CN106933778A - A kind of wind power combination forecasting method based on climbing affair character identification - Google Patents
A kind of wind power combination forecasting method based on climbing affair character identification Download PDFInfo
- Publication number
- CN106933778A CN106933778A CN201710052723.XA CN201710052723A CN106933778A CN 106933778 A CN106933778 A CN 106933778A CN 201710052723 A CN201710052723 A CN 201710052723A CN 106933778 A CN106933778 A CN 106933778A
- Authority
- CN
- China
- Prior art keywords
- climbing
- wind
- event
- power
- historical
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000009194 climbing Effects 0.000 title claims abstract description 118
- 238000013277 forecasting method Methods 0.000 title claims abstract description 17
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 45
- 238000000034 method Methods 0.000 claims abstract description 31
- 238000012360 testing method Methods 0.000 claims abstract description 25
- 238000007906 compression Methods 0.000 claims abstract description 13
- 230000006835 compression Effects 0.000 claims abstract description 13
- 239000000284 extract Substances 0.000 claims abstract description 8
- 238000000605 extraction Methods 0.000 claims abstract description 6
- 238000007619 statistical method Methods 0.000 claims abstract description 4
- 230000005611 electricity Effects 0.000 claims description 30
- 239000011159 matrix material Substances 0.000 claims description 14
- 210000002569 neuron Anatomy 0.000 claims description 9
- 238000001228 spectrum Methods 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 5
- 238000004138 cluster model Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 244000025254 Cannabis sativa Species 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000010183 spectrum analysis Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 238000005192 partition Methods 0.000 claims 1
- 238000013459 approach Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 230000001172 regenerating effect Effects 0.000 description 3
- 238000013144 data compression Methods 0.000 description 2
- 230000007935 neutral effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000002028 Biomass Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000005034 decoration Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000012252 genetic analysis Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 238000009987 spinning Methods 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2155—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling
-
- 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
- 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
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Business, Economics & Management (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Software Systems (AREA)
- Computational Mathematics (AREA)
- Economics (AREA)
- General Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Marketing (AREA)
- Evolutionary Computation (AREA)
- Operations Research (AREA)
- Evolutionary Biology (AREA)
- Algebra (AREA)
- Bioinformatics & Computational Biology (AREA)
- Databases & Information Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Tourism & Hospitality (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Computing Systems (AREA)
- Quality & Reliability (AREA)
Abstract
The invention discloses a kind of wind power combination forecasting method based on climbing affair character identification, comprise the following steps：First, the historical wind speed and power data of wind power plant are processed using wavelet de-noising method, respectively obtains smooth curve；2nd, the curve that will be obtained carries out the feature recognition and extraction of climbing event using compression algorithm to it；3rd, the characteristic value of the climbing event of the wind power for being obtained step 2 using the method for fuzzy clustering is classified；4th, different climbing event types is trained respectively using statistical method and sets up forecast model；5th, the real-time wind speed test data and realtime power test data to wind power plant carry out step one to two, extract the characteristic value of the climbing event of real-time wind speed and power test data；6th, real-time wind speed and the characteristic value of the climbing event of power test data of the forecast model for being obtained using step 4 to extracting are classified, and are predicted using forecast model, finally obtain final combined prediction result.
Description
Technical field
It is more particularly to a kind of based on wind power plant climbing affair character the present invention relates to operation and control of electric power system field
The wind power combination forecasting method of identification.
Background technology
With the increasingly depleted and increasingly serious, wind energy, the sun of energy quagmire of the non-renewable resources such as coal, oil
The regenerative resources such as energy, tide energy and biomass energy worldwide more receive much concern.Wind-power electricity generation is regenerative resource
Technology is most ripe in generation technology, most Development volue regenerative resource.Development wind-powered electricity generation adjusts energy for ensureing energy security
Source structure, mitigates environmental pollution, realizes that sustainable development etc. all has very important significance.
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 safety and economic operation of power network
Big challenge.Wind speed and wind power are accurately predicted in advance, the pressure of power system peak regulation, frequency modulation can be alleviated, had
Effect improves receiving ability of the power network to wind-powered electricity generation.
At present, with the raising of wind-powered electricity generation permeability in power network, scheduling and safety and stability of the wind-electricity integration to power system
Bring very big challenge so that the Accurate Prediction of wind power becomes particularly important.Although the power forecasting method of routine is
Wind power plant service requirement is reached, but when the i.e. wind power climbing event of instantaneous notable fluctuation for wind power occur, can be led
Net instantaneity or permanent fault are sent a telegraph, great economic loss is caused.Therefore, study the characteristic of wind-powered electricity generation climbing and standard is carried out to it
Really prediction, on the one hand can improve the accuracy that power network is estimated grid connected wind power power, contribute to power scheduling in advance to wind-powered electricity generation
Climbing event is taken precautions against, and safeguards the stable operation of power network；On the other hand conventional power unit reply wind-powered electricity generation climbing event can be reduced
Regulation and control burden, contribute to reduce power network spinning reserve capacity, reduce operating cost, the peace grid-connected to the extensive concentration of wind-powered electricity generation
Full operation of stabilizing the economy is significant.
The content of the invention
In order to overcome the above-mentioned deficiencies of the prior art, the invention provides a kind of wind-powered electricity generation based on climbing affair character identification
Power combination Forecasting Methodology.The method has taken into full account the popular research problem of current wind power prediction, the i.e. climbing of wind-powered electricity generation
The power prediction of event.Using the compressed data means in a kind of brand-new data storage field, can be more rapidly and effectively
Carry out the feature recognition and extraction of wind-powered electricity generation climbing event.Additionally, according to climbing event complex genesis, with non-linear, hysteresis quality,
The characteristics of accurate Mathematical Modeling is difficult to set up to characterize, using the method for fuzzy clustering, concentrates from historical sample and searches out phase
Like event classification and be trained and set up forecast model.The method strong adaptability, can effectively improve the prediction of wind-powered electricity generation
Precision.The technical solution adopted in the present invention considers following factor：
1st, wind power plant historical wind speed data and historical power data；
2nd, the climbing direction of wind-powered electricity generation climbing event, climbing rate, climbing amplitude, time of climb and climbing time started；
3rd, the history year rule change of local wind farm data.
On the basis of factors above, a kind of wind power combination forecasting method based on climbing affair character identification, bag
Include following steps：
S1, the historical wind speed and historical power data of wind power plant are processed using wavelet de-noising method, respectively obtain light
Sliding historical wind speed curve and historical power curve, to eliminate the interference of grass；
S2, using compression algorithm, the smooth historical wind speed curve and historical power curve obtained to step S1 are climbed
The feature recognition of slope event, then will identify that the climbing event come carries out characteristics extraction；
S3, the characteristic value of the climbing event for being extracted step S2 using the method for fuzzy clustering are classified, and are made
Optimum cluster model, obtains different climbing event types；
S4, the different climbing event type obtained to step S3 using statistical method are trained and set up prediction respectively
Model；
S5, real-time wind speed test data and realtime power test data to wind power plant carry out the process of step S1 to S2,
Extract the characteristic value of the climbing event of real-time wind speed test data and realtime power test data；
S6, the forecast model obtained using step S4 are tested the real-time wind speed test data and realtime power that extract
The characteristic value of the climbing event of data is classified, and is predicted using forecast model, finally obtains final combined prediction
As a result.
On the basis of such scheme, the process of wavelet de-noising is described in step S1：
S11, the time series to historical wind speed data and historical power data carry out Fourier's treatment, by discrete signal
Switch to continuous signal；
S12, power spectrumanalysis is carried out to continuous signal, draw its spectrum waveform；
The spectrum waveform feature of S13, analysis continuous signal, and then waveform is fitted, parametric equation is obtained after fitting,
Solution draws Power Spectrum Distribution result；
Inverse transformation is carried out after S14, the parameter of change equation, filtered result is drawn.
On the basis of such scheme, compression algorithm described in step S2 is revolving door (SDT) algorithm, and the revolving door is calculated
Method is a kind of trends of straight line compression algorithm.
Event of climbing on the basis of such scheme, described in step S2 is based on based on Fast Field (FFP) algorithm
The climbing event definition of FFP algorithms is：If P is wind-powered electricity generation historical power time series,It is smooth by the wind-powered electricity generation after denoising
Historical power curve,It is the wind-powered electricity generation historical power smooth sequence after being changed through SDT algorithms, the historical power smooth sequence
For recognizing climbing event, it is calculated by formula (1)：
Wherein SDT is the SQL with compression algorithm principle；Δ E is responsive parameter, and value is smaller, be can recognize that
The climbing event procedure duration it is shorter, climbing height it is lower；Value is bigger, the recognizable climbing event procedure duration
More long, climbing height is higher.
When the conversion signal of wind-powered electricity generation historical power time seriesAbsolute value greater-than match climb event criterion
During threshold parameter λ, i.e.,
WhenDuring for negative value, what is identified is lower climbing event, whenBe on the occasion of when, what is identified is upper climbing thing
Part.
On the basis of such scheme, the threshold value of the upper climbing event is 20.0%cap, and the threshold value of lower climbing event is
15.0%cap.
On the basis of such scheme, the characteristic value includes：Climbing rate Rr, climbing amplitude RA, time started RTsWith hold
Continuous time RTl。
On the basis of such scheme, the specific steps that step S3 is classified characteristic value using the method for fuzzy clustering
It is as follows：
S31, assume that each climbing event has a m characteristic value, and n sample composition eigenmatrix：
S32, the eigenmatrix in S31 is normalized：
S33, the fuzzy membership that each characteristic value is calculated using the method for fuzzy clustering：
Wherein,It is i-th data to the j degree of membership of cluster centre；I=1,2 .., m, j=1,2 .., c, c are number
According to hard plot clusters number；vijAnd σijCenter and the width of Gaussian function are represented respectively；
S34, optimum cluster model is made by the nearest principle of fuzzy membership, obtain different climbing event types.
On the basis of such scheme, using ExtremeLearningMachine (ELM) algorithm to different climbing event classes in step S4
Type is trained respectively, and regulation ELM algorithm parameters make various climbing events set up an optimal Clustering Model.
Need adjustment input weights and hidden layer biasing different from conventional function approximation theory, in ExtremeLearningMachine algorithm only
Wanting activation primitive can infinitely lead, and input weights and hidden layer biasing can be randomly assigned, once the value being randomly assigned is being calculated
Calligraphy learning starts to be updated in algorithm parameter, and hidden layer output matrix H is just uniquely determined and keeps constant, instruction
Practice a neutral net to be equivalent to find a least square solution for linear system, if the number of hidden layer node and input
Sample number is equal, then this Single hidden layer feedforward neural networks (SLFN) just can approach these training examples with zero error.
The ExtremeLearningMachine algorithm is concretely comprised the following steps：
1) input weights and hidden layer biasing are randomly assigned；
For N number of different training sample { xj,yj, xjIt is n × 1 dimensional input vector, yjIt is m × 1 dimension output vector, tool
There is D hidden node and can infinitely lead activation primitivegX the SLFN structures of () can approach this N number of sample with zero error, that is, exist
βi,ωiAnd bi, make ∑ | | yj-rj| |=0, can also be written as form：
In formula：ωiIt is n × 1 dimensional vector, represents the connection weight of input layer and the neuron of hidden layer；βiFor m × 1 tie up to
Amount, represents the connection weight of hidden layer and output layer neuron；biIt is the biasing of hidden layer neuron；rjIt is the output of network；g
X () is activation primitive, can be " sig ", " rbf ", the diversified forms such as " sin ".
2) hidden layer output matrix H is calculated；
This N number of equation can be write as matrix form：
H β=R (7)
Wherein, β is output layer weight matrix, and R is desired output；
3) global optimum's output weights of ELM algorithms are calculated；
H+It is the generalized inverse matrix of H.
Beneficial effect：
Wind power combined prediction the method based on climbing affair character identification of the present invention, has taken into full account mesh
The power prediction of the climbing event of the popular research problem of preceding wind power prediction, i.e. wind-powered electricity generation.Stored up using a kind of brand-new data
The compressed data means in field are deposited, the feature recognition and extraction of wind-powered electricity generation climbing event can be more rapidly and effectively carried out.This
Outward, according to climbing event complex genesis, with non-linear, hysteresis quality, it is difficult to the characteristics of setting up accurate Mathematical Modeling to characterize,
Concentrated from historical sample using fuzzy clustering and search out the classification of similar case and be trained and set up forecast model.The method
Strong adaptability, can effectively improve the precision of prediction of wind-powered electricity generation.
Brief description of the drawings
The present invention has drawings described below：
A kind of schematic flow sheets of the wind power combination forecasting method based on climbing affair character identification of Fig. 1.
Specific embodiment
Above-mentioned is only the general introduction of technical solution of the present invention, in order to better understand technological means of the invention, below
With reference to accompanying drawing, the present invention is described in further detail with specific embodiment.
As shown in figure 1, a kind of wind power combination forecasting method based on climbing affair character identification of the present invention,
Comprise the following steps：
Step S1. is obtained for the historical wind speed data and historical power data of wind power plant after being processed using wavelet de-noising method
To smooth historical wind speed curve and historical power curve, to eliminate the interference of grass.
Useful signal is usually expressed as low frequency signal or relatively steady, and noise signal is usually expressed as high frequency letter
Number.After being decomposed to noisy primary signal using wavelet de-noising method, noisy part is concentrated mainly in high-frequency signal.First to going through
The time series of history air speed data and historical power data carries out Fourier's treatment, and discrete signal is switched into continuous signal.And
Power spectrumanalysis is carried out to continuous signal, its spectrum waveform is drawn, the spectrum waveform feature of continuous signal is analyzed, and then to waveform
It is fitted, parametric equation is obtained after fitting, solution draws Power Spectrum Distribution result.Contravariant is carried out after the parameter for changing equation
Change, draw filtered result.
Step S2. uses compression algorithm, to being processed by step S1 after resulting smooth historical wind speed curve and
Historical power curve carries out the feature recognition of climbing event, then will identify that the climbing event come carries out characteristics extraction.
Using revolving door (Swinging door trending, the SDT) algorithm in data compression algorithm, SDT algorithms are
A kind of trends of straight line compression algorithm, it is that it is possible to find out straight line as most long as possible and becomes compared with the advantage of other compression algorithms
Gesture carries out data compression, eliminates influence of the noise to data so that the identification of climbing event is more complete, so that as far as possible
Avoid the missing of feature recognition and cross identification situation and occur.
Climbing event definition based on FFP algorithms is：If P is wind-powered electricity generation historical power time series,It is by after denoising
The smooth historical power curve of wind-powered electricity generation,It is the wind-powered electricity generation historical power smooth sequence after being changed through SDT algorithms, the history
Power smooth sequence can be used to recognize climbing event, is calculated by formula (1)：
Wherein SDT is the SQL with compression algorithm principle；Δ E is responsive parameter, and value is smaller, be can recognize that
The climbing event procedure duration it is shorter, climbing height it is lower；Value is bigger, the recognizable climbing event procedure duration
More long, climbing height is higher.
When the conversion signal of wind-powered electricity generation historical power time seriesAbsolute value greater-than match climb event criterion
During threshold parameter λ, i.e.,
WhenDuring for negative value, what is identified is lower climbing event, whenBe on the occasion of when, what is identified is upper climbing thing
Part.This definition needs given two parameter lambdas and Δ E.λ is the threshold value for determining whether to occur climbing event, the threshold of upper climbing event
It is 20.0%cap to be worth, and the threshold value of lower climbing event is 15.0%cap.Additionally, will identify that the climbing event come carries out characteristic value
Extract, characteristic value includes：Climbing rate Rr, climbing amplitude RA, time started RTsWith duration RTl。
The characteristic value of the climbing event that step S3. is extracted step S2 using the method for fuzzy clustering is classified,
Optimum cluster model is made, different climbing event types are obtained.
Because the factor that influence climbing event occurs is intricate, the physical mechanism for being formed to it so far is not done also completely
It is clear.Therefore, it is a basic thought for the system with typical blur based on wind-powered electricity generation climbing event, by genetic analysis, system
Meter analysis method is combined with Fuzzy Analysis, it is established that the fuzzy relation between climbing event and corresponding climbing characteristic value,
Carry out cluster analysis.
Assuming that each climbing event has m characteristic value, and n sample constitutes eigenmatrix：
Because dimension is inconsistent, first the eigenmatrix in S31 is normalized：
The fuzzy membership of each characteristic value is calculated using fuzzy clustering algorithm：
Wherein,It is i-th data to the j degree of membership of cluster centre；I=1,2 .., m, j=1,2 .., c, c are number
According to hard plot clusters number, 4 are taken herein；vijAnd σijCenter and the width of Gaussian function are represented respectively.It is nearest by fuzzy membership
Principle can make optimum cluster model, obtain different climbing event types.
Step S4. is trained and sets up forecast model respectively using statistical method to different climbing event types.
Different climbing event types is trained using ExtremeLearningMachine (ELM) algorithm, regulation ELM algorithm parameters make
Various climbing events set up an optimal Clustering Model,
Need adjustment input weights and hidden layer biasing different from conventional function approximation theory, in ExtremeLearningMachine algorithm only
Wanting activation primitive can infinitely lead, and input weights and hidden layer biasing can be randomly assigned, once the value being randomly assigned is being calculated
Calligraphy learning starts to be updated in algorithm parameter, and hidden layer output matrix H is just uniquely determined and keeps constant, instruction
Practice a neutral net to be equivalent to find a least square solution for linear system, if the number of hidden layer node and input
Sample number is equal, then this Single hidden layer feedforward neural networks (SLFN) just can approach these training examples with zero error.
The ExtremeLearningMachine algorithm is concretely comprised the following steps：
1) input weights and hidden layer biasing are randomly assigned；
For N number of different training sample { xj,yj, xjIt is n × 1 dimensional input vector, yjIt is m × 1 dimension output vector, tool
There is D hidden node and can infinitely lead activation primitivegX the SLFN structures of () can approach this N number of sample with zero error, that is, exist
βi,ωiAnd bi, make ∑ | | yj-rj| |=0, can also be written as form：
In formula：ωiIt is n × 1 dimensional vector, represents the connection weight of input layer and the neuron of hidden layer；βiFor m × 1 tie up to
Amount, represents the connection weight of hidden layer and output layer neuron；biIt is the biasing of hidden layer neuron；rjIt is the output of network；g
X () is activation primitive, can be " sig ", " rbf ", the diversified forms such as " sin ".
2) hidden layer output matrix H is calculated；
This N number of equation can be write as matrix form：
H β=R (7)
Wherein β is output layer weight matrix, and R is desired output；
3) global optimum's output weights of ELM algorithms are calculated；
H+It is the generalized inverse matrix of H.
Step S5. carries out the mistake of step S1 to S2 to the real-time wind speed test data and realtime power test data of wind power plant
Journey, extracts the characteristic value of the climbing event of real-time wind speed test data and realtime power test data.
According to step S1 to S2 real-time wind speed test data and realtime power test data are carried out same treatment and point
Analysis, extracts the characteristic value of climbing event：Climbing rate Rr, climbing amplitude RA, time started RTsWith duration RTl。
Real-time wind speed test data and real-time work(that step S6. is extracted using the forecast model that step S4 is obtained to step S5
The characteristic value of the climbing event of rate test data is classified, and is predicted using forecast model, finally obtains final group
Conjunction predicts the outcome.
Using same criteria for classification, the real-time wind speed test that the forecast model obtained according to step S4 is extracted to step S5
The characteristic value of the climbing event of data and realtime power test data is classified, respectively with the ExtremeLearningMachine for having trained
Algorithm is tested with four kinds of corresponding forecast models.
The above, is only preferred embodiments of the invention, and any formal limitation, ability are not made to the present invention
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 domain in.
The content not being described in detail in this specification belongs to prior art known to professional and technical personnel in the field.
Claims (9)
1. it is a kind of based on climbing affair character identification wind power combination forecasting method, it is characterised in that：Comprise the following steps：
S1, the historical wind speed and historical power data of wind power plant are processed using wavelet de-noising method, respectively obtained smooth
Historical wind speed curve and historical power curve, to eliminate the interference of grass；
S2, using compression algorithm, the smooth historical wind speed curve and historical power curve obtained to step S1 carry out climbing thing
The feature recognition of part, then will identify that the climbing event come carries out characteristics extraction；
S3, the characteristic value of the climbing event for being extracted step S2 using the method for fuzzy clustering are classified, and are made optimal
Clustering Model, obtains different climbing event types；
S4, the different climbing event type obtained to step S3 using statistical method are trained and set up prediction mould respectively
Type；
S5, real-time wind speed test data and realtime power test data to wind power plant carry out the process of step S1 to S2, extract
The characteristic value of the climbing event of real-time wind speed test data and realtime power test data；
S6, the forecast model obtained using step S4 are to the real-time wind speed test data and realtime power test data that extract
The characteristic value of climbing event classified, and be predicted using forecast model, finally obtain final combined prediction result.
It is 2. as claimed in claim 1 to be based on the wind power combination forecasting method that climbing affair character is recognized, it is characterised in that：
The process of wavelet de-noising is described in step S1：
S11, the time series to historical wind speed data and historical power data carry out Fourier's treatment, and discrete signal is switched to
Continuous signal；
S12, power spectrumanalysis is carried out to continuous signal, draw its spectrum waveform；
The spectrum waveform feature of S13, analysis continuous signal, and then waveform is fitted, parametric equation is obtained after fitting, solve
Draw Power Spectrum Distribution result；
Inverse transformation is carried out after S14, the parameter of change equation, filtered result is drawn.
It is 3. as claimed in claim 1 to be based on the wind power combination forecasting method that climbing affair character is recognized, it is characterised in that：
Compression algorithm described in step S2 is revolving door algorithm, and the revolving door algorithm is a kind of trends of straight line compression algorithm.
It is 4. as claimed in claim 1 to be based on the wind power combination forecasting method that climbing affair character is recognized, it is characterised in that：
Based on Fast Field algorithm, the climbing event definition based on Fast Field algorithm is event of climbing described in step S2：If P is wind
Electric historical power time series,Be by the smooth historical power curve of the wind-powered electricity generation after denoising,It is to turn through SDT algorithms
Wind-powered electricity generation historical power smooth sequence after changing, the historical power smooth sequence is used for recognizing climbing event, is calculated by formula (1)
Obtain：
Wherein SDT is the SQL with compression algorithm principle；Δ E is responsive parameter, and value is smaller, and recognizable climbs
The slope event procedure duration is shorter, and climbing height is lower；Value is bigger, and the recognizable climbing event procedure duration gets over
Long, climbing height is higher；
When the conversion signal of wind-powered electricity generation historical power time seriesAbsolute value greater-than match climb the threshold value of event criterion
During parameter lambda, whenDuring for negative value, what is identified is lower climbing event, whenBe on the occasion of when, what is identified is upper climbing thing
Part.
It is 5. as claimed in claim 4 to be based on the wind power combination forecasting method that climbing affair character is recognized, it is characterised in that：
The threshold value of the upper climbing event is 20.0%cap, and the threshold value of lower climbing event is 15.0%cap.
It is 6. as claimed in claim 1 to be based on the wind power combination forecasting method that climbing affair character is recognized, it is characterised in that：
Characteristic value includes described in step S2：Climbing rate Rr, climbing amplitude RA, time started RTsWith duration RTl。
It is 7. as claimed in claim 6 to be based on the wind power combination forecasting method that climbing affair character is recognized, it is characterised in that：
Using the method for fuzzy clustering by comprising the following steps that characteristic value is classified in step S3：
S31, assume that each climbing event has a m characteristic value, and n sample composition eigenmatrix：
S32, the eigenmatrix in S31 is normalized：
S33, the fuzzy membership that each characteristic value is calculated using the method for fuzzy clustering：
Wherein,It is i-th data to the j degree of membership of cluster centre；I=1,2 .., m, j=1,2 .., c, c are that data are hard
Partition clustering number；vijAnd σijCenter and the width of Gaussian function are represented respectively；
S34, optimum cluster model is made by the nearest principle of fuzzy membership, obtain different climbing event types.
It is 8. as claimed in claim 1 to be based on the wind power combination forecasting method that climbing affair character is recognized, it is characterised in that：
Different climbing event types is trained respectively using ExtremeLearningMachine algorithm in step S4, ExtremeLearningMachine algorithm is adjusted
Parameter makes various climbing events set up an optimal Clustering Model.
It is 9. as claimed in claim 8 to be based on the wind power combination forecasting method that climbing affair character is recognized, it is characterised in that：
The ExtremeLearningMachine algorithm is concretely comprised the following steps：
1) input weights and hidden layer biasing are randomly assigned；
For N number of different training sample { xj,yj, xjIt is n × 1 dimensional input vector, yjIt is m × 1 dimension output vector, with D
Hidden node and can infinitely lead the Single hidden layer feedforward neural networks structure of activation primitive g (x) and this N number of sample can be approached with zero error
, that is, there is β in thisi,ωiAnd bi, make ∑ | | yj-rj| |=0, can also be written as form：
In formula：ωiIt is n × 1 dimensional vector, represents the connection weight of input layer and the neuron of hidden layer；βiIt is m × 1 dimensional vector,
Represent the connection weight of hidden layer and output layer neuron；biIt is the biasing of hidden layer neuron；rjIt is the output of network；g(x)
It is activation primitive；
2) hidden layer output matrix H is calculated；
Write N number of equation as matrix form：
H β=R (7)
β is output layer weight matrix, and R is desired output；
3) global optimum's output weights of ELM algorithms are calculated；
H+It is the generalized inverse matrix of H.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710052723.XA CN106933778A (en) | 2017-01-22 | 2017-01-22 | A kind of wind power combination forecasting method based on climbing affair character identification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710052723.XA CN106933778A (en) | 2017-01-22 | 2017-01-22 | A kind of wind power combination forecasting method based on climbing affair character identification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106933778A true CN106933778A (en) | 2017-07-07 |
Family
ID=59423706
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710052723.XA Pending CN106933778A (en) | 2017-01-22 | 2017-01-22 | A kind of wind power combination forecasting method based on climbing affair character identification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106933778A (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107947200A (en) * | 2017-12-04 | 2018-04-20 | 长沙理工大学 | Method for determining frequency deviation coefficient of system containing wind power |
| CN108154260A (en) * | 2017-12-15 | 2018-06-12 | 南京信息工程大学 | A kind of short-term wind power forecast method |
| CN109033027A (en) * | 2018-08-08 | 2018-12-18 | 长春工程学院 | A kind of high speed fitful wind leads to the prediction technique of climbing event under wind power |
| CN109725627A (en) * | 2018-12-06 | 2019-05-07 | 南京信息工程大学 | A kind of wind power climbing event detection system and detection method |
| CN109978222A (en) * | 2019-01-04 | 2019-07-05 | 国网山东省电力公司电力科学研究院 | A kind of wind-powered electricity generation climbing probability of happening prediction technique and system based on Bayesian network |
| CN110046756A (en) * | 2019-04-08 | 2019-07-23 | 东南大学 | Short-time weather forecasting method based on Wavelet Denoising Method and Catboost |
| CN110245866A (en) * | 2019-06-14 | 2019-09-17 | 南京信息工程大学 | A kind of wind power climbing event detecting method |
| CN110348637A (en) * | 2019-07-12 | 2019-10-18 | 哈尔滨工业大学 | A kind of consideration field-net factor wind-powered electricity generation climbing event method for early warning |
| CN110766215A (en) * | 2019-10-17 | 2020-02-07 | 东北电力大学 | Wind power climbing event prediction method based on feature adaptive selection and WDNN |
| CN111414354A (en) * | 2020-03-12 | 2020-07-14 | 广西电网有限责任公司 | Wind power plant data processing method and device and computer readable storage medium |
| CN112070160A (en) * | 2020-09-08 | 2020-12-11 | 合肥工业大学 | Multilayer collaborative real-time classification early warning method for wind power climbing event |
| CN112069721A (en) * | 2019-06-11 | 2020-12-11 | 上海电机学院 | Wind power climbing event prediction method |
| CN112270439A (en) * | 2020-10-28 | 2021-01-26 | 国能日新科技股份有限公司 | Ultra-short-term wind power prediction method and device, electronic equipment and storage medium |
| CN112529275A (en) * | 2020-12-02 | 2021-03-19 | 中国矿业大学 | Wind power climbing event prediction method based on feature extraction and deep learning |
| CN112766583A (en) * | 2021-01-25 | 2021-05-07 | 湖北省气象服务中心（湖北省专业气象服务台） | Wind power short-term prediction method and system based on climbing event feature recognition |
| CN113297805A (en) * | 2021-06-30 | 2021-08-24 | 南京信息工程大学 | Wind power climbing event indirect prediction method |
| CN113486593A (en) * | 2021-07-22 | 2021-10-08 | 南京工程学院 | Wind-storage-hydrogen hybrid energy storage control method based on wind power climbing prediction |
| CN114036465A (en) * | 2021-11-26 | 2022-02-11 | 辽宁工程技术大学 | Air door opening and closing disturbance identification method based on wind speed monitoring |
| CN117293825A (en) * | 2023-11-24 | 2023-12-26 | 天津电力工程监理有限公司 | Wind speed and power hybrid prediction method and device considering wind power climbing |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103679282A (en) * | 2013-09-30 | 2014-03-26 | 清华大学 | Prediction method for wind power ramp |
| CN103955779A (en) * | 2014-05-23 | 2014-07-30 | 武汉大学 | Wind power climbing event probability scene prediction method |
| CN105654207A (en) * | 2016-01-07 | 2016-06-08 | 国网辽宁省电力有限公司锦州供电公司 | Wind power prediction method based on wind speed information and wind direction information |
| CN105719029A (en) * | 2016-03-11 | 2016-06-29 | 中国农业大学 | Combined wind power prediction method based on wind speed fluctuation characteristic extraction |
| CN106203693A (en) * | 2016-07-05 | 2016-12-07 | 华北电力大学 | A kind of system and method for Power Output for Wind Power Field climbing event prediction |
-
2017
- 2017-01-22 CN CN201710052723.XA patent/CN106933778A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103679282A (en) * | 2013-09-30 | 2014-03-26 | 清华大学 | Prediction method for wind power ramp |
| CN103955779A (en) * | 2014-05-23 | 2014-07-30 | 武汉大学 | Wind power climbing event probability scene prediction method |
| CN105654207A (en) * | 2016-01-07 | 2016-06-08 | 国网辽宁省电力有限公司锦州供电公司 | Wind power prediction method based on wind speed information and wind direction information |
| CN105719029A (en) * | 2016-03-11 | 2016-06-29 | 中国农业大学 | Combined wind power prediction method based on wind speed fluctuation characteristic extraction |
| CN106203693A (en) * | 2016-07-05 | 2016-12-07 | 华北电力大学 | A kind of system and method for Power Output for Wind Power Field climbing event prediction |
Non-Patent Citations (5)
| Title |
|---|
| SUJAY CHOUBEY ETC.: "wind power ramp events classification using extreme learning machines", 《PROCEEDINGS IN ADAPTATION，LEARNING AND OPTIMIZATION》 * |
| 刘红柳等: "基于小波变换和支持向量机的风电功率爬坡事件识别与预测", 《东北电力大学学报》 * |
| 孔韦韦等: "《图像融合技术—基于多分辨率非采样理论与方法》", 31 July 2015 * |
| 苏新宁等: "《数据挖掘理论与技术》", 30 June 2003 * |
| 赵志远: "风电功率爬坡事件的识别方法研究", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107947200A (en) * | 2017-12-04 | 2018-04-20 | 长沙理工大学 | Method for determining frequency deviation coefficient of system containing wind power |
| CN108154260A (en) * | 2017-12-15 | 2018-06-12 | 南京信息工程大学 | A kind of short-term wind power forecast method |
| CN109033027A (en) * | 2018-08-08 | 2018-12-18 | 长春工程学院 | A kind of high speed fitful wind leads to the prediction technique of climbing event under wind power |
| CN109033027B (en) * | 2018-08-08 | 2022-07-29 | 长春工程学院 | Method for predicting wind power down-grade climbing event caused by high-speed gust |
| CN109725627A (en) * | 2018-12-06 | 2019-05-07 | 南京信息工程大学 | A kind of wind power climbing event detection system and detection method |
| CN109978222A (en) * | 2019-01-04 | 2019-07-05 | 国网山东省电力公司电力科学研究院 | A kind of wind-powered electricity generation climbing probability of happening prediction technique and system based on Bayesian network |
| CN109978222B (en) * | 2019-01-04 | 2021-04-27 | 国网山东省电力公司电力科学研究院 | Bayesian network-based wind power climbing event probability prediction method and system |
| CN110046756A (en) * | 2019-04-08 | 2019-07-23 | 东南大学 | Short-time weather forecasting method based on Wavelet Denoising Method and Catboost |
| CN112069721A (en) * | 2019-06-11 | 2020-12-11 | 上海电机学院 | Wind power climbing event prediction method |
| CN112069721B (en) * | 2019-06-11 | 2024-02-27 | 上海电机学院 | Wind power climbing event prediction method |
| CN110245866A (en) * | 2019-06-14 | 2019-09-17 | 南京信息工程大学 | A kind of wind power climbing event detecting method |
| CN110348637A (en) * | 2019-07-12 | 2019-10-18 | 哈尔滨工业大学 | A kind of consideration field-net factor wind-powered electricity generation climbing event method for early warning |
| CN110348637B (en) * | 2019-07-12 | 2022-10-11 | 哈尔滨工业大学 | Wind power climbing event early warning method considering field-network factors |
| CN110766215A (en) * | 2019-10-17 | 2020-02-07 | 东北电力大学 | Wind power climbing event prediction method based on feature adaptive selection and WDNN |
| CN110766215B (en) * | 2019-10-17 | 2022-04-01 | 东北电力大学 | Wind power climbing event prediction method based on feature adaptive selection and WDNN |
| CN111414354A (en) * | 2020-03-12 | 2020-07-14 | 广西电网有限责任公司 | Wind power plant data processing method and device and computer readable storage medium |
| CN112070160A (en) * | 2020-09-08 | 2020-12-11 | 合肥工业大学 | Multilayer collaborative real-time classification early warning method for wind power climbing event |
| CN112270439A (en) * | 2020-10-28 | 2021-01-26 | 国能日新科技股份有限公司 | Ultra-short-term wind power prediction method and device, electronic equipment and storage medium |
| CN112270439B (en) * | 2020-10-28 | 2024-03-08 | 国能日新科技股份有限公司 | Ultra-short-term wind power prediction method and device, electronic equipment and storage medium |
| CN112529275B (en) * | 2020-12-02 | 2023-09-22 | 中国矿业大学 | Wind power climbing event prediction method based on feature extraction and deep learning |
| CN112529275A (en) * | 2020-12-02 | 2021-03-19 | 中国矿业大学 | Wind power climbing event prediction method based on feature extraction and deep learning |
| CN112766583A (en) * | 2021-01-25 | 2021-05-07 | 湖北省气象服务中心（湖北省专业气象服务台） | Wind power short-term prediction method and system based on climbing event feature recognition |
| CN112766583B (en) * | 2021-01-25 | 2024-02-02 | 湖北省气象服务中心（湖北省专业气象服务台） | Wind power short-term prediction method and system based on climbing event feature recognition |
| CN113297805A (en) * | 2021-06-30 | 2021-08-24 | 南京信息工程大学 | Wind power climbing event indirect prediction method |
| CN113486593B (en) * | 2021-07-22 | 2023-09-22 | 南京工程学院 | Wind-storage-hydrogen hybrid energy storage control method based on wind power climbing prediction |
| CN113486593A (en) * | 2021-07-22 | 2021-10-08 | 南京工程学院 | Wind-storage-hydrogen hybrid energy storage control method based on wind power climbing prediction |
| CN114036465A (en) * | 2021-11-26 | 2022-02-11 | 辽宁工程技术大学 | Air door opening and closing disturbance identification method based on wind speed monitoring |
| CN114036465B (en) * | 2021-11-26 | 2024-04-12 | 辽宁工程技术大学 | Wind door opening and closing disturbance identification method based on wind speed monitoring |
| CN117293825A (en) * | 2023-11-24 | 2023-12-26 | 天津电力工程监理有限公司 | Wind speed and power hybrid prediction method and device considering wind power climbing |
| CN117293825B (en) * | 2023-11-24 | 2024-02-13 | 天津电力工程监理有限公司 | Wind speed and power hybrid prediction method and device considering wind power climbing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106933778A (en) | A kind of wind power combination forecasting method based on climbing affair character identification | |
| Petković et al. | Wind farm efficiency by adaptive neuro-fuzzy strategy | |
| CN108832619A (en) | Transient stability evaluation in power system method based on convolutional neural networks | |
| CN105719029B (en) | A kind of wind power combination forecasting method based on fluctuations in wind speed feature extraction | |
| CN110994604B (en) | Power system transient stability assessment method based on LSTM-DNN model | |
| CN106447086B (en) | One kind being based on the pretreated wind power combination forecasting method of wind farm data | |
| CN104037776B (en) | The electric network reactive-load capacity collocation method of random inertial factor particle swarm optimization algorithm | |
| CN104036073B (en) | Double-fed wind power plant dynamic equivalence modeling method suitable for active power characteristic analysis | |
| CN108549962B (en) | Wind power prediction method based on historical segmented sequence search and time sequence sparsification | |
| CN104933483A (en) | Wind power forecasting method dividing based on weather process | |
| CN105069236B (en) | Consider the broad sense load joint probability modeling method of wind power plant node space correlation | |
| CN108229732A (en) | ExtremeLearningMachine wind speed ultra-short term prediction method based on error correction | |
| CN106803135A (en) | The Forecasting Methodology and device of a kind of photovoltaic power generation system output power | |
| CN103336998B (en) | A kind of wind energy turbine set fluctuation of power stabilizes the optimized calculation method of target value | |
| CN109242136A (en) | A kind of micro-capacitance sensor wind power Chaos-Genetic-BP neural network prediction technique | |
| CN111784543A (en) | System model for predicting short-term photovoltaic power generation system power generation based on hidden Markov model | |
| CN106611243A (en) | Residual correction method for wind speed prediction based on GARCH (Generalized ARCH) model | |
| CN110148935A (en) | A kind of power grid static security nargin appraisal procedure based on shot and long term Memory Neural Networks | |
| CN106712065A (en) | Prediction method of indexes of influence of electric card charging on power distribution network active loss | |
| CN111652478B (en) | Umbrella algorithm-based power system voltage stability evaluation misclassification constraint method | |
| CN104915788B (en) | A method of considering the Electrical Power System Dynamic economic load dispatching of windy field correlation | |
| CN113486593B (en) | Wind-storage-hydrogen hybrid energy storage control method based on wind power climbing prediction | |
| CN108694475B (en) | Short-time-scale photovoltaic cell power generation capacity prediction method based on hybrid model | |
| Almasoudi et al. | Nonlinear coordination strategy between renewable energy sources and fuel cells for frequency regulation of hybrid power systems | |
| CN111160386B (en) | Data-driven power island detection method based on active reactive power disturbance |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Ye Lin Inventor after: Teng Jingzhu Inventor after: Zhang Cihang Inventor after: Chen Chaoyu Inventor before: Ye Lin Inventor before: Teng Jingzhu Inventor before: Zhang Cihang Inventor before: Chen Chaoyu |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170707 |