization regression power curve method, average monthly precision comparison is conducted on produced prediction results, one of the methods is selected to serve as an optimum single algorithm for the categories, and trained models are obtained. Same classification and modeling are conducted on test samples, corresponding optimum single algorithms are selected for different models for respective prediction, and finally final prediction results are obtained through combination.
Description
Technical field
The present invention relates to a kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction, belong to operation and control of electric power system field.
Background technology
Along with wind-powered electricity generation permeability in the whole power system of China steps up, the problems such as Control of Voltage caused by its undulatory property, intermittence and randomness, active power dispatch and system stability are more and more prominent, wind power prediction is possible not only to reduce system reserve capacity and energy storage accurately, reduce system operation cost, also contribute to the impact alleviating wind power integration to electrical network simultaneously, improve operation of power networks reliability.
The intermittent nature of nature wind energy determines wind power and has very strong undulatory property, and along with the continuous increase of wind energy turbine set quantity and installed capacity, once be connected to the grid by wind-powered electricity generation, this power swing will bring huge challenge to the safety and economic operation of electrical network.In advance wind speed and wind power are predicted accurately, it is possible to alleviate the pressure of power system peak regulation, frequency modulation, be effectively improved the electrical network receiving ability to wind-powered electricity generation.
At present, both at home and abroad wind-power electricity generation is predicted that the research of problem is more and more extensive.Generally believe, wind speed change at random the wind power fluctuation change caused is the main uncertain factor of the power system stability containing wind-power electricity generation, control problem.Therefore study the wind velocity variation law of wind energy turbine set, set up rational wind farm wind velocity variation model, further the power producing characteristics of Wind turbines is predicted particularly important.
Traditional prediction method is mainly based upon the prediction of short-term and ultra-short term, such as Forecasting Methodologies such as the physical method based on numerical weather forecast, the Kalman filtering of Corpus--based Method means, artificial neural network, chaology, extreme learning machine, support vector machine.But these methods are all that historical wind speed and history actual power are set up model, and then following wind power is predicted.This means that the input of forecast model has significant difference from the mapping relations of output under different wind friction velocities.Therefore, single under different wind friction velocities forecast model precision of prediction has much room for improvement.
Additionally, when wind speed generation acute variation, generation also will increase in frequent degree and the intensity of the fluctuations in wind speed of this time period.As time goes on, set up model just for the historical data in a certain special time period and can not meet the dynamic characteristic of the fluctuations in wind speed in All Time sequence.And, the size of time window has vital impact for extraction and the classification of data characteristics, and too big and too little window is all unfavorable for feature analysis.
Summary of the invention
For the defect existed in prior art, it is an object of the invention to provide a kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction.
For reaching object above, the present invention considers following factor:
1, wind energy turbine set NWP (numerical weather forecast) wind speed, actual power data;
2, fluctuations in wind speed characteristic;
3, the width of predicted time window;
4, the fluctuation characteristic corresponding to the different parameters of multifractal spectra.
On the basis of factors above, a kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction, comprise the following steps:
A. the wind speed of training sample collection is normalized, to eliminate the difference of the amplitude that the interference such as noise bring.
B. the wind speed after normalization is set up time window, and in this time window, carrying out multi-fractal analysis of spectrum, the major parameter of multi-fractal analysis of spectrum includes the asymmetry parameter S of singular index α, the Singularity spectrum function f (α) of multifractal spectra and Singularity spectrum function.
C. the width ω of the singular index α interval of each time window of com-parison and analysis, peak difference Δ f (α) of the Singularity spectrum function f (α) of multifractal spectra and the asymmetry parameter S of Singularity spectrum function.
D. according to the step C parameter [ω being previously mentioned, Δ f (α), S] wind speed is classified, and the parameter [ω corresponding to each time window, Δ f (α), S] adjust the size of time window further so that and each time window can completely and accurately comprise specific wind speed type.
E. the wind speed type that step D is divided by extreme learning machine, support vector machine and optimized regression power curve method is adopted to be sequentially carried out training, and produced predicting the outcome is carried out monthly accuracy comparison, select one of which method as the optimum monomer algorithm of this wind speed type, it is trained modeling, obtains the model trained.
F. carry out identical classification and modeling according to the standard of above-mentioned training sample classification model construction to testing sample, and corresponding optimum monomer algorithm in different Model Selection step E is predicted respectively, be finally combined obtaining final predicting the outcome.
On the basis of such scheme, wind speed { ν described in step At, t=1,2 ... the normalization formula of n} is as follows:
On the basis of such scheme, the step of multi-fractal analysis of spectrum described in step B includes:
B1, calculating singular index α: the singular index α local singularity representing wind speed, definition D (i) is i i square region, its central point be I (x, y), thenWherein, μ is defined in the probability measure on [0,1] [0,1], i=2n+1, n=0, and 1 ...;
B2, Singularity spectrum function f (α): f (α) calculating multifractal spectra represent the overall singularity of wind speed, and computational methods are: for each point, (x, singular index y) calculate αmax=max (α (i, j)), αmin(α (i, j)), by [α for=minmin, αmax] it is divided into N number of interval, obtain the central point singular value in each interval accordingly, replace the value of other points in interval by this value, then obtain f (α) according to formula (5);
If μ is defined in the probability measure on [0,1] [0,1], Dn is the increasing sequence of a positive integer composition, then define
WhereinIt is μ (II, j, nThe summation of the point of) ≠ 0;When the limit exists, if
Definition f1(α) convert for the Legendre of τ (q)
On the basis of such scheme, the interval of ω singular index described in step C is [αmin, αmax], ω=αmax-αmin.ω is more big, and expression wind speed profile is more uneven, and fluctuations in wind speed is more violent, and when the amplitude of fluctuations in wind speed diminishes, α~f (α) composes the trend substantially narrowed.
On the basis of such scheme, Δ f described in step C (α) represents that wind speed is in the ratio of crest, wave trough position number, Δ f (α)>0 represents that wind speed is more in crest, the top of spectrum is relatively round and smooth, Δ f (α)<0 represents that wind speed is more in trough, and the top of spectrum is relatively sharp.
On the basis of such scheme, described in step C during S=0, singular spectrum is symmetrical;During S > 0, singular spectrum peak value is to the right, and wind speed has the trend of increase;< when 0, singular spectrum peak value is to the left, and wind speed has the trend of reduction for S.
On the basis of such scheme, in step D, selecting four kinds of feature wind processes as criteria for classification, the wind speed type of division includes, A: wind speed ascent stage, B: the wind speed decreased stage, C: fluctuate the stage gently, D: spike phase.
On the basis of such scheme, time window described in step D includes the time window of the fixing optimum window of width and variable-width;
Specifically comprising the following steps that of the optimal width of the time window of selection variable-width
Step D1: the starting point at air speed data starts, makes time window widthI=1,2 ... n, whereinComprise front 10 data points: p1, p2 ... p10;
Step D2: carry out multifractal spectra analysis on TWZt, and extraction obtains characteristic parameter [ωt,Δf(α)t,St], and be equal toOn characteristic parameter [ωi,Δf(α)i,Si];
Step D3: time window continues slip and obtainsWhereinComprise 10 data points: p11, p12 ... P20, it is carried out feature extraction and obtains [ωi+1,Δf(α)i+1,Si+1], according to the step C analysis done, investigate Δ f (α)iWith Δ f (α)i+1、SiAnd Si+1Positive and negative;
Step D31: if Δ f (α)iWith Δ f (α)i+1、SiWith Si+1Positive and negative all identical each other, then showWithWind speed have the identical trend that rises or falls, enter step D32;If difference, enter step D4;
Step D32: investigate ωiAnd ωi+1If it is poor result Δ ω less than a certain threshold value η, then shows that two time period degree of fluctuation are also suitable, now update time windowAnd indicate that TWZt belongs to ascent stage or decline stage;All training datas are carried out the average different index interval Δ ω that multi-fractal analysis of spectrum draws by threshold value ηAll, make η=Δ ωAll;
Step D321: if Δ ω is more than η, then still makeAnd extract characteristic parameter, the characteristic parameter [ω after being updatedt,Δf(α)t,St], then make i=i+1, repeat step D3;
Step D4: if Δ f (α)iWith Δ f (α)i+1、SiWith Si+1Positive and negative differing from each other, then showWithWind speed trend different, now have two kinds of situations: one isWithBeing in the mild fluctuation stage, two is the critical time intervals of wind speed raising and lowering.Now consider next small time windowCalculate the population variance δ of these three small time window;
Wherein, n is equal to i+2,It is the mean wind speed of the n-th little time window,It it is the wind speed grand mean of front n little time window;
Step D41: if δ is less than setting value δ0, thenWithIt is in the mild fluctuation stage, now updates time windowSetting value δ0It it is the average variance being calculated all training datas drawing.
Step D42: if δ >=setting value δ0, thenMake TWZt+1 andAs the starting point that next time window is chosen, orderRepeat step D2.
Beneficial effect: a kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction of the present invention, not only compensate for traditional annual unified Modeling and cannot simulate the mapping relations drawback of significant diversity under different wind friction velocities inputting and exporting, and solve the problem that forecast model precision of prediction single under different wind friction velocities is not high.
Accompanying drawing explanation
Fig. 1 is the present invention schematic flow sheet based on the wind power combination forecasting method of fluctuations in wind speed feature extraction.
Fig. 2 is the schematic diagram of the different multifractal spectra corresponding to fluctuations in wind speed feature.
Fig. 3 is the sliding time window schematic diagram of variable-width.
Fig. 4 is combinational algorithm schematic flow sheet.
Detailed description of the invention
Above-mentioned is only the general introduction of technical solution of the present invention, and in order to enable those skilled in the art to better understand the technological means of the present invention, below in conjunction with accompanying drawing, the present invention is described in further detail with detailed description of the invention.
As Figure 1-4,
The wind speed of training sample collection is normalized by step A., to eliminate the difference of the amplitude that the interference such as noise bring.Wind speed { νt, t=1,2 ... the normalization formula of n} is as follows:
Wind speed after normalization is set up time window by step B., and carries out multi-fractal analysis of spectrum in this time window.
Calculating singular index α: the singular index α local singularity representing wind speed, definition D (i) is i i square region, its central point be I (x, y), then(wherein, μ is defined in the probability measure on [0,1] [0,1], i=2n+1, n=0, and 1 ...).I is relevant with the location of calculating, generally takes i≤3, and at this moment (x y) reflects the singularity of local, takes big i and reflect singularity widely α.
Singularity spectrum function f (α): f (α) that calculate multifractal spectra represents the overall singularity of wind speed, and computational methods are: for each point, (x, singular index y) calculate αmax=max (α (i, j)), αmin(α (i, j)), by [α for=minmin, αmax] it is divided into N number of interval, obtain the central point singular value in each interval accordingly, replace the value of other points in interval by this value, then obtain f (α) according to formula (5).
If μ is defined in the probability measure on [0,1] [0,1], Dn is the increasing sequence of a positive integer composition, then define
WhereinIt is μ (II, j, nThe summation of the point of) ≠ 0.When the limit exists, if
Definition f1(α) convert for the Legendre of τ (q)
The width ω of the singular index α interval of each time window of step C. com-parison and analysis, the peak difference f (α) of the Singularity spectrum function of multifractal spectra, the asymmetry parameter S of Singularity spectrum function.
Choosing three major parameters to describe the degree of strength of multi-fractal character, first is singular index interval [αmin, αmax] width ω=αmax-αmin, ω is more big, and expression wind speed profile is more uneven, and fluctuations in wind speed is more violent, and when the amplitude of fluctuations in wind speed diminishes, α~f (α) composes the trend substantially narrowed.Second is peak difference Δ f (α) of Singularity spectrum function, Δ f (α) represents that wind speed is in the ratio of crest, wave trough position number, Δ f (α)>0 represents that wind speed is more in crest, the top of spectrum is relatively round and smooth, Δ f (α)<0 represents that wind speed is more in trough, and the top of spectrum is relatively sharp.3rd is the asymmetry parameter S of Singularity spectrum function.During S=0, singular spectrum is symmetrical;During S > 0, singular spectrum peak value is to the right, and wind speed has the trend of increase;< when 0, singular spectrum peak value is to the left, and wind speed has the trend of reduction for S.
Step D. is according to the step C parameter [ω being previously mentioned, Δ f (α), S] wind speed is classified, and the parameter [ω corresponding to each time window, Δ f (α), S] adjust the size of time window further so that and each time window can completely and accurately comprise specific wind speed type.
The selection of number of categories is to affect the factor that wind speed classification results is less, selects four kinds of feature wind processes as criteria for classification, and the wind speed type of division is mainly, A: wind speed ascent stage, B: wind speed decreased stage, C: fluctuate the stage gently, D: spike phase.
The width of time window is to affect the factor that wind speed classification results is bigger, the selection of time window has two kinds: a kind of optimum window being width and fixing, another kind is the time window of variable-width, and now the width of window participates in the classified counting of wind speed together with ω, Δ f (α), S.
First make time window width (TimeWindowSize, TWZ) equal to fixed value Bt, i.e. a TWZt=Bt, (t=1,2 ... n).Wherein, because wind energy has bigger randomness, the effective range of fluctuations in wind speed rule is limited, thus undulatory property analysis to take a time range unsuitable long.And wind energy turbine set short-term wind-electricity power predicted time resolution employed herein is every 15 minutes points.So adopting 10 data points as the width of small time window Bt.Specifically comprising the following steps that of the optimal width of the time window of selection variable-width
Step D1: the starting point at air speed data starts, makes time window width(i=1,2 ... n) (whereinComprise front 10 data points (p1, p2 ... p10));
Step D2: carry out multifractal spectra analysis on TWZt, and extraction obtains characteristic parameter [ωt,Δf(α)t,St], and be equal toOn characteristic parameter [ωi,Δf(α)i,Si]。
Step D3: time window continues slip and obtains small time window(wherein ), it is carried out feature extraction and obtains [ωi+1,Δf(α)i+1,Si+1], according to the step C analysis done, investigate Δ f (α)iWith Δ f (α)i+1、SiWith Si+1Positive and negative.
Step D31: if Δ f (α)iWith Δ f (α)i+1、SiWith Si+1Positive and negative all identical each other, then showWithWind speed have the identical trend that rises or falls, enter step D32;If difference, enter step 4.
Step D32: investigate ωiAnd ωi+1If it is poor result Δ ω less than a certain threshold value η, shows that two time period degree of fluctuation are also suitable, now update time windowAnd indicate that TWZt belongs to ascent stage or decline stage.All training datas are carried out the average different index interval Δ ω that multi-fractal analysis of spectrum draws by threshold value ηAll, make η=Δ ωAll。
Step D321: if Δ ω >=η, then still makeAnd extract characteristic parameter, the characteristic parameter [ω after being updatedt,Δf(α)t,St], then make i=i+1 repeat step 3.
Step D4: if these two groups of numbers is positive and negative differing from each other, then showWithWind speed trend different, now have two kinds of situations: one isWithBeing in the mild fluctuation stage, two are in the critical time intervals of wind speed raising and lowering.
Consider next small time windowCalculate the population variance δ of these three small time window.
Wherein, n is equal to i+2 in this example,It is the mean wind speed of the n-th little time window,It it is the wind speed grand mean of front n little time window.
Step D41: if δ is less than setting value δ0, thenWithIt is in the mild fluctuation stage, now updates time windowSetting value δ0It it is the average variance being calculated all training datas drawing.
Step D42: if δ >=setting value δ0, thenMake TWZt+1 andAs the starting point that next time window is chosen, orderRepeat step D2.
Pass through above step, it is possible to wind speed is divided into ascent stage, decline stage according to the standard divided, fluctuates the stage gently.The dynamic width carrying out adaptive adjustment window according to current feature in time passage process.
Step E. adopts extreme learning machine, support vector machine and optimized regression power curve method that the wind speed type divided is sequentially carried out training, and produced predicting the outcome is carried out monthly accuracy comparison, select one of which method as the optimum monomer algorithm of this wind speed type, it is trained modeling, obtains the model trained.
By extreme learning machine (ELM), support vector machine (SVM), optimized regression power (ORPC) method, the historical data of tetra-kinds of models of A, B, C, D being tested respectively, every kind of Model Selection one of which method is as the optimal algorithm of this model.
Step F. carries out identical classification and modeling according to the standard of above-mentioned training sample classification model construction to testing sample, and corresponding optimum monomer algorithm in different Model Selection step E is predicted respectively, is finally combined obtaining final predicting the outcome.
According to step A to C, test data are carried out same process and analysis, same criteria for classification classification is adopted, it was predicted that model A: wind speed ascent stage, it was predicted that Model B: wind speed decreased stage according to step D, forecast model C: flat wave ejector half wind speed, it was predicted that model D: spike type wind speed.Test with corresponding tetra-kinds of forecast models of A, B, C, D with the extreme learning machine (ELM), support vector machine (SVM) or optimized regression power (ORPC) algorithm that have trained respectively.
The above, be only the preferred embodiments of the present invention, not the present invention is done any pro forma restriction, and those skilled in the art utilize the technology contents of the disclosure above to make a little simple modification, equivalent variations or decoration, all fall within protection scope of the present invention.
The content not being described in detail in this specification belongs to the known prior art of professional and technical personnel in the field.
Claims (9)
1. the wind power combination forecasting method based on fluctuations in wind speed feature extraction, it is characterised in that comprise the following steps:
A. the wind speed of training sample collection is normalized, to eliminate the difference of the amplitude that noise jamming is brought;
B. the wind speed after normalization is set up time window, and in this time window, carrying out multi-fractal analysis of spectrum, the major parameter of multi-fractal analysis of spectrum includes the asymmetry parameter S of singular index α, the Singularity spectrum function f (α) of multifractal spectra and Singularity spectrum function;
C. the width ω of the singular index α interval of each time window of com-parison and analysis, peak difference Δ f (α) of the Singularity spectrum function f (α) of multifractal spectra and the asymmetry parameter S of Singularity spectrum function;
D. according to the step C parameter [ω being previously mentioned, Δ f (α), S] wind speed is classified, and the parameter [ω corresponding to each time window, Δ f (α), S] adjust the size of time window further so that and each time window can completely and accurately comprise specific wind speed type;
E. the wind speed type that step D is divided by extreme learning machine, support vector machine and optimized regression power curve method is adopted to be sequentially carried out training, and produced predicting the outcome is carried out monthly accuracy comparison, select one of which method as the optimum monomer algorithm of this wind speed type, it is trained modeling, obtains the model trained;
F. carry out identical classification and modeling according to the standard of above-mentioned training sample classification model construction to testing sample, and corresponding optimum monomer algorithm in different Model Selection step E is predicted respectively, be finally combined obtaining final predicting the outcome.
2. the wind power combination forecasting method based on fluctuations in wind speed feature extraction as claimed in claim 1, it is characterised in that wind speed { ν described in step At, t=1,2 ... the normalization formula of n} is as follows:
3. the wind power combination forecasting method based on fluctuations in wind speed feature extraction as claimed in claim 1, it is characterised in that the step of multi-fractal analysis of spectrum described in step B includes:
B1, calculating singular index α: the singular index α local singularity representing wind speed, definition D (i) is i i square region, its central point be I (x, y), thenWherein, μ is defined in the probability measure on [0,1] [0,1], i=2n+1, n=0, and 1 ...;
B2, Singularity spectrum function f (α): f (α) calculating multifractal spectra represent the overall singularity of wind speed, and computational methods are: for each point, (x, singular index y) calculate αmax=max (α (i, j)), αmin(α (i, j)), by [α for=minmin, αmax] it is divided into N number of interval, obtain the central point singular value in each interval accordingly, replace the value of other points in interval by this value, then obtain f (α) according to formula (5);
If μ is defined in the probability measure on [0,1] [0,1], Dn is the increasing sequence of a positive integer composition, then define
Wherein* it is μ (II, j, nThe summation of the point of) ≠ 0;When the limit exists, if
Definition f1(α) convert for the Legendre of τ (q)
4. the wind power combination forecasting method based on fluctuations in wind speed feature extraction as claimed in claim 1, it is characterised in that the interval of ω singular index described in step C is [αmin, αmax], ω=αmax-αmin。
5. the wind power combination forecasting method based on fluctuations in wind speed feature extraction as claimed in claim 1, it is characterized in that, Δ f described in step C (α) represents that wind speed is in the ratio of crest, wave trough position number, Δ f (α)>0 represents that wind speed is more in crest, the top of spectrum is relatively round and smooth, Δ f (α)<0 represents that wind speed is more in trough, and the top of spectrum is relatively sharp.
6. the wind power combination forecasting method based on fluctuations in wind speed feature extraction as claimed in claim 1, it is characterised in that described in step C during S=0, singular spectrum is symmetrical;During S > 0, singular spectrum peak value is to the right, and wind speed has the trend of increase;< when 0, singular spectrum peak value is to the left, and wind speed has the trend of reduction for S.
7. the wind power combination forecasting method based on fluctuations in wind speed feature extraction as claimed in claim 1, it is characterized in that, step D select four kinds of feature wind processes as criteria for classification, the wind speed type divided includes, A: wind speed ascent stage, B: wind speed decreased stage, C: fluctuate the stage gently, D: spike phase.
8. the wind power combination forecasting method based on fluctuations in wind speed feature extraction as claimed in claim 1, it is characterised in that time window described in step D includes the time window of the fixing optimum window of width and variable-width.
9. the wind power combination forecasting method based on fluctuations in wind speed feature extraction as claimed in claim 8, it is characterised in that specifically comprising the following steps that of the optimal width of the time window of selection variable-width
Step D1: the starting point at air speed data starts, makes time window widthI=1,2 ... n, whereinComprise front 10 data points: p1, p2 ... p10;
Step D2: carry out multifractal spectra analysis on TWZt, and extraction obtains characteristic parameter [ωt,Δf(α)t,St], and be equal toOn characteristic parameter [ωi,Δf(α)i,Si];
Step D3: time window continues slip and obtainsWhereinComprise 10 data points: p11, p12 ... P20, it is carried out feature extraction and obtains [ωi+1,Δf(α)i+1,Si+1], according to the step C analysis done, investigate Δ f (α)iWith Δ f (α)i+1、SiAnd Si+1Positive and negative;
Step D31: if Δ f (α)iWith Δ f (α)i+1、SiWith Si+1Positive and negative all identical each other, then showWithWind speed have the identical trend that rises or falls, enter step D32;If difference, enter step D4;
Step D32: investigate ωiAnd ωi+1If it is poor result Δ ω less than a certain threshold value η, then shows that two time period degree of fluctuation are also suitable, now update time windowAnd indicate that TWZt belongs to ascent stage or decline stage;All training datas are carried out the average different index interval Δ ω that multi-fractal analysis of spectrum draws by threshold value ηAll, make η=Δ ωAll;
Step D321: if Δ ω is more than η, then still makeAnd extract characteristic parameter, the characteristic parameter [ω after being updatedt,Δf(α)t,St], then make i=i+1, repeat step D3;
Step D4: if Δ f (α)iWith Δ f (α)i+1、SiWith Si+1Positive and negative differing from each other, then showWithWind speed trend different, now have two kinds of situations: one isWithBeing in the mild fluctuation stage, two is the critical time intervals of wind speed raising and lowering;Now consider next small time windowCalculate the population variance δ of these three small time window;
Wherein, n is equal to i+2,It is the mean wind speed of the n-th little time window,It it is the wind speed grand mean of front n little time window;
Step D41: if δ is less than setting value δ0, thenWithIt is in the mild fluctuation stage, now updates time windowSetting value δ0It it is the average variance being calculated all training datas drawing;
Step D42: if δ >=setting value δ0, thenMake TWZt+1 andAs the starting point that next time window is chosen, orderRepeat step D2.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610140115.XA CN105719029B (en) | 2016-03-11 | 2016-03-11 | A kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610140115.XA CN105719029B (en) | 2016-03-11 | 2016-03-11 | A kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105719029A true CN105719029A (en) | 2016-06-29 |
CN105719029B CN105719029B (en) | 2019-08-27 |
Family
ID=56158716
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610140115.XA Active CN105719029B (en) | 2016-03-11 | 2016-03-11 | A kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105719029B (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106933778A (en) * | 2017-01-22 | 2017-07-07 | 中国农业大学 | A kind of wind power combination forecasting method based on climbing affair character identification |
CN107491833A (en) * | 2017-07-21 | 2017-12-19 | 清华大学 | A kind of method and apparatus of photovoltaic power generation power prediction |
CN107895202A (en) * | 2017-10-24 | 2018-04-10 | 天津大学 | Short-term wind speed forecasting method based on various visual angles wind speed mode excavation |
CN108763584A (en) * | 2018-06-11 | 2018-11-06 | 北京天泽智云科技有限公司 | A kind of method and its system of the filtering of wind power curve scatterplot |
CN110110892A (en) * | 2019-04-01 | 2019-08-09 | 新奥数能科技有限公司 | Power consumer load forecasting model selection method and device |
CN110334310A (en) * | 2019-07-01 | 2019-10-15 | 北京源清慧虹信息科技有限公司 | A kind of the determination method and determining device of power of fan curve |
CN111008725A (en) * | 2019-10-23 | 2020-04-14 | 中国农业大学 | Meteorological factor fluctuation feature extraction method for short-term wind power prediction |
CN111382787A (en) * | 2020-03-06 | 2020-07-07 | 芯薇(上海)智能科技有限公司 | Target detection method based on deep learning |
CN111915079A (en) * | 2020-07-30 | 2020-11-10 | 国电南瑞科技股份有限公司 | Hybrid KNN wind power prediction method and system |
CN112085254A (en) * | 2020-08-05 | 2020-12-15 | 中国地质大学(武汉) | Prediction method and model based on multi-fractal cooperative measurement gating cycle unit |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140172329A1 (en) * | 2012-12-17 | 2014-06-19 | Junshan Zhang | System and method for wind generation forecasting |
CN103903067A (en) * | 2014-04-09 | 2014-07-02 | 上海电机学院 | Short-term combination forecasting method for wind power |
CN104899665A (en) * | 2015-06-19 | 2015-09-09 | 国网四川省电力公司经济技术研究院 | Wind power short-term prediction method |
CN105046349A (en) * | 2015-06-25 | 2015-11-11 | 广东电网有限责任公司电力科学研究院 | Wind power prediction method considering wake effect |
-
2016
- 2016-03-11 CN CN201610140115.XA patent/CN105719029B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140172329A1 (en) * | 2012-12-17 | 2014-06-19 | Junshan Zhang | System and method for wind generation forecasting |
CN103903067A (en) * | 2014-04-09 | 2014-07-02 | 上海电机学院 | Short-term combination forecasting method for wind power |
CN104899665A (en) * | 2015-06-19 | 2015-09-09 | 国网四川省电力公司经济技术研究院 | Wind power short-term prediction method |
CN105046349A (en) * | 2015-06-25 | 2015-11-11 | 广东电网有限责任公司电力科学研究院 | Wind power prediction method considering wake effect |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106933778A (en) * | 2017-01-22 | 2017-07-07 | 中国农业大学 | A kind of wind power combination forecasting method based on climbing affair character identification |
CN107491833B (en) * | 2017-07-21 | 2020-05-15 | 清华大学 | Photovoltaic power generation power prediction method and device |
CN107491833A (en) * | 2017-07-21 | 2017-12-19 | 清华大学 | A kind of method and apparatus of photovoltaic power generation power prediction |
CN107895202A (en) * | 2017-10-24 | 2018-04-10 | 天津大学 | Short-term wind speed forecasting method based on various visual angles wind speed mode excavation |
CN107895202B (en) * | 2017-10-24 | 2022-03-04 | 天津大学 | Short-term wind speed prediction method based on multi-view wind speed mode mining |
CN108763584A (en) * | 2018-06-11 | 2018-11-06 | 北京天泽智云科技有限公司 | A kind of method and its system of the filtering of wind power curve scatterplot |
CN108763584B (en) * | 2018-06-11 | 2021-11-02 | 北京天泽智云科技有限公司 | Method and system for filtering scattered points of wind power curve |
CN110110892A (en) * | 2019-04-01 | 2019-08-09 | 新奥数能科技有限公司 | Power consumer load forecasting model selection method and device |
CN110334310A (en) * | 2019-07-01 | 2019-10-15 | 北京源清慧虹信息科技有限公司 | A kind of the determination method and determining device of power of fan curve |
CN110334310B (en) * | 2019-07-01 | 2023-04-07 | 北京源清慧虹信息科技有限公司 | Method and device for determining power curve of fan |
CN111008725A (en) * | 2019-10-23 | 2020-04-14 | 中国农业大学 | Meteorological factor fluctuation feature extraction method for short-term wind power prediction |
CN111008725B (en) * | 2019-10-23 | 2022-06-24 | 中国农业大学 | Meteorological factor fluctuation feature extraction method for short-term wind power prediction |
CN111382787A (en) * | 2020-03-06 | 2020-07-07 | 芯薇(上海)智能科技有限公司 | Target detection method based on deep learning |
CN111915079A (en) * | 2020-07-30 | 2020-11-10 | 国电南瑞科技股份有限公司 | Hybrid KNN wind power prediction method and system |
CN111915079B (en) * | 2020-07-30 | 2022-08-26 | 国电南瑞科技股份有限公司 | Hybrid KNN wind power prediction method and system |
CN112085254A (en) * | 2020-08-05 | 2020-12-15 | 中国地质大学(武汉) | Prediction method and model based on multi-fractal cooperative measurement gating cycle unit |
CN112085254B (en) * | 2020-08-05 | 2023-06-13 | 中国地质大学(武汉) | Prediction method and model based on multi-fractal cooperative measurement gating circulation unit |
Also Published As
Publication number | Publication date |
---|---|
CN105719029B (en) | 2019-08-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105719029A (en) | Combined wind power prediction method based on wind speed fluctuation characteristic extraction | |
CN106529814B (en) | Distributed photovoltaic ultra-short term prediction method based on Adaboost clustering and Markov chain | |
CN102663513B (en) | Utilize the wind power combined prediction modeling method of grey relational grade analysis | |
CN106933778A (en) | A kind of wind power combination forecasting method based on climbing affair character identification | |
CN106447086B (en) | One kind being based on the pretreated wind power combination forecasting method of wind farm data | |
CN103065202B (en) | Wind power plant ultrashort term wind speed prediction method based on combination kernel function | |
CN104573879A (en) | Photovoltaic power station output predicting method based on optimal similar day set | |
CN106374465B (en) | Short-term wind power forecast method based on GSA-LSSVM model | |
CN105426956A (en) | Ultra-short-period photovoltaic prediction method | |
CN104933483A (en) | Wind power forecasting method dividing based on weather process | |
CN108520269A (en) | A kind of wind speed forecasting method and forecasting wind speed system | |
CN110826791A (en) | Hybrid wind power prediction method based on long-time and short-time memory neural network | |
CN104573876A (en) | Wind power plant short-period wind speed prediction method based on time sequence long memory model | |
CN108549962B (en) | Wind power prediction method based on historical segmented sequence search and time sequence sparsification | |
CN103996079B (en) | Wind power weighting predication method based on conditional probability | |
CN109636003A (en) | A kind of power-system short-term load forecasting method of high precision of prediction | |
CN105574615A (en) | Spatial correlation and genetic algorithm (GA) based wind power forecast method for wavelet-BP neural network | |
CN110674864B (en) | Wind power abnormal data identification method comprising synchronous phasor measurement device | |
CN112633565B (en) | Photovoltaic power set interval prediction method | |
CN106611243A (en) | Residual correction method for wind speed prediction based on GARCH (Generalized ARCH) model | |
CN105005708A (en) | Generalized load characteristic clustering method based on AP clustering algorithm | |
CN111242353A (en) | Wind power combined prediction modeling and prediction method | |
CN111784543A (en) | System model for predicting short-term photovoltaic power generation system power generation based on hidden Markov model | |
CN107742000B (en) | Boiler combustion oxygen content modeling method | |
CN109193791B (en) | Wind power convergence tendency state-based quantification method based on improved shape value |
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 |