CN107507097A - A kind of short-term wind power prediction method - Google Patents
A kind of short-term wind power prediction method Download PDFInfo
- Publication number
- CN107507097A CN107507097A CN201710532750.7A CN201710532750A CN107507097A CN 107507097 A CN107507097 A CN 107507097A CN 201710532750 A CN201710532750 A CN 201710532750A CN 107507097 A CN107507097 A CN 107507097A
- Authority
- CN
- China
- Prior art keywords
- wind speed
- prediction
- model
- forecasting
- wind
- 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
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000000354 decomposition reaction Methods 0.000 claims description 39
- 230000006870 function Effects 0.000 claims description 39
- 238000004422 calculation algorithm Methods 0.000 claims description 22
- 241000255581 Drosophila <fruit fly, genus> Species 0.000 claims description 6
- 230000001932 seasonal effect Effects 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 230000005611 electricity Effects 0.000 description 8
- 238000005457 optimization Methods 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- 230000000052 comparative effect Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000011160 research Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 230000006872 improvement Effects 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 3
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 239000002245 particle Substances 0.000 description 3
- 238000004445 quantitative analysis Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 108700010295 Drosophila for Proteins 0.000 description 1
- 102000008297 Nuclear Matrix-Associated Proteins Human genes 0.000 description 1
- 108010035916 Nuclear Matrix-Associated Proteins Proteins 0.000 description 1
- 241000255588 Tephritidae Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003556 assay Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000005684 electric field Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 230000019637 foraging behavior Effects 0.000 description 1
- 238000013277 forecasting method Methods 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012843 least square support vector machine Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 210000000299 nuclear matrix Anatomy 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
-
- 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"
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A30/00—Adapting or protecting infrastructure or their operation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of short-term wind power prediction method, this method comprises the following steps:(1) training dataset is obtained, including historical wind speed time series and historical weather data, model training is carried out using historical wind speed time series and obtains forecasting wind speed model, while carries out model training using historical wind speed time series and historical weather data and obtains wind power prediction model;(2) predictive data set, historical wind speed time series of the predictive data set including the setting time section before the pending wind power prediction time limit and the data of weather forecast in the prediction time limit are obtained;(3) the historical wind speed time series that prediction data is concentrated is inputted to forecasting wind speed model and obtains predicting the prediction of wind speed time series in the time limit;(4) data of weather forecast that prediction of wind speed time series and prediction data are concentrated is inputted to wind power prediction model to the wind power prediction value for obtaining predicting the time limit.Compared with prior art, prediction result of the present invention is more accurate credible.
Description
Technical field
The present invention relates to a kind of wind power forecasting method, more particularly, to a kind of short-term wind power prediction method.
Background technology
Wind power prediction is the basis for assessing wind power plant running status, and its random fluctuation characteristic is brought to electric power netting safe running
Challenge.In order to improve the predictable of output of wind electric field, ensure the reliable and stable operation of power system security, alleviate power system and adjust
Peak, frequency modulation pressure, it is necessary to improve the precision of microgrid short term power prediction.Recent domestic is on the main base of wind power prediction
Historical data, numerical weather prediction (Numerical Weather Prediction, NWP), geographical position in wind power plant and
Weather element, wind speed-wind power conversion characteristic, with reference to forecast models such as physics, statistics and combinations, realize Multiple Time Scales
Wind power prediction.
Ripe NWP systems must be relied on using physical model, it is quantitative analysis processing wind farm wind velocity, wind direction, temperature, big
The history meteorological data such as air humidity degree and air pressure, carry it into power and turn curve and obtain actual power, this method has NWP renewal speed
Limitation, be generally used for maintenance of fan or debugging, be generally used for the applying of wind power.Statistical regression and study are managed
By the mapping relations being based between historical statistical data, real-time monitor value and power output, strong with generalization ability, Wuxi is examined
The tool on worry blower fan periphery a little, is widely used in the short-term forecast of wind power as characteristic etc..But statistical regression method excessively relies on
Historical data, it is not suitable for the situation of small data sample, and faces weather environment complicated and changeable, the scope of application of model needs
Improve.Thought of the Statistical learning-based approaches based on machine learning, compensate for the deficiency of statistic law, improve the flexibility of model.At present
The emphasis of research is wind power to wind speed, the wave motion response of wind direction, the research for wind speed for Future Development emphasis.
The content of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide a kind of wind power is short-term
Forecasting Methodology.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of short-term wind power prediction method, this method comprise the following steps:
(1) training dataset is obtained, described training dataset includes historical wind speed time series and historical weather data,
Model training is carried out using historical wind speed time series and obtains forecasting wind speed model, while using historical wind speed time series and is gone through
History weather data carries out model training and obtains wind power prediction model;
(2) predictive data set is obtained, described predictive data set includes the setting before the pending wind power prediction time limit
The historical wind speed time series and the data of weather forecast in the prediction time limit of period;
(3) the historical wind speed time series that prediction data is concentrated is inputted to forecasting wind speed model and obtains predicting in the time limit
Prediction of wind speed time series;
(4) data of weather forecast that prediction of wind speed time series and prediction data are concentrated is inputted to wind power prediction mould
Type obtains predicting the wind power prediction value in time limit.
Establishing forecasting wind speed model in step (1) is specially:
(11) decomposed using the historical wind speed time series for being concentrated training data based on set empirical mode decomposition method
For the n wind speed subsequence that frequency domain is stable;
(12) n wind speed subsequence is subjected to phase space reconfiguration and obtains n wind speed reconstruct submatrix;
(13) n forecasting wind speed model is established;
(14) n wind speed reconstruct submatrix is inputted to forecasting wind speed model, Mei Gefeng as a training sample
Fast forecast model one forecasting wind speed subsequence of corresponding output;
(15) optimizing is carried out to Parameters for Phase Space Reconstruction and forecasting wind speed model parameter to solve to obtain optimal phase space reconfiguration
Parameter and forecasting wind speed model parameter.
Described forecasting wind speed model is least square method supporting vector machine forecasting wind speed model.
Step (3) is specially:
(31) decomposed using the historical wind speed time series for being concentrated prediction data based on set empirical mode decomposition method
For the n forecasting wind speed subsequence that frequency domain is stable;
(32) phase space reconfiguration is carried out by the optimal Parameters for Phase Space Reconstruction of determination respectively to n forecasting wind speed subsequence
Obtain n wind speed reconstruct prediction submatrix;
(33) n wind speed reconstruct prediction submatrix as prediction input and is inputted to determining optimal wind speed forecast model
The least square method supporting vector machine forecasting wind speed model of parameter obtains n forecasting wind speed subsequence;
(34) n forecasting wind speed subsequence is overlapped to obtain prediction of wind speed time series.
It is described to be specially based on set empirical mode decomposition method:
(a) historical wind speed time series to be decomposed is added to the white noise sequence of Normal Distribution, forms new wind
Fast target sequence;
(b) empirical mode decomposition is carried out to wind speed target sequence, obtains n-1 intrinsic mode function component Ci(t) it is and surplus
1 remaining residual component rn(t), i=1,2 ... n-1;
(c) renewal white noise sequence repeats step (a)~(b) and decomposes to obtain p until carrying out p empirical mode decomposition
Group intrinsic mode function component and residual component;
(d) average and the n wind speed stable as frequency domain are correspondingly asked for p group intrinsic mode function components and residual component
Subsequence.
Empirical mode decomposition is specially in step (b):
(b1) wind speed target sequence is denoted as X (t), defines the adequate condition of intrinsic mode function component, number i is decomposed in order
=1;
(b2) iterations k=1 is made;
(b3) wind speed target sequence X (t) is fitted to obtain the upper of wind speed target sequence using 3 batten difference functions
Envelope likAnd lower envelope line l (t)ik(t);
(b4) coenvelope line l is soughtikAnd lower envelope line l (t)ik(t) median is:
mik(t)=[lik(t)+lik(t)]/2;
(b5) h is madeik(t)=X (t)-mik(t), if h i k(t) meet the adequate condition of intrinsic mode function component, then perform
Step (b6), otherwise, another k=k+1, return to step (b3);
(b6) h is judgedik(t) whether monotonicity is met, if then by hik(t) it is used as residual component rn(t) and terminate, otherwise
By hik(t) intrinsic mode function component, i.e. C are used asi(t)=hik(t) step (b7), is performed;
(b7) X (t)-C is madei(t) as new wind speed target sequence X (t), with seasonal i=i+1, k=1, return to step
(b3)。
The adequate condition of described intrinsic mode function component includes:Intrinsic mode function component zero crossing number and part
Extreme value points at most differ 1, and the intrinsic mode function component average in the range of domain tends to 0.
Parameters for Phase Space Reconstruction includes embedded dimension m and delay time T in step (15).
Wind speed prediction model parameterses include regularization parameter γ and kernel functional parameter σ in step (15).
Described carries out optimizing solution to Parameters for Phase Space Reconstruction and forecasting wind speed model parameter using drosophila optimized algorithm.
Compared with prior art, the invention has the advantages that:
(1) present invention is used when establishing forecasting wind speed model and carrying out actual wind speed prediction based on set empirical modal
Decomposition method (EEMD) realizes multi-resolution decomposition, and the sequence signal of nonlinear and nonstationary can be decomposed into various different scales by it
Intrinsic mode function (IMF) and a residual components, can effectively solve the big problem of fluctuation for predicting waveform, avoid waveform decomposition
During the modal overlap phenomenon that is likely to occur, improve the robustness of model prediction, improve the precision of wind power prediction.
(2) forecasting wind speed model of the present invention uses least square method supporting vector machine model, forecast model precision of prediction height;
(3) sought when carrying out Parameters for Phase Space Reconstruction and forecasting wind speed model parameter solves using drosophila optimized algorithm
Excellent solution, effectively determines adjustable parameter, improves convergence rate, expands hunting zone, improves the forecasting accuracy of model,
Further improve the precision of wind power prediction.
Brief description of the drawings
Fig. 1 is the FB(flow block) of short-term wind power prediction method of the present invention;
Fig. 2 is the forecasting wind speed comparative result figure of EEMD forecasting wind speeds model and EMD forecasting wind speed models;
Fig. 3 is the forecasting wind speed comparative result figure of EEMD forecasting wind speeds model and LS-SVM forecasting wind speed models;
Fig. 4 is the ideal output power matched curve of 1.5MW blower fans;
Fig. 5 is wind speed-wind power matched curve scatter diagram;
Fig. 6 is the wind power prediction Comparative result of EEMD wind power prediction models and EMD wind power prediction models
Figure;
Fig. 7 is that the wind power of EEMD wind power prediction models and the LS-SVM wind power prediction models of the present invention is pre-
Survey comparative result figure.
Embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.
Embodiment
A kind of short-term wind power prediction method, this method comprise the following steps:
(1) training dataset is obtained, training dataset includes historical wind speed time series and historical weather data, using going through
History wind speed time series carries out model training and obtains forecasting wind speed model, while utilizes historical wind speed time series and weather history
Data carry out model training and obtain wind power prediction model;
(2) predictive data set is obtained, predictive data set includes the setting time section before the pending wind power prediction time limit
Historical wind speed time series and prediction the time limit in data of weather forecast;
(3) the historical wind speed time series that prediction data is concentrated is inputted to forecasting wind speed model and obtains predicting in the time limit
Prediction of wind speed time series;
(4) data of weather forecast that prediction of wind speed time series and prediction data are concentrated is inputted to wind power prediction mould
Type obtains predicting the wind power prediction value in time limit.
Establishing forecasting wind speed model in step (1) is specially:
(11) decomposed using the historical wind speed time series for being concentrated training data based on set empirical mode decomposition method
For the n wind speed subsequence that frequency domain is stable;
(12) n wind speed subsequence is subjected to phase space reconfiguration and obtains n wind speed reconstruct submatrix;
(13) n forecasting wind speed model is established;
(14) n wind speed reconstruct submatrix is inputted to forecasting wind speed model, Mei Gefeng as a training sample
Fast forecast model one forecasting wind speed subsequence of corresponding output;
(15) optimizing is carried out to Parameters for Phase Space Reconstruction and forecasting wind speed model parameter to solve to obtain optimal phase space reconfiguration
Parameter and forecasting wind speed model parameter.
Forecasting wind speed model is least square method supporting vector machine forecasting wind speed model.
Step (3) is specially:
(31) decomposed using the historical wind speed time series for being concentrated prediction data based on set empirical mode decomposition method
For the n forecasting wind speed subsequence that frequency domain is stable;
(32) phase space reconfiguration is carried out by the optimal Parameters for Phase Space Reconstruction of determination respectively to n forecasting wind speed subsequence
Obtain n wind speed reconstruct prediction submatrix;
(33) n wind speed reconstruct prediction submatrix as prediction input and is inputted to determining optimal wind speed forecast model
The least square method supporting vector machine forecasting wind speed model of parameter obtains n forecasting wind speed subsequence;
(34) n forecasting wind speed subsequence is overlapped to obtain prediction of wind speed time series.
It is specially based on set empirical mode decomposition method:
(a) historical wind speed time series to be decomposed is added to the white noise sequence of Normal Distribution, forms new wind
Fast target sequence;
(b) empirical mode decomposition is carried out to wind speed target sequence, obtains n-1 intrinsic mode function component Ci(t) it is and surplus
1 remaining residual component rn(t), i=1,2 ... n-1;
(c) renewal white noise sequence repeats step (a)~(b) and decomposes to obtain p until carrying out p empirical mode decomposition
Group intrinsic mode function component and residual component;
(d) average and the n wind speed stable as frequency domain are correspondingly asked for p group intrinsic mode function components and residual component
Subsequence.
Empirical mode decomposition is specially in step (b):
(b1) wind speed target sequence is denoted as X (t), defines the adequate condition of intrinsic mode function component, number i is decomposed in order
=1;
(b2) iterations k=1 is made;
(b3) wind speed target sequence X (t) is fitted to obtain the upper of wind speed target sequence using 3 batten difference functions
Envelope likAnd lower envelope line l (t)ik(t);
(b4) coenvelope line l is soughtikAnd lower envelope line l (t)ik(t) median is:
mik(t)=[lik(t)+lik(t)]/2;
(b5) h is madeik(t)=X (t)-mik(t), if h i k(t) meet the adequate condition of intrinsic mode function component, then perform
Step (b6), otherwise, another k=k+1, return to step (b3);
(b6) h is judgedik(t) whether monotonicity is met, if then by hik(t) it is used as residual component rn(t) and terminate, otherwise
By hik(t) intrinsic mode function component, i.e. C are used asi(t)=hik(t) step (b7), is performed;
(b7) X (t)-C is madei(t) as new wind speed target sequence X (t), with seasonal i=i+1, k=1, return to step
(b3)。
The adequate condition of intrinsic mode function component includes:Intrinsic mode function component zero crossing number and Local Extremum
Number at most differs 1, and the intrinsic mode function component average in the range of domain tends to 0.
Parameters for Phase Space Reconstruction includes embedded dimension m and delay time T in step (15).
Wind speed prediction model parameterses include regularization parameter γ and kernel functional parameter σ in step (15).
Optimizing solution is carried out using drosophila optimized algorithm to Parameters for Phase Space Reconstruction and forecasting wind speed model parameter.
For synthesis, short-term wind power prediction method of the invention as shown in figure 1, this method includes two large divisions, its
Middle A is the forecasting wind speed in the prediction time limit, and B is the wind power prediction in the prediction time limit.
Specifically, A includes:A1:The acquisition of predictive data set, A2:It is pre- using being incited somebody to action based on set empirical mode decomposition method
The historical wind speed time series surveyed in data set is decomposed into n stable forecasting wind speed subsequence IMF of frequency domain1、IMF2、……
IMFn, A3:Phase space reconfiguration, A4 are carried out to n forecasting wind speed subsequence:Data input values after phase space reconfiguration are instructed in advance
The least square method supporting vector machine forecasting wind speed model (LS-SVM models) perfected, A5:It is predicted to obtain n wind speed respectively pre-
Survey subsequence, A6:N forecasting wind speed subsequence is overlapped to obtain prediction of wind speed time series.It is pre- that wind speed is carried out again in addition
Its concrete mode is similar when forecasting wind speed is actually carried out when surveying model (LS-SVM models) training, and difference is that step A1 is replaced
Acquisition training dataset is changed to, and A2~A6 steps are carried out using training dataset, while obtains optimized parameter.
B includes:B1:Obtain the data of weather forecast in the prediction time limit, B2:The prediction of wind speed time series that will be obtained in A
And the data of weather forecast that B1 is obtained is inputted to the good wind power prediction model of training in advance, B3:It is pre- according to wind power
Model is surveyed to be predicted to obtain the wind power prediction value in the prediction time limit.
The present embodiment is modeled using microgrid energy management misconduct platform as research object for wind-powered electricity generation short term power.Should
Platform wind generator system simulates blower fan capacity 1.5MW, sampled data resolution ratio are 5min, there is provided each prediction day 288 is pre-
Survey data point, including the characteristic such as wind speed, wind direction, temperature, atmospheric pressure, relative humidity.Wind-power electricity generation short term power is predicted
Mainly for power prediction of the predicted time within 72 hours (hr), the present invention is limited to not on the research of wind power prediction
Carry out 24hr, model establish substantially process be roughly divided into the meteorological historical data of collection, data analysis and process, screening input because
Son, establish the several processes of forecast model.
1) present invention uses microgrid experiment porch Supervisory control and data acquisition (Supervisory Control And Data
Acquisition, SCADA) historical data work of the blower fan that provides of system when run between 27 days~March 9 January in 2016
For experimental data, building for wind power prediction model is carried out, it includes the meteorological data of NWP offers, and its resolution ratio is
5min。
In order to realize the Accurate Prediction of wind-power electricity generation wind speed and wind power, first have to carry out the collection of history meteorological data.Wind
Speed, wind direction, atmospheric pressure etc. are the key factors for influenceing wind energy, so needing that these meteorological elements are detected and recorded.
In order to obtain accurately and reliably meteorological element, meteorological sensor performance indications must are fulfilled for certain requirement.Specific meteorological element
Monitoring technology index is shown in Table 1.
The meteorological element monitoring technology index of table 1
In actual moving process, some failures inevitably occur for data collecting system, cause some abnormal datas
Generation, so as to have a strong impact on the precision of wind-powered electricity generation forecast model.Therefore, will, it is necessary to be pre-processed to the data of system acquisition
Missing data supplement is complete, abnormal data is rationally replaced, to guarantee the service requirement for meeting forecast model.
2) present invention uses drosophila optimized algorithm (Fruit fly Optimization Algorithm, FOA), and FOA is calculated
Method is a kind of global search learning algorithm, compared to current several ripe optimized learning algorithms, such as genetic algorithm (Genetic
Algorithm, GA), particle cluster algorithm (Particle swarm optimization, PSO), its have adjustable parameter it is less,
The advantages that robustness is preferable, search time is short, pace of learning is fast, but the algorithm development is later, and FOA also has very big development at present
Space.FOA algorithms are able to develop out a kind of algorithm of global optimizing by drosophila foraging behavior.
3) wind speed, wind power signal have the characteristics of unstable, fluctuation is big, are often caused in wind power prediction
Irregular fluctuation, the problem of so as to cause power sequence responding ability deficiency.Therefore, the present invention is based on signal Scale Decomposition dimensionality reduction
Thought, it is proposed that a kind of method based on modified empirical mode decomposition is used for establishing forecast model, so as to effectively handling
Nonlinear sequential wind power signal, improve the precision of power prediction.Empirical mode decomposition (EMD) is a kind of for signal point
The self-adapting data method for digging of analysis.It is by the way that non-linear sequence to be decomposed into the intrinsic mode function components of some different scales
(Intrinsic Mode Function, IMF) and a residual components, to obtain stationary sequence, realize the drop of data dimension
Dimension.Wherein IMF components need to meet 2 conditions:Signal zero-crossing number at most differs 1 with local extremum points;Whole definition
Serial mean in the range of domain tends to 0.
EMD itself yardstick based on signal in decomposable process, remains the property of initial data, is especially suitable for for locating
The larger wind speed waveform of fluctuation is managed, can apply to any kind of time series signal in theory, is decomposing complicated fluctuation
Nonlinear data when be based on 3 conditions:1. at least there are 2 extreme values (maximum and minimum value) in the waveform of sequence;2. data
Local temporal characteristic is determined by the time scale of adjacent extreme point, and is unique;3. if extreme point is not present in signal, but
Flex point be present, then can ask for single order or higher differentiation obtains extreme value, then decomposition result is obtained by integrating.
4) informational influence included for initial data by the process of EMD Decomposition Sequences is very big, so multiple at some
Modal overlap occurs in the IMF sequences for decomposing to obtain under miscellaneous characteristic signal, i.e., includes very big otherness in single IMF samples
Time scale, or different IMF samples have similar characteristic time scale, cause the waveform aliasing of the two sequences, mutually it
Between influence distinguish, it is difficult to recognize.
The monitoring and collection of wind-powered electricity generation data often occur signal interruption, noise and equipment fault and cause abnormal pulsers to be done
Disturb, the IMF components of mistake can be caused during mode decomposition, modal overlap phenomenon easily occur, so as to can not realize compared with
For preferable effect.Therefore the present invention carries out the decomposition of air speed data using set empirical mode decomposition method.
In wind-power electricity generation, blower fan group is influenceed by wind energy, and obtained kinetic energy is converted into electric energy.The wind energy of wind power plant is by wind
The influence of a variety of Meteorological Characteristics factors such as speed, wind direction, temperature, air pressure and humidity, if unprocessed directly by these meteorologic factors
As the input of forecast model, model multiplicity can be caused too high, influence the Generalization Capability of model, reduce the robustness of model.Cause
This, before forecast model is established, it should analyze the property of characteristic, extraction is significant with wind power output correlation
Input the factor.
(1) wind speed:Earth surface is by solar radiation, the pressure gradient-force for causing nonuniform heating to be formed, and its measurement refers to
Mark is wind speed.Forecasting wind speed is significant for improving overall stably operating wind power field and improvement wind power prediction,
For wind power, wind speed, which is undoubtedly, influences maximum Meteorological Characteristics factor.The fluctuation of wind speed and uneven stability research
It is the key of wind power prediction.Wind energy and air velocity it is cube directly proportional, its relational expression is:
W=ρ Av3/2
In formula:W is wind energy;V is wind speed;ρ is atmospheric density;A is the cross-sectional area of gaseous exchange.
(2) wind direction:When carrying out resource situation assay to the environment of wind power plant, wind regime data not only include wind speed, and also
The change of wind direction need to be considered.Wind direction generally comprises 16 orientation:N (north), NNE (north north east), NE (east northeast), ENE (east-north-east),
E (east), ESE (east-south-east), SE (east southeast), SSE (South South-East), S (south), SSW (South South west), SW (Nan Xi), WSW (cc
South), W (west), WNW (cc north), NW (northwest), NNW (cc north).In order to more intuitively portray this change, generally use
Wind rose carries out the statistics of wind energy resources measurement data.In order to record the property with quantitative analysis wind direction, wind direction is converted
For sinusoidal and cosine expression way.
In summary, the main weather factor for influenceing wind power is wind speed, and wind speed is according to geographical position and time
Difference, its wind direction has larger variation.The coefficient correlation of calculation of wind speed and atmospheric pressure and relative air humidity is respectively
0.54 and 0.33 or so, this shows that, when carrying out the regression analysis of wind speed time series, multiple relevant weather factors need to be considered
Feature.
EEMD adds the white noise sequence of Normal Distribution in EMD decomposes clock signal waveform, effectively suppresses
Modal aliasing problem.But two important parameters therein:The amplitude coefficient α and population mean number m of white noise selection, it is right
There is important physical significance in the whole structure of mode decomposition.ByThe relation between two parameter can be obtained, wherein e is
Overall resolution error caused by during addition white noise, it is:
Consider the equilibrium relation between two parameter, it is clear that α values can not be excessive, because excessive white noise amplitude coefficient can be led
There is larger error after causing mode decomposition, the amplitude versus frequency characte of primary signal can be covered when serious, loses the purpose of decomposition;If α
That chooses is too small, although being favorably improved the precision of classification, can cause to be not enough to the Local Extremum for changing primary signal, from
And the purpose that original signal information is obtained by changing signal local time span can not be realized.From operation time cost consideration, m
Value is not the bigger the better, and can increase calculating cost.
Because in e≤0.01, resolution error caused by residual noise can reach ideal effect, sequence reconstruct
In stable state, therefore the present invention can use and EEMD parameters are determined based on improvement FOA optimized algorithms, its fitness letter
Number may be defined as fα,m(e), i.e. representative function f represents that evaluation index is overall resolution error e by parameter alpha and m.
When optimizing selection to two parameter, domain is limited first, if noise amplitude α ∈ [0.1,0.3], population mean
Number m >=100, it is then final that two parameter optimizition value is respectively α=0.18, m=200 by FOA optimizing.
2592 data sample points are training set before regulation, and rear 288 data sample points are test set, will using EEMD
Wind speed time series carries out mode decomposition, the parameter for optimizing to obtain using FOA above in the wind-powered electricity generation sample data of SCADA records
Value:α=0.18, m=200.Finally give 9 intrinsic mode function component IMF1~IMF9With 1 residual components rn(t) amount to
10 waveforms, the IMF components after decomposition are compared with original wind series, and its fluctuating change is more steady, and spectrum signature is also by IMF
Component characterizes successively from high frequency to low frequency.
After EEMD decomposes to obtain more stable subsequence, its stability is strengthened.Because wind velocity signal is not only right
There is sensitiveness in initial condition, and aperiodic motion be present, therefore the interior of wind series can be analyzed using chaology
In attribute, that is, postpone coordinate state space reconstruction method.
Acquisition of time delay τ and embedded dimension the m determination for information after reconstruction attractor is particularly significant, this 2 parameters
Selection the precision of prediction can be directly affected.If dimension is too low, coincidence, which can occur, for information causes attractor to occur certainly
It is intersecting, it is too high, amount of calculation can be increased;If delay time T is too short, correlation can be excessively close between each point coordinates in reconstruction attractor
Collection, then 2 coordinate components x (i+j τ) in phase space vector and x (i+ (j+1) τ) numerically closely, can not be formed compared with
High identification, so as to which independent coordinate components can not be provided;If delay time T is oversize, chaos attractor track is two
Projection in component direction just loses correlation, it is therefore desirable to determines a suitable delay using suitable method
Time τ, so as to reach a kind of balance between the two in independent and correlation.
The present invention determines delay time T and embedded dimension m two parameters using FOA optimized algorithms are improved.Firstly, because
The air speed data sample time resolution ratio of SCADA collections is 5min, so the delay time T of minimum of computation unit is 5min;Its
It is secondary, to ensure two parameter Optimum Matching, here in conjunction with least square method supporting vector machine (Least Square-Support
Vector Machine, LS-SVM) forecast model is built, while consider 2 parameters of LS-SVM models:Regularization parameter γ and
Kernel functional parameter σ, using the predicted value error of each subsequence as final optimization pass target function value.
The SVMs (SVM) of standard is the convex quadratic programming problem of a linear inequality constraint.LS-SVM is SVM
Improvement, be using linear least-squares system replace quadratic programming be used as loss function, with equality constraint substitute standard support to
The inequality constraints of amount machine, effectively increase and calculate the time, reduce amount of calculation, improve generalization ability.Due to forecasting wind speed
Subsequence quantity is more, and amount of calculation is larger, and the present invention establishes forecasting wind speed model using LS-SVM.
LS-SVM principles are:If training sampleWherein xiFor i-th of sample, yi∈ { -1 ,+1 } is supporting vector
Machine exports, and the inequality constraints of SVMs is:
s.t. yi(ωTφ(xi)+b)≥1-ξi,ξi>=0, i=1,2 ..., l,
In formula:ω is the amount of being sent to of hyperplane, and C is punishment parameter, and ξ is slack variable.
Equality constraint is changed into, is:
In formula:γ is weight constant, similar to punishment parameter C, finds optimal hyperlane for balancing, minimizes deviation
Amount;ekFor error vector.
Former problem is converted into using Lagrange Multiplier Methods extreme-value problem is asked to parameter alpha:
To parameter w, b, ek、αkDerivation obtains:
Can list system of linear equations by above formula is:
Wherein, nuclear matrix is:
Finally giving LS-SVM classification expression formula is:
Because the formula is system of linear equations, therefore solving speed is accelerated, improve the generalization ability of model.
On the affected several parameters of precision of prediction in LS-SVM models:Model data is embedded in dimension m, time delay
τ, regularization parameter γ and kernel functional parameter σ, determine that defining kernel function herein is using based on follow-on FOA optimized algorithms
Gaussian kernel function exp (- | | xi-xj||2/(2σ)).The optimization process of model parameter is as follows:
(1) initialization model parameter, wherein:Dimension d is set to 2, and time delay τ is 1 (sampled point 5min), and regularization is joined
Number γ and kernel functional parameter σ is set to be randomized initial value;
(2) modified FOA Optimal Parameters models are established, fitness function is defined and is defined by training data sample mean square deviation
Then;
(3) optimal parameter after extraction optimization, bring prediction day test data into model, finally give prediction result.
The present invention needed when forecasting wind speed model is established to Parameters for Phase Space Reconstruction and forecasting wind speed model parameter
Optimize, the Parameters for Phase Space Reconstruction for specifically solving to obtain using the progress optimizing of drosophila optimized algorithm includes embedded dimension m
It is as shown in table 2 to include regularization parameter γ and kernel functional parameter σ with delay time T and forecasting wind speed model parameter.
The Space Reconstruction parameter of table 2 and forecasting wind speed model parameter table
In order to verify superiority of the modified FOA used in the present invention for parameter optimization, the present invention additionally uses heredity
Algorithm (GA), particle cluster algorithm (PSO) and above-mentioned parameter is separately optimized without improved FOA.
During using modified FOA Optimized model parameters, its iterations is minimum, and only 4 times with regard to that can reach convergence;And without
If improved FOA realizes convergence, iteration is needed 8 times;For GA algorithms in Optimal Parameters, convergence rate is identical with improved FOA, but suitable
Answer angle value maximum;Smaller result can be finally obtained using PSO Optimal Parameters, but reach stable state convergence to need 14 left sides of iteration
It is right, it is clear that in terms of the optimization of model parameter, follow-on FOA whole structures are even more ideal.The FOA kinds group hunting after improvement
Scope (width) is wider, searching route is shorter, is advantageous to the selection of Optimal Parameters, avoids being absorbed in local optimum, when shortening calculating
Between.
After improved FOA algorithm optimizations parameter, the phase space reconfiguration of original wind speed clock signal mode decomposition has been obtained
Parameter and LS-SVM model parameters, the forecasting wind speed model based on LS-SVM is established with reference to these parameter values.Based on set experience
IMF1~IMF3 in subsequence after mode decomposition is larger because frequency characteristic fluctuates, and it predicts that waveform has larger mistake
Difference, and IMF4~IMF9 and residual components rn(t) than shallower, frequency is relatively stable, is advantageous to the fitting of curve, predicts ripple
The recurrence that shape realizes true wind series substantially will can obtain the predicted value of final wind speed after the superposition of forecasting wind speed subsequence.This
Invention is respectively adopted carries out short-term forecast based on EEMD models, EMD models and LS-SVM models to wind speed, and this is proved by contrasting
Invention uses the validity of method.Fig. 2 is the wind speed value pair obtained using EEMD models and EMD models progress forecasting wind speed
Than figure, Fig. 3 is the wind speed value comparison diagram obtained using EEMD models and LS-SVM models progress forecasting wind speed.With reference to Fig. 2
With Fig. 3 it can be seen that the forecasting wind speed model accuracy based on EEMD models is higher, the recurrence of air speed value have effectively achieved.
In order to more intuitively assess the overall estimated performance of this 3 kinds of models, the present invention is respectively by the ripple of 3 kinds of models
Shape error quantization is analyzed, and employs mean absolute error percentage (Mean Absolute Percentage Error, MAPE)
Contrasted with root-mean-square error (Root Mean Square Error, RMSE), it is specific to predict that error assessment index is shown in Table
3。
The forecasting wind speed error assessment index of table 3
Both error criterion expression formulas are:
In formula:Hh(i) it is predicted value, Hy(i) it is measured value, Z is sampled point number.MAPE can assessment models prediction miss
Poor degree, and RMSE can weigh the overall precision of prediction result.
From the forecasting wind speed error index value of table 3 and Fig. 2 and Fig. 3 forecasting wind speed comparison of wave shape:Using based on EEMD and
EMD method is to forecast model its precision of prediction that wind series established after mode decomposition higher than directly using LS-SVM moulds
The result of type, this explanation is for this mode decomposition energy for having the time series unstable, fluctuation is big, considering waveform of wind speed
Enough it is effectively improved the precision of prediction;Using the Comparative result of EEMD models and EMD models it can be found that the model after improving is one
Determine to improve precision in degree, effectively prevent the interference that this phenomenon of modal overlap is decomposed for waveform, demonstrate the present invention
The reasonability of institute's established model.
Wind speed characteristic quantity is the primary variables of wind power transformation model, and the ultra-short term precision of prediction for improving wind speed can have
Effect improves short-term wind power fluctuating change and responds insufficient problem.Consider wind power relevant weather characteristic factor, provided using NWP
Data of weather forecast combination EEMD short-term wind speed forecasting results, consider that blower fan group can obtain power and be:
P=ρ ACpv3/2
In formula:P is wind wheel power output;V is wind speed;ρ is atmospheric density;A is swept area of rotor, i.e., vertical with wind direction
Plane on, wind wheel rotate when generate round projected area;CpFor power coefficient, maximum usage factor C is takenp,max=
16/27。
For single blower fan, A is constant, it is generally recognized that atmospheric density is also maintained at steady state value, while considers blower fan
Peak power output, the incision wind speed v of blower faninWith cut-out wind speed voff, energy management system of micro-grid simulation can be obtained
The ideal output power matched curve of 1.5MW blower fans is as shown in Figure 4.Because during fan operation, actual contribute can not possibly be strict
Desired power curve is obeyed, blower fan is influenceed by extraneous enchancement factor, it may appear that some floatings, wind speed-wind power matched curve
Scatter diagram is as shown in Figure 5.
By wind speed-wind power matched curve linearisation, 4 stages are divided into:
(1) the initial low wind speed stage (v<vin), it is not enough to drive blower fan power output less than incision wind speed;
(2) wind speed ascent stage (v inin≤v≤vN), more than incision wind speed and it is less than rated wind speed vNWhen, less wind
Speed change can produce obvious power output;
(3) high wind speed saturation stage (vN≤v≤voff), when more than rated wind speed but being less than cut-out wind speed, blower fan output work
Rate is that steady state value is rated power PN, and the change of wind speed will not cause the change of power;
(4) blower fan operation-stopping stage (v>voff), during more than cut-out wind speed, for protection blower fan, now blower fan should stop work
Make, power output 0.
Functional relation is between wind speed-wind power can be obtained:
Finally, after wind speed-wind power inversion cuver founding mathematical models, wind power can be tried to achieve with reference to wind speed value
Predicted value.
Fig. 6 is the wind power prediction comparative result figure of EEMD models and EMD models, and Fig. 7 is EEMD models and the present invention
LS-SVM models wind power prediction comparative result figure.
Similarly, in order to more intuitively assess overall estimated performance of this 3 kinds of models to wind power, respectively by 3
The waveform error quantitative analysis of kind model, employs mean absolute error percentage (Mean Absolute Percentage
Error, MAPE) and root-mean-square error (Root Mean Square Error, RMSE) contrasted, it is specific to predict that error is commented
Valency index is shown in Table 4.
The wind power prediction error assessment index of table 4
Can be obtained by table 4, Fig. 6 and Fig. 7, the short-term wind power prediction model accuracy established using EEMD be higher than using EMD and
LS-SVM models, this is due to that wind series signal decomposition is more stable waveform by EEMD models, is advantageous to returning for waveform
Return, and avoid caused modal overlap between the subsequence after decomposing.The big external environment of fluctuations in wind speed is in blower fan
Under, the precision of power prediction can not be effectively improved only for historical wind speed data, the wind-powered electricity generation forecast model that the present invention establishes passes through
Indirect method improves the model of prediction of wind speed, finally improves the precision of wind power prediction.
Claims (10)
1. a kind of short-term wind power prediction method, it is characterised in that this method comprises the following steps:
(1) training dataset is obtained, described training dataset includes historical wind speed time series and historical weather data, uses
Historical wind speed time series carries out model training and obtains forecasting wind speed model, while utilizes historical wind speed time series and history day
Destiny obtains wind power prediction model according to model training is carried out;
(2) predictive data set is obtained, described predictive data set includes the setting time before the pending wind power prediction time limit
The historical wind speed time series and the data of weather forecast in the prediction time limit of section;
(3) the historical wind speed time series that prediction data is concentrated is inputted to forecasting wind speed model and obtains predicting the prediction in the time limit
Wind speed time series;
(4) data of weather forecast that prediction of wind speed time series and prediction data are concentrated is inputted to wind power prediction model and obtained
To the wind power prediction value in prediction time limit.
2. a kind of short-term wind power prediction method according to claim 1, it is characterised in that step establishes wind in (1)
Fast forecast model is specially:
(11) the historical wind speed time series that training data is concentrated is decomposed into frequency using based on set empirical mode decomposition method
N stable wind speed subsequence of domain;
(12) n wind speed subsequence is subjected to phase space reconfiguration and obtains n wind speed reconstruct submatrix;
(13) n forecasting wind speed model is established;
(14) inputted using n wind speed reconstruct submatrix as a training sample pre- to forecasting wind speed model, each wind speed
Survey model one forecasting wind speed subsequence of corresponding output;
(15) optimizing is carried out to Parameters for Phase Space Reconstruction and forecasting wind speed model parameter to solve to obtain optimal Parameters for Phase Space Reconstruction
With forecasting wind speed model parameter.
A kind of 3. short-term wind power prediction method according to claim 2, it is characterised in that described forecasting wind speed mould
Type is least square method supporting vector machine forecasting wind speed model.
4. a kind of short-term wind power prediction method according to claim 3, it is characterised in that step (3) is specially:
(31) the historical wind speed time series that prediction data is concentrated is decomposed into frequency using based on set empirical mode decomposition method
N stable forecasting wind speed subsequence of domain;
(32) phase space reconfiguration is carried out respectively to n forecasting wind speed subsequence by the optimal Parameters for Phase Space Reconstruction of determination to obtain
N wind speed reconstruct prediction submatrix;
(33) n wind speed reconstruct prediction submatrix as prediction input and is inputted to determining optimal wind speed prediction model parameterses
Least square method supporting vector machine forecasting wind speed model obtain n forecasting wind speed subsequence;
(34) n forecasting wind speed subsequence is overlapped to obtain prediction of wind speed time series.
5. a kind of short-term wind power prediction method according to claim 4, it is characterised in that described to be passed through based on set
Testing mode decomposition method is specially:
(a) historical wind speed time series to be decomposed is added to the white noise sequence of Normal Distribution, forms new wind speed mesh
Mark sequence;
(b) empirical mode decomposition is carried out to wind speed target sequence, obtains n-1 intrinsic mode function component CiAnd remaining 1 (t)
Individual residual component rn(t), i=1,2 ... n-1;
(c) renewal white noise sequence repeats step (a)~(b) until carrying out p empirical mode decomposition decomposes to obtain p groups originally
Levy mode function component and residual component;
(d) average and the n wind speed sequence stable as frequency domain are correspondingly asked for p group intrinsic mode function components and residual component
Row.
A kind of 6. short-term wind power prediction method according to claim 5, it is characterised in that Empirical Mode in step (b)
State is decomposed:
(b1) wind speed target sequence is denoted as X (t), defines the adequate condition of intrinsic mode function component, number i=1 is decomposed in order;
(b2) iterations k=1 is made;
(b3) wind speed target sequence X (t) is fitted to obtain the coenvelope of wind speed target sequence using 3 batten difference functions
Line likAnd lower envelope line l (t)ik(t);
(b4) coenvelope line l is soughtikAnd lower envelope line l (t)ik(t) median is:
mik(t)=[lik(t)+lik(t)]/2;
(b5) h is madeik(t)=X (t)-mik(t), if hik(t) meet the adequate condition of intrinsic mode function component, then perform step
(b6), otherwise, another k=k+1, return to step (b3);
(b6) h is judgedik(t) whether monotonicity is met, if then by hik(t) it is used as residual component rn(t) and terminate, otherwise by hik
(t) intrinsic mode function component, i.e. C are used asi(t)=hik(t) step (b7), is performed;
(b7) X (t)-C is madei(t) as new wind speed target sequence X (t), with seasonal i=i+1, k=1, return to step (b3).
A kind of 7. short-term wind power prediction method according to claim 5, it is characterised in that described intrinsic mode letter
The adequate condition of number component includes:Intrinsic mode function component zero crossing number at most differs 1 with local extremum points, definition
Intrinsic mode function component average in the range of domain tends to 0.
A kind of 8. short-term wind power prediction method according to claim 2, it is characterised in that phase space in step (15)
Reconstruction parameter includes embedded dimension m and delay time T.
9. a kind of short-term wind power prediction method according to claim 2, it is characterised in that wind speed is pre- in step (15)
Surveying model parameter includes regularization parameter γ and kernel functional parameter σ.
10. a kind of short-term wind power prediction method according to claim 2, it is characterised in that described to phase space
Reconstruction parameter and forecasting wind speed model parameter carry out optimizing solution using drosophila optimized algorithm.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710532750.7A CN107507097A (en) | 2017-07-03 | 2017-07-03 | A kind of short-term wind power prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710532750.7A CN107507097A (en) | 2017-07-03 | 2017-07-03 | A kind of short-term wind power prediction method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107507097A true CN107507097A (en) | 2017-12-22 |
Family
ID=60678906
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710532750.7A Pending CN107507097A (en) | 2017-07-03 | 2017-07-03 | A kind of short-term wind power prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107507097A (en) |
Cited By (31)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108427854A (en) * | 2018-03-29 | 2018-08-21 | 深圳市智物联网络有限公司 | A kind of data analysing method and its relevant device |
CN108537372A (en) * | 2018-03-27 | 2018-09-14 | 中南大学 | A kind of Yaw control method of wind direction prediction technique and wind power generating set |
CN109063930A (en) * | 2018-08-30 | 2018-12-21 | 上海电力学院 | A kind of dynamic wind power plant general power prediction technique based on clustering |
CN109086916A (en) * | 2018-07-16 | 2018-12-25 | 国家气象中心 | A kind of convection weather nowcasting method and device based on multi-source observation data |
CN109146131A (en) * | 2018-07-11 | 2019-01-04 | 国网河北省电力有限公司 | A kind of wind-power electricity generation prediction technique a few days ago |
CN109886488A (en) * | 2019-02-21 | 2019-06-14 | 南方电网科学研究院有限责任公司 | Distributed wind power plant layered hybrid short-term prediction method considering wind speed time lag |
CN110263971A (en) * | 2019-05-14 | 2019-09-20 | 西安理工大学 | Super short-period wind power combination forecasting method based on support vector machines |
CN110378541A (en) * | 2019-08-08 | 2019-10-25 | 上海交通大学 | A kind of short-term multistep forecasting method of wind power and device |
CN110390435A (en) * | 2019-07-24 | 2019-10-29 | 广东工业大学 | A kind of wind power forecasting method, device, equipment and readable storage medium storing program for executing |
CN110428100A (en) * | 2019-07-31 | 2019-11-08 | 新疆新能集团有限责任公司乌鲁木齐电力建设调试所 | A kind of blower short-term power generation power prediction technique |
CN110457800A (en) * | 2019-07-30 | 2019-11-15 | 天津大学 | Consider the trunnion axis blower wind speed power output translation method of machinery inertial |
CN111160621A (en) * | 2019-12-06 | 2020-05-15 | 江苏方天电力技术有限公司 | Short-term wind power prediction method integrating multi-source information |
CN111222707A (en) * | 2020-01-13 | 2020-06-02 | 湖北工业大学 | Wind speed prediction method based on time series mutation error correction |
CN111486991A (en) * | 2020-05-09 | 2020-08-04 | 国家电网有限公司 | Overheating risk early warning device and early warning method for 10kV high-voltage switch cabinet |
CN111666458A (en) * | 2020-06-22 | 2020-09-15 | 中国船级社质量认证公司 | Fitting method for power curve of wind turbine generator |
CN111862538A (en) * | 2020-08-03 | 2020-10-30 | 中铁二院工程集团有限责任公司 | Large wind early warning method and system for long-span arch bridge construction period |
CN111915070A (en) * | 2020-07-17 | 2020-11-10 | 天津大学 | Prediction method for k-nearest neighbor day-ahead wind speed by adopting atmospheric pressure spatial correlation |
CN112072691A (en) * | 2019-06-10 | 2020-12-11 | 上海电机学院 | Fan control method based on seasonal decomposition and support vector machine wind power prediction |
CN112085272A (en) * | 2020-09-07 | 2020-12-15 | 国网甘肃省电力公司电力科学研究院 | Wind power prediction method |
CN112101616A (en) * | 2020-08-10 | 2020-12-18 | 国网山东省电力公司青岛供电公司 | Wind power short-term prediction method and system based on inherent time scale decomposition |
CN112348255A (en) * | 2020-11-06 | 2021-02-09 | 湖南大学 | Ultra-short-term wind power prediction method based on wavelet time-frequency imaging |
CN112669168A (en) * | 2020-12-15 | 2021-04-16 | 国网辽宁省电力有限公司阜新供电公司 | Short-term wind power prediction method |
CN112712202A (en) * | 2020-12-29 | 2021-04-27 | 广东电网有限责任公司 | Short-term wind power prediction method and device, electronic equipment and storage medium |
CN113447671A (en) * | 2021-07-15 | 2021-09-28 | 中煤科工集团重庆研究院有限公司 | Roadway section wind speed detection method based on high-frequency and low-frequency ultrasonic waves |
CN113554203A (en) * | 2020-04-24 | 2021-10-26 | 国能日新科技股份有限公司 | Wind power prediction method and device based on high-dimensional gridding and LightGBM |
CN113704558A (en) * | 2021-08-25 | 2021-11-26 | 成都信息工程大学 | Automatic meteorological station automatic clustering method and system based on differential sequence |
CN113836756A (en) * | 2021-11-29 | 2021-12-24 | 山东华尚电气有限公司 | Intelligent monitoring method and system for annealing process of three-dimensional wound core transformer |
CN113836801A (en) * | 2021-09-13 | 2021-12-24 | 湖北工业大学 | Prediction method based on CEEMD and improved SSA-LSSVM |
CN114118499A (en) * | 2020-08-25 | 2022-03-01 | 华北电力大学 | Short-term wind power prediction method and device |
CN114462710A (en) * | 2022-02-11 | 2022-05-10 | 华润电力技术研究院有限公司 | Short-term prediction method, device and medium for fan generated power |
CN117039895A (en) * | 2023-10-09 | 2023-11-10 | 西安热工研究院有限公司 | Wind power prediction method and system for energy storage auxiliary black start |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2013002194A1 (en) * | 2011-06-29 | 2013-01-03 | 三菱重工業株式会社 | Maintenance device for wind power generator and maintenance method for wind power generator |
CN103425867A (en) * | 2013-06-24 | 2013-12-04 | 武汉大学 | Short-term wind power combination prediction method |
CN104077632A (en) * | 2014-06-23 | 2014-10-01 | 华北电力大学 | Wind power field power prediction method based on deep neural network |
-
2017
- 2017-07-03 CN CN201710532750.7A patent/CN107507097A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2013002194A1 (en) * | 2011-06-29 | 2013-01-03 | 三菱重工業株式会社 | Maintenance device for wind power generator and maintenance method for wind power generator |
CN103425867A (en) * | 2013-06-24 | 2013-12-04 | 武汉大学 | Short-term wind power combination prediction method |
CN104077632A (en) * | 2014-06-23 | 2014-10-01 | 华北电力大学 | Wind power field power prediction method based on deep neural network |
Non-Patent Citations (3)
Title |
---|
CAI HONG ZHU 等: "Short-term wind speed forecasting by using chaotic theory and SVM", 《MECHANICS AND MATERIALS》 * |
张金会: "基于支持向量机的智能数据挖掘算法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
朱霄珣 等: "基于PSO参数优化的LS-SVM风速预测方法研究", 《中国电机工程学报》 * |
Cited By (46)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108537372B (en) * | 2018-03-27 | 2020-06-26 | 中南大学 | Wind direction prediction method and yaw control method of wind generating set |
CN108537372A (en) * | 2018-03-27 | 2018-09-14 | 中南大学 | A kind of Yaw control method of wind direction prediction technique and wind power generating set |
CN108427854B (en) * | 2018-03-29 | 2022-06-07 | 深圳市智物联网络有限公司 | Data analysis method and related equipment thereof |
CN108427854A (en) * | 2018-03-29 | 2018-08-21 | 深圳市智物联网络有限公司 | A kind of data analysing method and its relevant device |
CN109146131A (en) * | 2018-07-11 | 2019-01-04 | 国网河北省电力有限公司 | A kind of wind-power electricity generation prediction technique a few days ago |
CN109086916A (en) * | 2018-07-16 | 2018-12-25 | 国家气象中心 | A kind of convection weather nowcasting method and device based on multi-source observation data |
CN109086916B (en) * | 2018-07-16 | 2021-04-02 | 国家气象中心 | Convection weather nowcasting method and device based on multi-source observation data |
CN109063930A (en) * | 2018-08-30 | 2018-12-21 | 上海电力学院 | A kind of dynamic wind power plant general power prediction technique based on clustering |
CN109886488A (en) * | 2019-02-21 | 2019-06-14 | 南方电网科学研究院有限责任公司 | Distributed wind power plant layered hybrid short-term prediction method considering wind speed time lag |
CN109886488B (en) * | 2019-02-21 | 2022-10-18 | 南方电网科学研究院有限责任公司 | Distributed wind power plant layered hybrid short-term prediction method considering wind speed time lag |
CN110263971A (en) * | 2019-05-14 | 2019-09-20 | 西安理工大学 | Super short-period wind power combination forecasting method based on support vector machines |
CN112072691A (en) * | 2019-06-10 | 2020-12-11 | 上海电机学院 | Fan control method based on seasonal decomposition and support vector machine wind power prediction |
CN110390435A (en) * | 2019-07-24 | 2019-10-29 | 广东工业大学 | A kind of wind power forecasting method, device, equipment and readable storage medium storing program for executing |
CN110457800B (en) * | 2019-07-30 | 2023-06-16 | 天津大学 | Wind speed and output conversion method of horizontal axis fan considering mechanical inertia |
CN110457800A (en) * | 2019-07-30 | 2019-11-15 | 天津大学 | Consider the trunnion axis blower wind speed power output translation method of machinery inertial |
CN110428100A (en) * | 2019-07-31 | 2019-11-08 | 新疆新能集团有限责任公司乌鲁木齐电力建设调试所 | A kind of blower short-term power generation power prediction technique |
CN110378541A (en) * | 2019-08-08 | 2019-10-25 | 上海交通大学 | A kind of short-term multistep forecasting method of wind power and device |
CN111160621A (en) * | 2019-12-06 | 2020-05-15 | 江苏方天电力技术有限公司 | Short-term wind power prediction method integrating multi-source information |
CN111160621B (en) * | 2019-12-06 | 2022-06-14 | 江苏方天电力技术有限公司 | Short-term wind power prediction method integrating multi-source information |
CN111222707A (en) * | 2020-01-13 | 2020-06-02 | 湖北工业大学 | Wind speed prediction method based on time series mutation error correction |
CN111222707B (en) * | 2020-01-13 | 2022-04-29 | 湖北工业大学 | Wind speed prediction method based on time series mutation error correction |
CN113554203A (en) * | 2020-04-24 | 2021-10-26 | 国能日新科技股份有限公司 | Wind power prediction method and device based on high-dimensional gridding and LightGBM |
CN113554203B (en) * | 2020-04-24 | 2023-12-26 | 国能日新科技股份有限公司 | Wind power prediction method and device based on high-dimensional meshing and LightGBM |
CN111486991A (en) * | 2020-05-09 | 2020-08-04 | 国家电网有限公司 | Overheating risk early warning device and early warning method for 10kV high-voltage switch cabinet |
CN111666458A (en) * | 2020-06-22 | 2020-09-15 | 中国船级社质量认证公司 | Fitting method for power curve of wind turbine generator |
CN111915070A (en) * | 2020-07-17 | 2020-11-10 | 天津大学 | Prediction method for k-nearest neighbor day-ahead wind speed by adopting atmospheric pressure spatial correlation |
CN111915070B (en) * | 2020-07-17 | 2022-05-31 | 天津大学 | Prediction method for k-nearest neighbor day-ahead wind speed by adopting atmospheric pressure spatial correlation |
CN111862538A (en) * | 2020-08-03 | 2020-10-30 | 中铁二院工程集团有限责任公司 | Large wind early warning method and system for long-span arch bridge construction period |
CN111862538B (en) * | 2020-08-03 | 2021-11-23 | 中铁二院工程集团有限责任公司 | Large wind early warning method and system for long-span arch bridge construction period |
CN112101616A (en) * | 2020-08-10 | 2020-12-18 | 国网山东省电力公司青岛供电公司 | Wind power short-term prediction method and system based on inherent time scale decomposition |
CN114118499A (en) * | 2020-08-25 | 2022-03-01 | 华北电力大学 | Short-term wind power prediction method and device |
CN112085272A (en) * | 2020-09-07 | 2020-12-15 | 国网甘肃省电力公司电力科学研究院 | Wind power prediction method |
CN112348255B (en) * | 2020-11-06 | 2024-04-09 | 湖南大学 | Ultra-short-term wind power prediction method based on wavelet time-frequency imaging |
CN112348255A (en) * | 2020-11-06 | 2021-02-09 | 湖南大学 | Ultra-short-term wind power prediction method based on wavelet time-frequency imaging |
CN112669168B (en) * | 2020-12-15 | 2023-09-05 | 国网辽宁省电力有限公司阜新供电公司 | Short-term prediction method for wind power |
CN112669168A (en) * | 2020-12-15 | 2021-04-16 | 国网辽宁省电力有限公司阜新供电公司 | Short-term wind power prediction method |
CN112712202A (en) * | 2020-12-29 | 2021-04-27 | 广东电网有限责任公司 | Short-term wind power prediction method and device, electronic equipment and storage medium |
CN113447671B (en) * | 2021-07-15 | 2022-09-23 | 中煤科工集团重庆研究院有限公司 | Roadway section wind speed detection method based on high-frequency and low-frequency ultrasonic waves |
CN113447671A (en) * | 2021-07-15 | 2021-09-28 | 中煤科工集团重庆研究院有限公司 | Roadway section wind speed detection method based on high-frequency and low-frequency ultrasonic waves |
CN113704558A (en) * | 2021-08-25 | 2021-11-26 | 成都信息工程大学 | Automatic meteorological station automatic clustering method and system based on differential sequence |
CN113704558B (en) * | 2021-08-25 | 2022-10-14 | 成都信息工程大学 | Automatic meteorological station automatic clustering method and system based on differential sequence |
CN113836801A (en) * | 2021-09-13 | 2021-12-24 | 湖北工业大学 | Prediction method based on CEEMD and improved SSA-LSSVM |
CN113836756A (en) * | 2021-11-29 | 2021-12-24 | 山东华尚电气有限公司 | Intelligent monitoring method and system for annealing process of three-dimensional wound core transformer |
CN114462710A (en) * | 2022-02-11 | 2022-05-10 | 华润电力技术研究院有限公司 | Short-term prediction method, device and medium for fan generated power |
CN117039895A (en) * | 2023-10-09 | 2023-11-10 | 西安热工研究院有限公司 | Wind power prediction method and system for energy storage auxiliary black start |
CN117039895B (en) * | 2023-10-09 | 2024-01-26 | 西安热工研究院有限公司 | Wind power prediction method and system for energy storage auxiliary black start |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107507097A (en) | A kind of short-term wind power prediction method | |
Lange et al. | Physical approach to short-term wind power prediction | |
US8606418B1 (en) | Wind prediction for wind farms through the use of weather radar | |
CN102184337B (en) | Dynamic combination analysis method of new energy generating capacity influenced by meteorological information | |
CN105046374A (en) | Power interval predication method based on nucleus limit learning machine model | |
CN108388962B (en) | Wind power prediction system and method | |
Yang et al. | Day-ahead forecasting of photovoltaic output power with similar cloud space fusion based on incomplete historical data mining | |
CN105373857A (en) | Photovoltaic power station irradiance prediction method | |
CN105654207A (en) | Wind power prediction method based on wind speed information and wind direction information | |
Li et al. | A model-based climatology analysis of wind power resources at 100-m height over the Bohai Sea and the Yellow Sea | |
CN103488869A (en) | Wind power generation short-term load forecast method of least squares support vector machine | |
Labati et al. | A decision support system for wind power production | |
CN106250656A (en) | The complicated landform wind field design platform of the big data of a kind of combination and method | |
CN103489046A (en) | Method for predicting wind power plant short-term power | |
CN106529724A (en) | Wind power prediction method based on grey-combined weight | |
CN103473621A (en) | Wind power station short-term power prediction method | |
CN106875037A (en) | Wind-force Forecasting Methodology and device | |
CN105303250A (en) | Wind power combination prediction method based on optimal weight coefficient | |
CN112862274A (en) | WRF-LES and Deepar combined wind power prediction method | |
CN105825002A (en) | Method for modeling dynamic equivalence of wind power farm based on dynamic grey-relevancy analysis method | |
Yang et al. | Photovoltaic power forecasting with a rough set combination method | |
Chen et al. | Comparison of three methods for short-term wind power forecasting | |
Wang et al. | Optimizing wind farm layout by addressing energy-variance trade-off: A single-objective optimization approach | |
CN114970952A (en) | Photovoltaic output short-term prediction method and system considering environmental factors | |
CN116307282A (en) | Wind power prediction method and system based on mechanism model and data driving |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171222 |