CN105719029B - A kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction - Google Patents
A kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction Download PDFInfo
- Publication number
- CN105719029B CN105719029B CN201610140115.XA CN201610140115A CN105719029B CN 105719029 B CN105719029 B CN 105719029B CN 201610140115 A CN201610140115 A CN 201610140115A CN 105719029 B CN105719029 B CN 105719029B
- Authority
- CN
- China
- Prior art keywords
- wind speed
- time window
- wind
- fluctuations
- feature extraction
- 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.)
- Active
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
- G06F18/24—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—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 kind of wind power combination forecasting methods based on fluctuations in wind speed feature extraction, comprising the following steps: the air speed data of training sample acquisition is normalized;To the wind speed settling time window after normalization, and multi-fractal spectrum analysis is carried out in the time window;Analyze the width ω of the singular index α value interval of more each time window, the peak difference Δ f (α) of Singularity spectrum function f (α), the asymmetry parameter S of f (α).Classified according to parameter [ω, Δ f (α), S] to wind speed, the size for a successive step time window of going forward side by side.It is successively trained using the classification of ExtremeLearningMachine, support vector machines and optimized regression power curve method to division, and monthly accuracy comparison is carried out to generated prediction result, optimal monomer algorithm of one of method as the category is selected, trained model is obtained.Identical classification and modeling are carried out to test sample, and select corresponding optimal monomer algorithm to predict respectively on different models, finally combination obtains final prediction result.
Description
Technical field
The present invention relates to a kind of wind power combination forecasting methods based on fluctuations in wind speed feature extraction, belong to electric system
Operation and control field.
Background technique
As permeability of the wind-powered electricity generation in the entire electric system in China steps up, fluctuation, intermittence and randomness
The problems such as caused voltage control, active power dispatch and system is stablized is more and more prominent, and accurate wind power prediction not only may be used
To reduce system reserve capacity and energy storage, system operation cost is reduced, while also contributing to mitigation wind power integration and power grid is rushed
It hits, improves operation of power networks reliability.
The intermittent nature of nature wind energy determines that wind power has very strong fluctuation, with wind-powered electricity generation number and dress
The continuous increase of machine capacity, once wind-powered electricity generation is connected to the grid, this power swing will bring huge to the safe and economic operation of power grid
Big challenge.Wind speed and wind power are accurately predicted in advance, the pressure of electric system peak regulation, frequency modulation can be alleviated, had
Effect improves power grid to the receiving ability of wind-powered electricity generation.
Currently, more and more extensive and deep for the research of wind-power electricity generation prediction project both at home and abroad.It generally believes, by wind speed
It is random change caused by wind power fluctuating change be that power system stability containing wind-power electricity generation, the main of control problem are not known
Factor.Therefore the wind velocity variation law for studying wind power plant, establishes reasonable wind farm wind velocity variation model, further to wind turbine
The power producing characteristics of group predict particularly important.
Traditional prediction method is mainly based upon short-term and ultra-short term prediction, such as the physics side based on numerical weather forecast
Method, the prediction such as Kalman filtering, artificial neural network, chaology, ExtremeLearningMachine, support vector machines based on statistical means
Method.But these methods are all model to be established to historical wind speed and history actual power, and then carry out to the following wind power
Prediction.This means that the input of prediction model has significant difference from the mapping relations of output under different wind friction velocities.Cause
This, single prediction model precision of prediction is to be improved under different wind friction velocities.
In addition, the frequent degree and strong journey in the fluctuations in wind speed of the period occurs when acute variation occurs for wind speed
Degree will also increase.Over time, model is established just for the historical data in a certain special time period not being able to satisfy entirely
The dynamic characteristic of fluctuations in wind speed in portion's time series.Moreover, the size of time window has the extraction and classification of data characteristics
Vital influence, too big and too small window be all unfavorable for signature analysis.
Summary of the invention
In view of the deficiencies in the prior art, the purpose of the present invention is to provide one kind to be based on fluctuations in wind speed feature extraction
Wind power combination forecasting method.
To achieve the above objectives, the present invention comprehensively considers following factor:
1, wind power plant NWP (numerical weather forecast) wind speed, actual power data;
2, fluctuations in wind speed characteristic;
3, the width of predicted time window;
4, 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, packet
Include following steps:
A. the wind speed of training sample acquisition is normalized, to eliminate the difference of the interference bring amplitude such as noise.
B. to the wind speed settling time window after normalization, and multi-fractal spectrum analysis is carried out in the time window,
The major parameter of multi-fractal spectrum analysis includes the Singularity spectrum function f (α) and Singularity spectrum function of singular index α, multifractal spectra
Asymmetry parameter S.
C. the width ω of the singular index α value interval of more each time window, the singular spectrum of multifractal spectra are analyzed
The peak difference Δ f (α) of function f (α) and the asymmetry parameter S of Singularity spectrum function.
D. the parameter [ω, Δ f (α), S] being previously mentioned according to step C classifies to wind speed, and according to each time window
Corresponding parameter [ω, Δ f (α), S] carrys out the size of further adjustment time window, enables each time window complete
It and accurately include specific wind speed type.
E. wind speed step D divided using ExtremeLearningMachine, support vector machines and optimized regression power curve method
Type is successively trained, and carries out monthly accuracy comparison to generated prediction result, selects one of method as this
The optimal monomer algorithm of wind speed type, is trained modeling, obtains trained model.
F. identical classification and modeling are carried out to test sample according to the standard of above-mentioned training sample classification model construction, and to not
It is predicted respectively with optimal monomer algorithm corresponding in model selection step E, is finally combined to obtain final prediction
As a result.
On the basis of above scheme, wind speed { ν described in step At, t=1,2 ... n } normalization formula it is as follows:
On the basis of above scheme, the step of multi-fractal spectrum analysis described in step B, includes:
B1, calculating singular index α: singular index α represent the local singularity of wind speed, and defining D (i) is the squared region i X i
Domain, central point are I (x, y), thenWherein, μ is defined in the probability in [0,1] X [0,1]
Estimate, i=2n+1, n=0,1 ...;
B2, calculate multifractal spectra Singularity spectrum function f (α): f (α) represent the global singularity of wind speed, calculation method
Are as follows: for the singular index of each point (x, y), calculate to obtain αmax=max (α (i, j)), αmin=min (α (i, j)), will
[αmin, αmax] it is divided into N number of section, the central point singular value in each section is accordingly obtained, replaces other points in section with the value
Value, f (α) is then found out according to formula (5);
If μ is defined in the probability measure in [0,1] X [0,1], Dn is the increasing sequence of positive integer composition, then fixed
Justice
WhereinIt is μ (II, j, nThe summation of the point of) ≠ 0;In the presence of the limit, if
Define f1The Legendre that (α) is τ (q) is converted
On the basis of above scheme, ω singular index value interval described in step C is [αmin, αmax], ω=αmax-
αmin.ω is bigger, and expression wind speed profile is more uneven, and fluctuations in wind speed is more violent, when the amplitude of fluctuations in wind speed becomes smaller, α~f (α)
Composing has the tendency that obviously narrowing.
On the basis of above scheme, Δ f (α) described in step C indicates that wind speed is in the ratio of wave crest, wave trough position number
Example, Δ f (α)>0 indicate that wind speed is more in wave crest, and the top of spectrum is relatively round and smooth, and Δ f (α)<0 indicates that wind speed is more located
It is relatively sharp in the top of trough, spectrum.
On the basis of above scheme, when S=0 described in step C, singular spectrum is symmetrical;When S > 0, unusual spectrum peak
To the right, wind speed has increased trend;When S < 0, unusual spectrum peak is to the left, and wind speed has the tendency that reduction.
On the basis of above scheme, in step D, select four kinds of feature wind processes as classification standard, the wind speed of division
Type includes A: wind speed ascent stage, B: the wind speed decreased stage, C: smooth fluctuation stage, D: spike phase.
On the basis of above scheme, time window described in step D includes the fixed optimal window of width and width can
The time window of change;
Specific step is as follows for the optimal width of the time window of selection variable-width:
Step D1: start in the starting point of air speed data, enable time window widthI=1,2 ... n, wherein
Include preceding 10 data points: p1, p2 ... p10;
Step D2: multifractal spectra analysis is carried out on TWZt, and extracts and obtains characteristic parameter [ωt,Δf(α)t,St],
And it is equal toOn characteristic parameter [ωi,Δf(α)i,Si];
Step D3: time window continues sliding and obtainsWhereinInclude 10 data points: p11, p12 ... P20,
Feature extraction is carried out to it and obtains [ωi+1,Δf(α)i+1,Si+1], according to the analysis that step C is done, investigate Δ f (α)iWith Δ f
(α)i+1、SiAnd Si+1It is positive and negative;
Step D31: if Δ f (α)iWith Δ f (α)i+1、SiWith Si+1It is positive and negative all identical each other, then showWithWind
Speed have it is identical rise or fall trend, enter step D32;D4 is entered step if different;
Step D32: ω is investigatediAnd ωi+1If its result Δ ω made the difference is less than a certain threshold value η, show two times
Section degree of fluctuation is also suitable, at this time renewal time windowAnd indicate that TWZt belongs to ascent stage and still declines
Stage;Threshold value η is the average different index value interval Δ ω that all training datas are carried out with multi-fractal spectrum analysis and is obtainedAll, enable
η=Δ ωAll;
Step D321: it if Δ ω is greater than η, still enablesAnd characteristic parameter is extracted, it obtains updating it
Characteristic parameter [ω afterwardst,Δf(α)t,St], i=i+1 is then enabled, step D3 is repeated;
Step D4: if Δ f (α)iWith Δ f (α)i+1、SiWith Si+1It is positive and negative differing from each other, then showWithWind
Fast trend is different, and there are two types of situations at this time: first is thatWithIn the smooth fluctuation stage, second is that wind speed raising and lowering
Critical time intervals.Next small time window is considered at this timeCalculate the population variance δ of these three small time windows;
Wherein, n is equal to i+2,It is the mean wind speed of n-th small time window,Be preceding n small time windows wind speed it is always equal
Value;
Step D41: if δ is less than setting value δ0, thenWithIn the smooth fluctuation stage, renewal time window at this timeSetting value δ0It is the average variance that all training datas are calculated.
Step D42: if δ >=setting value δ0, thenEnable TWZt+1 andIt is chosen as next time window
Starting point, enableRepeat step D2.
The utility model has the advantages that 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 can not simulation input and the mapping relations of output shown under different wind friction velocities
The drawbacks of otherness of work, and solve that single prediction model precision of prediction under different wind friction velocities is not high to ask
Topic.
Detailed description of the invention
Fig. 1 is that the present invention is based on the flow diagrams of the wind power combination forecasting method of fluctuations in wind speed feature extraction.
Fig. 2 is the schematic diagram of multifractal spectra corresponding to different fluctuations in wind speed features.
Fig. 3 is the sliding time window schematic diagram of variable-width.
Fig. 4 is combinational algorithm flow diagram.
Specific embodiment
The above is merely an overview of the technical solutions of the present invention, in order to enable those skilled in the art to better understand the present invention
Technological means, below in conjunction with attached drawing, the present invention is described in further detail with specific embodiment.
As shown in Figs 1-4,
The wind speed that step A. acquires training sample is normalized, to eliminate the difference of the interference bring amplitude such as noise
It is different.Wind speed { νt, t=1,2 ... n } normalization formula it is as follows:
Step B. carries out multifractal spectra point to the wind speed settling time window after normalization in the time window
Analysis.
It calculates singular index α: singular index α and represents the local singularity of wind speed, defining D (i) is i X i square region,
Its central point is I (x, y), then(wherein, μ is defined in the survey of the probability in [0,1] X [0,1]
Degree, i=2n+1, n=0,1 ...).I is related with the positioning of calculating, generally takes i≤3, and at this moment α (x, y) reflects local unusual
Property, take big i to reflect wider singularity.
Calculate multifractal spectra Singularity spectrum function f (α): f (α) represents the global singularity of wind speed, calculation method are as follows:
For the singular index of each point (x, y), α is calculated to obtainmax=max (α (i, j)), αmin=min (α (i, j)), by [αmin,
αmax] it is divided into N number of section, the central point singular value in each section is accordingly obtained, the value for replacing other in section to put with the value,
Then f (α) is found out according to formula (5).
If μ is defined in the probability measure in [0,1] X [0,1], Dn is the increasing sequence of positive integer composition, then fixed
Justice
WhereinIt is μ (II, j, nThe summation of the point of) ≠ 0.In the presence of the limit, if
Define f1The Legendre that (α) is τ (q) is converted
Step C. analyzes the width ω of the singular index α value interval of more each time window, the surprise of multifractal spectra
The peak difference f (α) of different spectral function, the asymmetry parameter S of Singularity spectrum function.
Three major parameters are chosen to describe the degree of strength of multi-fractal property, first is singular index value interval
[αmin, αmax] width ω=αmax-αmin, ω is bigger, and expression wind speed profile is more uneven, and fluctuations in wind speed is more violent, when wind speed wave
When dynamic amplitude becomes smaller, α~f (α), which is composed, to be had the tendency that obviously narrowing.Second be Singularity spectrum function peak difference Δ f (α),
Δ f (α) indicates that wind speed is in the ratio of wave crest, wave trough position number, and Δ f (α) > 0 indicates that wind speed is more in wave crest, spectrum
Top is relatively round and smooth, and Δ f (α) < 0 indicates that wind speed is more in trough, and the top of spectrum is relatively sharp.Third is singular spectrum
The asymmetry parameter S of function.When S=0, singular spectrum is symmetrical;When S > 0, unusual spectrum peak is to the right, wind speed have it is increased become
Gesture;When S < 0, unusual spectrum peak is to the left, and wind speed has the tendency that reduction.
The parameter [ω, Δ f (α), S] that step D. is previously mentioned according to step C classifies to wind speed, and according to each time
Parameter corresponding to window [ω, Δ f (α), S] carrys out the size of further adjustment time window, enables each time window
It completely and accurately include specific wind speed type.
The selection of classification number is to influence the lesser factor of wind speed classification results, selects four kinds of feature wind processes as contingency table
The wind speed type of standard, division is mainly A: wind speed ascent stage, B: the wind speed decreased stage, C: smooth fluctuation stage, D: spike rank
Section.
The width of time window is to influence the biggish factor of wind speed classification results, and there are two types of the selections of time window: a kind of
It is the fixed optimal window of width, another kind is the time window of variable-width, the at this time width of window and ω, Δ f (α), S mono-
Act the classified calculating for participating in wind speed.
Time window width (Time Window Size, TWZ) is enabled to be equal to a fixed value Bt, i.e. TWZt=Bt first,
(t=1,2 ... n).Wherein, because wind energy has biggish randomness, the effective range of fluctuations in wind speed rule is limited, institute
Take a time range unsuitable too long with fluctuation analysis.And wind power plant short-term wind-electricity power predicted time employed herein is differentiated
Rate is every 15 minutes points.So the width using 10 data points as small time window Bt.Select variable-width when
Between the optimal width of window specific step is as follows:
Step D1: start in the starting point of air speed data, enable time window width(i=1,2 ... n) (whereinInclude preceding 10 data points (p1, p2 ... p10));
Step D2: multifractal spectra analysis is carried out on TWZt, and extracts and obtains characteristic parameter [ωt,Δf(α)t,St],
And it is equal toOn characteristic parameter [ωi,Δf(α)i,Si]。
Step D3: time window continues sliding and obtains small time window(wherein ),
Feature extraction is carried out to it and obtains [ωi+1,Δf(α)i+1,Si+1], according to the analysis that step C is done, investigate Δ f (α)iWith Δ f
(α)i+1、SiWith Si+1It is positive and negative.
Step D31: if Δ f (α)iWith Δ f (α)i+1、SiWith Si+1It is positive and negative all identical each other, then showWithWind
Speed have it is identical rise or fall trend, enter step D32;4 are entered step if different.
Step D32: ω is investigatediAnd ωi+1, show two periods if its result made the difference Δ ω is less than a certain threshold value η
Degree of fluctuation is also suitable, at this time renewal time windowAnd indicate that TWZt belongs to ascent stage and still descends depression of order
Section.Threshold value η is the average different index value interval Δ ω that all training datas are carried out with multi-fractal spectrum analysis and is obtainedAll, enable η
=Δ ωAll。
Step D321: if Δ ω >=η, still enablesAnd characteristic parameter is extracted, after obtaining update
Characteristic parameter [ωt,Δf(α)t,St], then enable i=i+1 repeat step 3.
Step D4: if this two groups of numbers is positive and negative differing from each other, showWithWind speed trend it is different, at this time
There are two types of situations: first is thatWithIn the smooth fluctuation stage, second is that being 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 windows.
Wherein, n is equal to i+2 in this example,It is the mean wind speed of n-th small time window,It is preceding n small time windows
Wind speed grand mean.
Step D41: if δ is less than setting value δ0, thenWithIn the smooth fluctuation stage, renewal time window at this timeSetting value δ0It is the average variance that all training datas are calculated.
Step D42: if δ >=setting value δ0, thenEnable TWZt+1 andIt is chosen as next time window
Starting point, enableRepeat step D2.
Pass through above step, so that it may which wind speed is divided into ascent stage, decline stage, flat wave according to the standard divided
The dynamic stage.Dynamically according to current feature come the width of the adjustment window of adaptability during time passage.
Step E. is using ExtremeLearningMachine, support vector machines and optimized regression power curve method to the wind speed class divided
Type is successively trained, and carries out monthly accuracy comparison to generated prediction result, selects one of method as the wind
The optimal monomer algorithm of fast type, is trained modeling, obtains trained model.
Respectively with ExtremeLearningMachine (ELM), support vector machines (SVM), optimized regression power (ORPC) method to A, B, C, D
The historical data of four kinds of models is tested, and every kind of model selects optimal algorithm of one of method as the model.
Step F. carries out identical classification and modeling to test sample according to the standard of above-mentioned training sample classification model construction, and
Optimal monomer algorithm corresponding in different models selection step E is predicted respectively, is finally combined to obtain final
Prediction result.
Test data is similarly handled and analyzed according to step A to C, uses same contingency table according to step D
Quasi- classification, prediction model A: wind speed ascent stage, prediction model B: wind speed decreased stage, prediction model C: flat wave ejector half wind
Speed, prediction model D: spike type wind speed.Respectively with trained ExtremeLearningMachine (ELM), support vector machines (SVM) or excellent
Change recurrence power (ORPC) algorithm to be tested with tetra- kinds of prediction models of corresponding A, B, C, D.
The above is only preferred embodiments of the invention, is not intended to limit the present invention in any form, ability
Field technique personnel make a little simple modification, equivalent variations or decoration using the technology contents of the disclosure above, all fall within the present invention
Protection scope in.
The content being not described in detail in this specification belongs to the prior art well known to professional and technical personnel in the field.
Claims (8)
1. a kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction, which comprises the following steps:
A. the wind speed of training sample acquisition is normalized, to eliminate the difference of noise jamming bring amplitude;
B. to the wind speed settling time window after normalization, and multi-fractal spectrum analysis is carried out in the time window, it is multiple
The major parameter of fractal spectrum analysis include singular index α, multifractal spectra Singularity spectrum function f (α) and Singularity spectrum function pair
Title property parameter S;
C. the width ω of the singular index α value interval of more each time window, the Singularity spectrum function f of multifractal spectra are analyzed
The peak difference △ f (α) of (α) and the asymmetry parameter S of Singularity spectrum function;
D. the parameter [ω, △ f (α), S] being previously mentioned according to step C classifies to wind speed, and right according to each time window institute
The parameter [ω, △ f (α), S] answered carrys out the size of further adjustment time window, enables each time window completely and essence
True includes specific wind speed type;
E. wind speed type step D divided using ExtremeLearningMachine, support vector machines and optimized regression power curve method
It is successively trained, and monthly accuracy comparison is carried out to generated prediction result, select one of method as the wind speed
The optimal monomer algorithm of type, is trained modeling, obtains trained model;
F. identical classification and modeling are carried out to test sample according to the standard of above-mentioned training sample classification model construction, and to different moulds
Corresponding optimal monomer algorithm is predicted respectively in type selection step E, is finally combined to obtain final prediction result;
The step of multi-fractal spectrum analysis described in step B includes:
B1, calculating singular index α: singular index α represent the local singularity of wind speed, and defining D (i) is i X i square region,
Its central point is I (x, y), thenWherein, μ is defined in the survey of the probability in [0,1] X [0,1]
Degree, i=2n+1, n=0,1 ...;
B2, calculate multifractal spectra Singularity spectrum function f (α): f (α) represent the global singularity of wind speed, calculation method are as follows:
For the singular index of each point (x, y), α is calculated to obtainmax=max (α (i, j)), αmin=min (α (i, j)), by [αmin,
αmax] it is divided into N number of section, the central point singular value in each section is accordingly obtained, the value for replacing other in section to put with the value,
Then f (α) is found out according to formula (5);
If μ is defined in the probability measure in [0,1] X [0,1], Dn is the increasing sequence of positive integer composition, then defines
WhereinIt is μ (II, j, nThe summation of the point of) ≠ 0;In the presence of the limit, if
Define f1The Legendre that (α) is τ (q) is converted
2. as described in claim 1 based on the wind power combination forecasting method of fluctuations in wind speed feature extraction, which is characterized in that
Wind speed { ν described in step At, t=1,2 ... n } normalization formula it is as follows:
3. as described in claim 1 based on the wind power combination forecasting method of fluctuations in wind speed feature extraction, which is characterized in that
ω singular index value interval described in step C is [αmin, αmax], ω=αmax-αmin。
4. as described in claim 1 based on the wind power combination forecasting method of fluctuations in wind speed feature extraction, which is characterized in that
△ f (α) described in step C indicates that wind speed is in the ratio of wave crest, wave trough position number, and △ f (α) > 0 indicates that wind speed is more located
Relatively round and smooth in the top of wave crest, spectrum, △ f (α) < 0 indicates that wind speed is more in trough, and the top of spectrum is relatively sharp.
5. as described in claim 1 based on the wind power combination forecasting method of fluctuations in wind speed feature extraction, which is characterized in that
When S=0 described in step C, singular spectrum is symmetrical;When S > 0, unusual spectrum peak is to the right, and wind speed has increased trend;When S < 0,
Unusual spectrum peak is to the left, and wind speed has the tendency that reduction.
6. as described in claim 1 based on the wind power combination forecasting method of fluctuations in wind speed feature extraction, which is characterized in that
Select four kinds of feature wind processes as classification standard in step D, the wind speed type of division includes A: wind speed ascent stage, B: wind
Fast decline stage, C: smooth fluctuation stage, D: spike phase.
7. as described in claim 1 based on the wind power combination forecasting method of fluctuations in wind speed feature extraction, which is characterized in that
Time window described in step D includes the time window of width fixed optimal window and variable-width.
8. as claimed in claim 7 based on the wind power combination forecasting method of fluctuations in wind speed feature extraction, which is characterized in that
Specific step is as follows for the optimal width of the time window of selection variable-width:
Step D1: start in the starting point of air speed data, enable time window widthI=1,2 ... n, whereinInclude
Preceding 10 data points: p1, p2 ... p10;
Step D2: multifractal spectra analysis is carried out on TWZt, and extracts and obtains characteristic parameter [ωt,△f(α)t,St], and
It is equal toOn characteristic parameter [ωi,△f(α)i,Si];
Step D3: time window continues sliding and obtainsWhereinInclude 10 data points: p11, p12 ... P20, to its into
Row feature extraction obtains [ωi+1,△f(α)i+1,Si+1], according to the analysis that step C is done, investigate △ f (α)iWith △ f (α)i+1、Si
And Si+1It is positive and negative;
Step D31: if △ f (α)iWith △ f (α)i+1、SiWith Si+1It is positive and negative all identical each other, then showWithWind speed have
It is identical to rise or fall trend, enter step D32;D4 is entered step if different;
Step D32: ω is investigatediAnd ωi+1If its result △ ω made the difference is less than a certain threshold value η, show two period fluctuations
Degree is also suitable, at this time renewal time windowAnd indicate that TWZt belongs to ascent stage or decline stage;Threshold
Value η is the average different index value interval △ ω that all training datas are carried out with multi-fractal spectrum analysis and is obtainedAll, enable η=△
ωAll;
Step D321: it if △ ω is greater than η, still enablesAnd characteristic parameter is extracted, after being updated
Characteristic parameter [ωt,△f(α)t,St], i=i+1 is then enabled, step D3 is repeated;
Step D4: if △ f (α)iWith △ f (α)i+1、SiWith Si+1It is positive and negative differing from each other, then showWithWind speed trend
Different, there are two types of situations at this time: first is thatWithIn the smooth fluctuation stage, second is that wind speed raising and lowering it is critical when
Section;Next small time window is considered at this timeCalculate the population variance δ of these three small time windows;
Wherein, n is equal to i+2,It is the mean wind speed of n-th small time window,It is the wind speed grand mean of preceding n small time windows;
Step D41: if δ is less than setting value δ0, thenWithIn the smooth fluctuation stage, renewal time window at this timeSetting value δ0It is the average variance that all training datas are calculated;
Step D42: if δ >=setting value δ0, thenEnable TWZt+1 andIt is risen as what next time window was chosen
Point enablesRepeat 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 CN105719029A (en) | 2016-06-29 |
CN105719029B true 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) |
Families Citing this family (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 |
CN107491833B (en) * | 2017-07-21 | 2020-05-15 | 清华大学 | Photovoltaic power generation power prediction method and device |
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 |
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 |
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 |
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 |
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 |
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 |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
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 |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9460478B2 (en) * | 2012-12-17 | 2016-10-04 | Arizona Board Of Regents On Behalf Of Arizona State University | System and method for wind generation forecasting |
-
2016
- 2016-03-11 CN CN201610140115.XA patent/CN105719029B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
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 |
Also Published As
Publication number | Publication date |
---|---|
CN105719029A (en) | 2016-06-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105719029B (en) | A kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction | |
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 | |
CN102880810B (en) | Wind power prediction method based on time sequence and neural network method | |
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 | |
CN107609697B (en) | A kind of Wind power forecasting method | |
CN106499581B (en) | A kind of adaptive method for controlling torque of wind energy conversion system considering variation turbulent flow wind regime | |
CN108549962B (en) | Wind power prediction method based on historical segmented sequence search and time sequence sparsification | |
CN106875033A (en) | A kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting | |
CN104933483A (en) | Wind power forecasting method dividing based on weather process | |
CN112186761B (en) | Wind power scene generation method and system based on probability distribution | |
CN103835878A (en) | Maximum power point tracing control method based on neural network optimization starting rotating speed | |
CN110598983A (en) | Cascade reservoir optimal scheduling method of self-adaptive improved particle swarm algorithm | |
CN105134487A (en) | Wind turbine maximum power point tracking control method with turbulence frequency factor considered | |
CN104732300A (en) | Neural network wind power short-term forecasting method based on fuzzy partition theory | |
CN102749471B (en) | A kind of short-term wind speed, wind power forecasting method | |
CN110739726A (en) | multi-type power supply capacity long-term planning method considering offshore wind power access | |
CN106611243A (en) | Residual correction method for wind speed prediction based on GARCH (Generalized ARCH) model | |
CN109409019A (en) | A kind of pre-bending optimization method of pneumatic equipment bladess aerodynamic configuration | |
CN114386718A (en) | Wind power plant output power short-time prediction algorithm combined with particle swarm neural network | |
CN104484704A (en) | Wind power plant theoretical power determination method based on BP neural network | |
Zhang et al. | Ultra-short-term wind power forecasting model based on time-section fusion and pattern classification | |
CN103473607A (en) | Ultra-short-term wind power prediction method according to off-line track characteristic optimization and real-time extrapolation model matching | |
CN112803403B (en) | Offshore wind power clustering effect multi-level evaluation method based on time-frequency characteristics |
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 |