CN106682760A - Wind power climbing prediction method - Google Patents
Wind power climbing prediction method Download PDFInfo
- Publication number
- CN106682760A CN106682760A CN201610822667.9A CN201610822667A CN106682760A CN 106682760 A CN106682760 A CN 106682760A CN 201610822667 A CN201610822667 A CN 201610822667A CN 106682760 A CN106682760 A CN 106682760A
- Authority
- CN
- China
- Prior art keywords
- atom
- wind power
- signal
- prediction
- decomposition
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The invention provides a wind power climbing prediction method which includes the following steps: 1. conducting sparse decomposition on the amount of power while climbing; 2. conducting self-prediction on the atomic component which is obtained after decomposition, conducting radial basis function neural network prediction on the residual component; and 3. conducting linear adding on each prediction component to obtain the predicted value of a next moment. The method predicts the amount of wind power climbing by the sliding prediction method which combines the sparse decomposition and the radial basis function neural network, establishes a model for predicting wind power climbing events, can predict the amount of wind power climbing, and increases prediction precision.
Description
Technical field
The present invention relates to wind power climbing Forecasting Methodology, more particularly to it is a kind of refreshing based on atom Its Sparse Decomposition and radial direction base
The wind power climbing Forecasting Methodology of Jing networks.
Background technology
In recent years, the whole world all suffers from the multi-party surface pressure such as energy crisis and the horizontal continuous decrease of natural environment, how to enter
The adjustment of row energy resources structure and optimization, efficiently develop and use regenerative resource, have become the countries in the world energy and send out
The key components of exhibition strategy work.Wherein, in renewable energy power generation technology, with resource extensively, green non-pollution
The features such as wind-power electricity generation, have become the larger new energy power generation technology of level of development relative maturity and installed capacity.
Meanwhile, with the development of extensive high concentration degree wind energy turbine set, the regional fluctuation phenomenon of wind power output power is also got over
Come more serious, not only show that wind power has uncertainty, it may appear that the big climbing phenomenon of the fast amplitude of speed, affect short-term
Balance in electrical network between power supply and load, and the increasing of feature analysiss difficulty, wind-powered electricity generation so that active power of wind power field is exerted oneself
Power prediction precision is difficult to be lifted, therefore, the research of prediction of climbing to wind power has become more and more important research neck
Domain.
The content of the invention
The Forecasting Methodology it is an object of the invention to provide a kind of wind power is climbed, carries out the prediction of wind power climbing amount,
Improve precision of prediction.
The present invention is for the solution technical scheme that adopted of its technical problem,
A kind of wind power climbing Forecasting Methodology, it is characterised in that comprise the following steps：
(1) wind power climbing amount is carried out into Its Sparse Decomposition；
(2) the atom component obtained after decomposition is carried out, from prediction, residual component being carried out into radial base neural net prediction；
(3) each anticipation component is carried out into the predictive value that linear, additive obtains subsequent time.
In step (1), using the historical time sequence of wind power climbing amount as forecast model input, by sparse point
Solution method carries out the Its Sparse Decomposition of data：
Climbing amount time serieses are expressed as into the sparse linear combination of atom, its general expression is
In formula, ckIt is sparse coefficient,For an approximation signal of primary signal, by signal f be approximately represented as M it is former
The linear combination of son,
Constructed complete dictionary：
By selecting Gabor atoms, generating function g (t) is taken for Gaussian function
Again by the sliding-model control such as the translation to generating function, flexible, atom is generated, so as to obtain complete dictionary
Collection；
Atom Corresponding Sparse Algorithm adopts orthogonal matching pursuit algorithm, and to whole atoms that every step is decomposed process is orthogonalized,
Step is as follows：
1) first the atom for matching the most is chosen from over-complete dictionary of atoms, that is, is met
Assume R0=f, signal f is decomposed into following form：
Wherein,Represent R0It is rightProjection, i.e.,Match the most for first in the atom found out
The atom of signal, R1For the residual error obtained after first time signal decomposition,With R1It is orthogonal, obtains
Select the atom of near optimalSo that
Wherein, 0 ＜ α≤1,
α=1 is taken,
Next to residual error R1Carrying out identical step carries out computing, obtains
Meet
Treat after decomposed signal decomposed successively, just can obtain after such m+1 iteration
WhereinMeet
Assume that the atom for matching the most isThe process of Schimidt orthogonalization is carried out to it：
Residual error RMIn umUpper projection, i.e.,：
So as to the representation of signal f is changed into：
Then for signal f to be decomposed completes the Its Sparse Decomposition of OMP algorithms.
In step (2)：On the basis of atom is selected, by decomposition after the n atom component for obtaining, carry out atom point
Certainly the prediction of amount expression formula, a remaining residual component carries out the training of model, so as to pre- as the input of radial basis function network
Measure the predictive value of the corresponding subsequent time of residual component.
In step (3)：By n+1 anticipation component of prediction in step (2), these components are carried out into linear regression superposition
And amendment, the predictive value of the wind power climbing amount of the subsequent time as predicted.
It is an advantage of the current invention that the method is using the slip combined based on Its Sparse Decomposition method and radial base neural net
Forecasting Methodology carries out the prediction of wind power climbing amount, the forecast model of wind power climbing event is established, so as to enter sector-style
The prediction of electrical power climbing amount, improves precision of prediction.
Description of the drawings
Fig. 1 is the algorithm flow chart of Forecasting Methodology proposed by the present invention；
Fig. 2 is the forecast model flow chart of the method；
Fig. 3 is measured power curve and climbing amount schematic diagram；
Fig. 4 is pre- power scale climbing amount curve.
Specific embodiment
In order that technological means, creation characteristic, reached purpose and effect that the present invention is realized are easy to understand, tie below
Diagram and specific embodiment are closed, the present invention is expanded on further.
As shown in Figure 1 and Figure 2, its particular content of Forecasting Methodology proposed by the present invention is：(1) by wind power climbing amount
Historical time sequence carries out the Its Sparse Decomposition of data as the input of forecast model by Its Sparse Decomposition method.Its Sparse Decomposition is managed
Include that 2 is most of by main --- over-complete dictionary of atoms；Its Sparse Decomposition algorithm.Climbing amount time serieses are expressed as into the dilute of atom
Thin linear combination, its general expression is：
In formula, ckIt is sparse coefficient,For an approximation signal of primary signal, by signal f be approximately represented as M it is former
The linear combination of son
It is by selecting Gabor atoms, taking generating function g (t) for Gaussian function that the present invention constructed complete dictionary：
Again by the sliding-model control such as the translation to generating function, flexible, atom is generated, so as to obtain complete dictionary
Collection.
(2) atom Corresponding Sparse Algorithm adopts orthogonal matching pursuit algorithm (Orthogonal Matching Pursuit, OMP)
Whole atoms that every step is decomposed are orthogonalized process by algorithm, not only meet required precision but also convergence rate faster.Step
It is as follows：
1) first the atom for matching the most is chosen from over-complete dictionary of atoms, that is, is met
Assume R0=f, such signal f can be decomposed into following form：
Wherein,Represent R0It is rightProjection, i.e.,Match the most for first in the atom found out
The atom of signal, R1For the residual error obtained after first time signal decomposition.It is not difficult to find out,With R1It is orthogonal, in can be
Arrive：
In order that approximate error R1Energy it is minimum, it is necessary to select to causeMaximum
In the middle of infinite dimension, can not find under normal circumstancesExtreme value, be only possible to optimum selecting in some sense near
Like optimal atomSo that
Wherein, α is referred to as Optimization Factor, meets ＜ α≤1 of condition 0.
And in the case of finite dimensional, |<R,gγ>| there is maximum.Therefore, use in the case of finite dimensional
When, generally take α=1.
Next to residual error R1Carrying out identical step carries out computing, obtains
Meet
Treat after decomposed signal decomposed successively, just can obtain after such m+1 iteration
WhereinMeet：
The projecting direction orthogonalization that OMP algorithms pass through each step difference terms to iteration, realizes to selected all originals
Son is orthogonalized process, compared to the optimality that MP algorithms ensure iteration, and greatly reduces the number of times of iteration so as to receive
Hold back speed faster.
In OMP algorithms, it is assumed that the atom for matching the most isThe process of Schimidt orthogonalization is carried out to it：
Residual error RMIn umUpper projection, i.e.,：
So as to the representation of signal f is changed into：
Then for signal f to be decomposed completes the Its Sparse Decomposition of OMP algorithms.
(3) on the basis of above-mentioned steps select atom, by decomposition after the n atom component for obtaining, carry out atom
Certainly the prediction of weight expression, a remaining residual component carries out the training of model as the input of radial basis function network, so as to
Predict the predictive value of the corresponding subsequent time of residual component.
(4) by n+1 anticipation component of prediction in step (3), these components are carried out into linear regression superposition and amendment,
The predictive value of the wind power climbing amount of the subsequent time as predicted.
Fig. 3, Fig. 4 are Example Verification, and Fig. 3 is the wind power time serieses of 500 actual data of Shanghai wind energy turbine set,
With the time serieses of corresponding wind power climbing amount, 50 sample points after this model prediction, that is, Fig. 4 is obtained, and led to
Cross and contrasted with conventional wavelet neural network and the prediction of single radial base neural net, by Error Calculation, this
The error of Its Sparse Decomposition described in bright and radial base neural net Forecasting Methodology is minimum, and its root-mean-square error is 11.09%, is met
Precision of prediction.
Embodiment of above technology design only to illustrate the invention and feature, its object is to allow those skilled in the art
Member understands present disclosure and is carried out, and can not be limited the scope of the invention with this, all according to spirit of the invention
Equivalence changes or modification that essence is done, all should cover within the scope of the present invention.
Claims (4)
1. a kind of wind power is climbed Forecasting Methodology, it is characterised in that comprised the following steps：
(1) wind power climbing amount is carried out into Its Sparse Decomposition；
(2) the atom component obtained after decomposition is carried out, from prediction, residual component being carried out into radial base neural net prediction；
(3) each anticipation component is carried out into the predictive value that linear, additive obtains subsequent time.
2. a kind of wind power according to claim 1 is climbed Forecasting Methodology, it is characterised in that
In step (1), using the historical time sequence of wind power climbing amount as forecast model input, by Its Sparse Decomposition side
Method carries out the Its Sparse Decomposition of data：
Climbing amount time serieses are expressed as into the sparse linear combination of atom, its general expression is
In formula, ckIt is sparse coefficient,For an approximation signal of primary signal, signal f is approximately represented as into M atom
Linear combination,
Constructed complete dictionary：
By selecting Gabor atoms, generating function g (t) is taken for Gaussian function
Again by the sliding-model control such as the translation to generating function, flexible, atom is generated, so as to obtain complete wordbook；
Atom Corresponding Sparse Algorithm adopts orthogonal matching pursuit algorithm, and to whole atoms that every step is decomposed process, step are orthogonalized
It is as follows：
1) first the atom for matching the most is chosen from over-complete dictionary of atoms, that is, is met
Assume R0=f, signal f is decomposed into following form：
Wherein,Represent R0It is rightProjection, i.e.,For in the atom found out first matched signal the most
Atom, R1For the residual error obtained after first time signal decomposition,With R1It is orthogonal, obtains
Select the atom of near optimalSo that
Wherein, 0 ＜ α≤1,
α=1 is taken,
Next to residual error R1Carrying out identical step carries out computing, obtains
Meet
Treat after decomposed signal decomposed successively, just can obtain after such m+1 iteration
WhereinMeet
Assume that the atom for matching the most isThe process of Schimidt orthogonalization is carried out to it：
Residual error RMIn umUpper projection, i.e.,：
So as to the representation of signal f is changed into：
Then for signal f to be decomposed completes the Its Sparse Decomposition of OMP algorithms.
3. a kind of wind power according to claim 1 is climbed Forecasting Methodology, it is characterised in that the process of step (2) is：
On the basis of atom is selected, by decomposition after the n atom component for obtaining, carry out atom weight expression from predict,
A remaining residual component carries out the training of model as the input of radial basis function network, so as to predict residual component correspondence
Subsequent time predictive value.
4. a kind of wind power according to claim 1 is climbed Forecasting Methodology, it is characterised in that the process of step (3) is：
By n+1 anticipation component of prediction in step (2), these components are carried out into linear regression superposition and amendment, as predicted
The predictive value of the wind power climbing amount of subsequent time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610822667.9A CN106682760A (en) | 2016-09-13 | 2016-09-13 | Wind power climbing prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610822667.9A CN106682760A (en) | 2016-09-13 | 2016-09-13 | Wind power climbing prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106682760A true CN106682760A (en) | 2017-05-17 |
Family
ID=58840248
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610822667.9A Pending CN106682760A (en) | 2016-09-13 | 2016-09-13 | Wind power climbing prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106682760A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108537380A (en) * | 2018-04-04 | 2018-09-14 | 福州大学 | A kind of Methods of electric load forecasting based on rarefaction representation |
| 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 |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103117546A (en) * | 2013-02-28 | 2013-05-22 | 武汉大学 | Ultrashort-term slide prediction method for wind power |
| CN105335796A (en) * | 2015-11-02 | 2016-02-17 | 华北电力大学 | System and method for predicting wind farm output power climbing event |
-
2016
- 2016-09-13 CN CN201610822667.9A patent/CN106682760A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103117546A (en) * | 2013-02-28 | 2013-05-22 | 武汉大学 | Ultrashort-term slide prediction method for wind power |
| CN105335796A (en) * | 2015-11-02 | 2016-02-17 | 华北电力大学 | System and method for predicting wind farm output power climbing event |
Non-Patent Citations (1)
| Title |
|---|
| 任凌志: "基于多尺度线调频基稀疏信号分解的风力发电机组故障诊断研究", 《中国优秀硕士学位论文全文数据库（工程科技II辑）》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108537380A (en) * | 2018-04-04 | 2018-09-14 | 福州大学 | A kind of Methods of electric load forecasting based on rarefaction representation |
| CN108537380B (en) * | 2018-04-04 | 2022-03-25 | 福州大学 | Power load prediction method based on sparse representation |
| 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 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Forecasting power output of photovoltaic system using a BP network method | |
| Zhang et al. | Wind speed prediction of IPSO-BP neural network based on lorenz disturbance | |
| Wang et al. | Decision tree based online stability assessment scheme for power systems with renewable generations | |
| CN106532778B (en) | Method for calculating maximum access capacity of distributed photovoltaic grid connection | |
| CN104584357B (en) | Being controlled using distributed energy sources reduces the loss of distribution system | |
| CN105429129B (en) | Intermittent energy power generation capacity confidence evaluation method considering network constraints | |
| Zhang et al. | Robust optimization for dynamic economic dispatch under wind power uncertainty with different levels of uncertainty budget | |
| CN108306303A (en) | A kind of consideration load growth and new energy are contributed random voltage stability assessment method | |
| CN103455716B (en) | A kind of power system voltage stabilization margin calculation method based on super short-period wind power prediction | |
| Mao et al. | A novel short-term wind speed prediction based on MFEC | |
| CN110148068A (en) | One kind being based on meteorological data similarity analysis and LSTM neural fusion photovoltaic plant ultra-short term power forecasting method | |
| CN110334870B (en) | Photovoltaic power station short-term power prediction method based on gated cyclic unit network | |
| CN105186499A (en) | Multi-target probabilistically optimal power flow fuzzy modelling and solving method for power distribution network | |
| CN104050517A (en) | Photovoltaic power generation forecasting method based on GRNN | |
| Liu et al. | PV generation forecasting with missing input data: A super-resolution perception approach | |
| CN107204615A (en) | A kind of method and system for realizing power prediction | |
| CN108336739A (en) | A kind of Probabilistic Load Flow on-line calculation method based on RBF neural | |
| CN108599239A (en) | A kind of droop control type isolated island micro-capacitance sensor voltage quality probability evaluation method of failure | |
| CN110009141B (en) | Climbing event prediction method and system based on SDAE feature extraction and SVM classification model | |
| CN112183641A (en) | Transient frequency stability assessment method and system integrating prediction-correction deep learning | |
| CN104268659A (en) | Photovoltaic power station generated power super-short-term prediction method | |
| CN106611243A (en) | Residual correction method for wind speed prediction based on GARCH (Generalized ARCH) model | |
| Abdullah et al. | A noniterative method to estimate load carrying capability of generating units in a renewable energy rich power grid | |
| CN103887813A (en) | Control method of wind power system operation based on wind power prediction uncertainty | |
| CN105140967B (en) | A kind of appraisal procedure of the demand of peak regulation containing New-energy power system |
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: 20170517 |