CN102750542A - Support vector regression machine wind speed combination forecast method with interpolation being smoothed and optimized - Google Patents
Support vector regression machine wind speed combination forecast method with interpolation being smoothed and optimized Download PDFInfo
- Publication number
- CN102750542A CN102750542A CN201210173133XA CN201210173133A CN102750542A CN 102750542 A CN102750542 A CN 102750542A CN 201210173133X A CN201210173133X A CN 201210173133XA CN 201210173133 A CN201210173133 A CN 201210173133A CN 102750542 A CN102750542 A CN 102750542A
- Authority
- CN
- China
- Prior art keywords
- wind speed
- interpolation
- wind
- support vector
- optimized
- 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
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a support vector regression machine wind speed combination forecast method with interpolation being smoothed and optimized. The method comprises the following steps of: step I, preprocessing obtained initial data of wind speed at height of a hub of a fan in a wind power station, analyzing sample data, and selecting a time domain for smoothing and optimizing the interpolation; step II, carrying out interpolation smoothing and optimizing on a wind speed time sequence in the selected time domain; step III, carrying out phase space reconstruction on the wind speed sequence which is subjected to interpolation smoothing and optimizing so as to form a sample set required by modeling; and step IV, establishing a corresponding support vector regression machine wind speed combination forecast model with the interpolation being smoothed and optimized by utilizing the sample set obtained from the step III. According to the method, higher forecast accuracy can be obtained compared with a general counting method under the same data condition, the knowability and the controllability of wind power are increased, the grid-connected development of large-scale wind power is favored, and the benefits of the wind power station and a power dispatch department are guaranteed.
Description
Technical field
The present invention relates to the wind power technology field, relate in particular to the support vector regression wind speed combination forecasting method that a kind of interpolation smooth is optimized.
Background technology
Wind-powered electricity generation more and more receives the attention of various countries as a kind of clean reproducible energy, but because all random fluctuation property, intermittence and the uncontrollabilities of wind speed self have increased the uncertainty of the meritorious output of wind-powered electricity generation.This uncertainty will increase the difficulty of dispatching of power netwoks, influences the security and stability of electrical network, runs counter to the development trend that large-scale wind power is incorporated into the power networks.Predict it is the effective measures that address this problem accurately.
Wind speed has directly determined the electric energy of wind energy turbine set output; Therefore can be through the wind speed of axial fan hub height being predicted accurately the meritorious output situation of wind energy turbine set that obtains; Help electric system department and arrange more reasonably operation plan, thereby effectively alleviate the influence that wind-electricity integration brings power distribution network.
Summary of the invention
The technical matters that the present invention solves is to obtain higher forecasting wind speed precision.
In order to overcome the above problems, the support vector regression wind speed combination forecasting method that a kind of interpolation smooth is optimized may further comprise the steps:
Step 1: the wind speed primary data to the wind electric field blower hub height that obtained is carried out pre-service, and the analyzing samples data, the time domain of selecting the interpolation smooth to optimize;
Step 2: the wind speed time series in the times selected territory is carried out interpolation smooth optimization process;
Step 3: the wind series after the interpolation smooth optimization process is carried out phase space reconfiguration form the required sample set of modeling;
Step 4: the sample set that utilizes in the step 3 to be obtained is set up corresponding interpolation smooth and is optimized support vector regression wind speed combination forecasting.
Further, as a kind of preferred, select the time domain of interpolation smooth optimization specifically to may further comprise the steps described in the step 1: the wind speed time series after the pre-service to be carried out the compare of analysis that difference is made in step-by-step, promptly for the wind speed time series: Δ x=(x
2-x
1), be treated to: (x
2-x
1), (x
3-x
2) ..., (x
n-x
N-1), and give corresponding moment attribute to the sequence after handling, that is: supposition t=1 constantly corresponding wind speed difference in change be Δ x=(x
2-x
1), by that analogy, as corresponding wind speed difference in change Δ x=(x of the t=n-1 moment
n-x
N-1); To the analysis of sorting from big to small of this wind speed difference in change time series; Give the ordering sequence number to this sequence, that is: what the wind speed difference in change was maximum is x=1 number, by that analogy; The minimum sequence number of wind speed difference in change is x=n-1, according to selecting the maximum symmetrical left and right sides time neighborhood of wind speed difference in change or comprise from the maximum wind velocity difference in change to begin the time domain of selecting required interpolation smooth to optimize to the principle of the time domain between the x wind speed difference in change.
Further, as a kind of preferred, said step 2 specifically may further comprise the steps: for wind series x
1, x
2, x
3X
n, suppose constantly that at per two wind speed corresponding air speed value is x
mAnd x
M+1Between insert a new wind speed constantly, corresponding air speed value does
Thereby obtain a new wind speed time series, be made as v
1, v
2... V
2n-1
Further, as a kind of preferred, carrying out phase space reconfiguration when forming required sample set in the said step 3: the wind speed of supposing current time determines that by p historical wind speed relational expression v is arranged
m=f (v
M-p, v
M-p+1..., v
M-1) set up, wherein p is called the embedding dimension, need be divided into training sample set and test sample book collection by prediction then.When n sample data of reconstruct, the p dimension air speed data of input sample is v
P+n-1, when the numerical value of p+n-1 is even number v in season
P+n-1=v
P+n-2If odd number is not then done conversion.
Further, as a kind of preferred, concrete steps are following in the said step 4:
Step 4.1: the training sample set that obtains is inputed to BP neural network and SVMs respectively; Adopt corresponding algorithm to set up the forecasting wind speed model of corresponding BP neural network and SVMs; Model to being obtained carries out test analysis, obtains corresponding prediction output;
Step 4.2: the BP neural network of acquisition and the prediction output valve of SVMs forecasting wind speed model are carried out subsequent treatment, select output valve constantly, preserve prediction output and moment corresponding attribute, and carry out the predicated error analysis according to required prediction;
Step 4.3: the input of the output result of BP neural network after the interpolation smooth optimization and SVM prediction model being made up the forecasting wind speed model as SVR; Combine prediction true air speed value constantly to carry out phase space reconfiguration simultaneously and obtain the combined prediction sample set, set up the support vector regression wind speed combination forecasting that the interpolation smooth is optimized then.
Compared with prior art the invention has the beneficial effects as follows: when utilizing method of the present invention to carry out forecasting wind speed; The forecast model of being set up can obtain the precision of prediction higher than general statistical method under equal data qualification; Promote the knowability and the controllability of wind-powered electricity generation; Help the development of being incorporated into the power networks of large-scale wind power, ensure the interests of wind energy turbine set and power scheduling department.
Description of drawings
When combining accompanying drawing to consider; Through with reference to following detailed, can more completely understand the present invention better and learn wherein many attendant advantages easily, but accompanying drawing described herein is used to provide further understanding of the present invention; Constitute a part of the present invention; Illustrative examples of the present invention and explanation thereof are used to explain the present invention, do not constitute to improper qualification of the present invention, wherein:
The method flow diagram of Fig. 1 embodiment of the invention;
The support vector regression wind speed combination forecasting method design sketch that the smooth of Fig. 2 interpolation is optimized.
Embodiment
Followingly embodiments of the invention are described with reference to Fig. 1.
For make above-mentioned purpose, feature and advantage can be more obviously understandable, below in conjunction with accompanying drawing and embodiment the present invention done further detailed explanation.
Support vector regression wind speed combination forecasting method as shown in Figure 1, that a kind of interpolation smooth is optimized may further comprise the steps:
S1, prediction beginning;
S2, obtain original air speed data;
S3, pre-service;
S4, select suitable time domain;
S5, interpolation smooth are optimized;
S6, reconstructed sample collection;
S7, support vector (SVM) modeling;
S8, BP neural network (BP-ANN) modeling;
S9, S10 test model;
S11, subsequent treatment are selected test sample book;
S12, reconstructed sample collection;
S13, support vector regression (SVR) wind speed compositional modeling;
S14, error analysis are also exported the wind speed combination forecasting.
Embodiment
Following implementation is to illustrate the prediction scheme that on the MATLAB emulation platform, realizes.The predicted data that adopts is taken from the actual measurement air speed data of Shanxi Province in the operation wind energy turbine set, and target is that the short-term wind speed is predicted, time scale is chosen as 1 hour.Be carved into 09:00 on May 1 historical wind speed data constantly when selecting this wind field 00:00 on March 1st, 2008 as research object.Concrete performing step is following:
Step 1: the air speed data that investigation is obtained carries out pre-service, rejects bad value.And data are carried out normalization handle, with its through type y=(y
Max-y
Min) * (x-x
Min) ÷ (x
Max-x
Min)+y
MinBe transformed between [0,1].
Step 2: on the MATLAB platform, import and read sample data, the analyzing samples data, the time domain of selecting the interpolation smooth to optimize:
Wind speed time series after the pre-service is carried out step-by-step do poor comparing, promptly for the wind speed time series: Δ x=(x
2-x
1), be treated to: (x
2-x
1), (x
3-x
2) ..., (x
n-x
N-1), and give corresponding moment attribute to the sequence after handling, that is: supposition t=1 constantly corresponding wind speed difference in change be Δ x=(x
2-x
1), by that analogy, as corresponding wind speed difference in change Δ x=(x of the t=n-1 moment
n-x
N-1), to the analysis of sorting from big to small of wind speed transformation period sequence, selecting the maximum left and right sides neighborhood of wind speed difference in change is the time domain that required interpolation smooth is optimized, and promptly is carved into April 10 09:00 during 00:00 on April 1 constantly as the time domain of interpolation.
Step 3: the wind speed time series in the times selected territory is carried out interpolation smooth optimization process:
For wind series x
1, x
2, x
3X
n, suppose constantly that at per two wind speed corresponding air speed value is x
mAnd x
M+1Between insert a new wind speed constantly, corresponding air speed value does
Thereby obtain a new wind speed time series, be made as v
1, v
2... v
2n-1
Step 4: the sequence to after handling is carried out correlation analysis, and obtaining best embedding dimension is 6.Carry out the reconstruct of set of data samples then.But follow following principle:
Because embedding dimension is 6, when n sample data of reconstruct, the 6th dimension air speed data of input sample is v
6+n-1, when (6+n-1) is even number v in season
6+n-1=v
6+n-2
Step 5: the sample set after utilize optimizing is realized the building of forecasting wind speed model of BP neural network method and SVMs method respectively on the MATLAB platform:
(1) BP neural network method modeling: selection be a single latent layer BP network, the number of hidden nodes is 20, the output dimension is 1 dimension, selects learning rate changing momentum gradient algorithm, the average relative error value of corresponding sample prediction is 6.4901%.
(2) SVMs method modeling: the parameter that adopts the cross validation method to obtain model is c=4, g=0.5 (wherein c is a penalty coefficient, and g is the nuclear width), and the average relative error value of corresponding sample prediction is 6.8737%.
Step 6: subsequent treatment is carried out in the prediction output to above-mentioned steps, obtains corresponding constantly BP-ANN of forecasting wind speed and the prediction air speed value of SVM.
Step 7:, combine corresponding wind speed actual value constantly simultaneously, the input sample set of reconstruct built-up pattern respectively with the output of BP-ANN and SVM input as SVR.Utilize SVR to carry out training study, the training and testing effect of combination forecasting wind speed model is as shown in Figure 2.Model parameter is: c=4, g=0.0625 (wherein c is a penalty coefficient, and g is the nuclear width).Wind speed sample before combine handling simultaneously carries out comparison that predicting the outcome of BP neural network method and SVMs method modeling analysis carry out predicated error respectively shown in form 1.
The predicated error compare of analysis of the different Forecasting Methodologies of table 1
As stated, embodiments of the invention have been carried out explanation at length, but as long as not breaking away from inventive point of the present invention and effect in fact can have a lot of distortion, this will be readily apparent to persons skilled in the art.Therefore, such variation also all is included within protection scope of the present invention.
Claims (5)
1. the support vector regression wind speed combination forecasting method that the interpolation smooth is optimized is characterized in that, may further comprise the steps:
Step 1: the wind speed primary data to the wind electric field blower hub height that obtained is carried out pre-service, and the analyzing samples data, the time domain of selecting the interpolation smooth to optimize;
Step 2: the wind speed time series in the times selected territory is carried out interpolation smooth optimization process;
Step 3: the wind series after the interpolation smooth optimization process is carried out phase space reconfiguration form the required sample set of modeling;
Step 4: the sample set that utilizes in the step 3 to be obtained is set up corresponding interpolation smooth and is optimized support vector regression wind speed combination forecasting.
2. the support vector regression wind speed combination forecasting method that a kind of interpolation smooth according to claim 1 is optimized; It is characterized in that; Select the time domain of interpolation smooth optimization specifically to may further comprise the steps described in the step 1: the wind speed time series after the pre-service to be carried out the compare of analysis that difference is made in step-by-step, promptly for wind speed time series: x
1, x
2, x
3X
n, be treated to: (x
2-x
1), (x
3-x
2) ..., (x
n-x
N-1), and give corresponding moment attribute to the sequence after handling, that is: supposition t=1 constantly corresponding wind speed difference in change be Δ x=(x
2-x
1), by that analogy, as corresponding wind speed difference in change Δ x=(x of the t=n-1 moment
n-x
N-1); To the analysis of sorting from big to small of this wind speed difference in change time series; Give the ordering sequence number to this sequence, that is: what the wind speed difference in change was maximum is x=1 number, by that analogy; The minimum sequence number of wind speed difference in change is x=n-1, according to selecting the maximum symmetrical left and right sides time neighborhood of wind speed difference in change or comprise from the maximum wind velocity difference in change to begin the time domain of selecting required interpolation smooth to optimize to the principle of the time domain between the x wind speed difference in change.
3. the support vector regression wind speed combination forecasting method that 1 described a kind of interpolation smooth is optimized according to claims is characterized in that said step 2 specifically may further comprise the steps:
4. the support vector regression wind speed combination forecasting method that 1 described a kind of interpolation smooth is optimized according to claims; It is characterized in that; Carrying out phase space reconfiguration when forming required sample set in the said step 3: the wind speed of supposing current time determines that by p historical wind speed relational expression v is arranged
m=f (v
M-p, v
M-p+1..., v
M-1) set up, wherein p is called the embedding dimension, need be divided into training sample set and test sample book collection by prediction then, and when n sample data of reconstruct, the p dimension air speed data of input sample is v
P+n-1, when the numerical value of p+n-1 is even number v in season
P+n-1=v
P+n-2If odd number is not then done conversion.
5. the support vector regression wind speed combination forecasting method that 1 described a kind of interpolation smooth is optimized according to claims is characterized in that concrete steps are following in the said step 4:
Step 4.1: the training sample set that obtains is inputed to BP neural network and SVMs respectively; Adopt corresponding algorithm to set up the forecasting wind speed model of corresponding BP neural network and SVMs; Model to being obtained carries out test analysis, obtains corresponding prediction output;
Step 4.2: the BP neural network of acquisition and the prediction output valve of SVMs forecasting wind speed model are carried out subsequent treatment, select output valve constantly, preserve prediction output and moment corresponding attribute, and carry out the predicated error analysis according to required prediction;
Step 4.3: the input of the output result of BP neural network after the interpolation smooth optimization and SVM prediction model being made up the forecasting wind speed model as SVR; Combine prediction true air speed value constantly to carry out phase space reconfiguration simultaneously and obtain the combined prediction sample set, set up the support vector regression wind speed combination forecasting that the interpolation smooth is optimized then.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210173133XA CN102750542A (en) | 2012-05-29 | 2012-05-29 | Support vector regression machine wind speed combination forecast method with interpolation being smoothed and optimized |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210173133XA CN102750542A (en) | 2012-05-29 | 2012-05-29 | Support vector regression machine wind speed combination forecast method with interpolation being smoothed and optimized |
Publications (1)
Publication Number | Publication Date |
---|---|
CN102750542A true CN102750542A (en) | 2012-10-24 |
Family
ID=47030707
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210173133XA Pending CN102750542A (en) | 2012-05-29 | 2012-05-29 | Support vector regression machine wind speed combination forecast method with interpolation being smoothed and optimized |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102750542A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103514488A (en) * | 2013-10-08 | 2014-01-15 | 哈尔滨工程大学 | Electrical power system short-term load forecasting device and method based on combination forecasting model |
CN105354763A (en) * | 2015-11-16 | 2016-02-24 | 华北电力科学研究院有限责任公司 | Method and device for measuring wind speed of incoming flow of wind turbine generator |
CN110210343A (en) * | 2019-05-20 | 2019-09-06 | 深圳供电局有限公司 | Big data face recognition method and system and readable storage medium |
CN113291055A (en) * | 2021-04-14 | 2021-08-24 | 西安理工大学 | Artificial intelligent flexographic printing pressure prediction method |
CN113408071A (en) * | 2021-06-22 | 2021-09-17 | 鲁能集团有限公司 | Wind turbine generator tower attitude prediction method and system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101661530A (en) * | 2009-09-30 | 2010-03-03 | 华北电力大学(保定) | Method for acquiring steady-state equivalent wind speed and generated power in wind power station based on correlation analysis |
CN102102626A (en) * | 2011-01-30 | 2011-06-22 | 华北电力大学 | Method for forecasting short-term power in wind power station |
CN102184337A (en) * | 2011-06-07 | 2011-09-14 | 中国电力科学研究院 | Dynamic combination analysis method of new energy generating capacity influenced by meteorological information |
CN102184453A (en) * | 2011-05-16 | 2011-09-14 | 上海电气集团股份有限公司 | Wind power combination predicting method based on fuzzy neural network and support vector machine |
-
2012
- 2012-05-29 CN CN201210173133XA patent/CN102750542A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101661530A (en) * | 2009-09-30 | 2010-03-03 | 华北电力大学(保定) | Method for acquiring steady-state equivalent wind speed and generated power in wind power station based on correlation analysis |
CN102102626A (en) * | 2011-01-30 | 2011-06-22 | 华北电力大学 | Method for forecasting short-term power in wind power station |
CN102184453A (en) * | 2011-05-16 | 2011-09-14 | 上海电气集团股份有限公司 | Wind power combination predicting method based on fuzzy neural network and support vector machine |
CN102184337A (en) * | 2011-06-07 | 2011-09-14 | 中国电力科学研究院 | Dynamic combination analysis method of new energy generating capacity influenced by meteorological information |
Non-Patent Citations (1)
Title |
---|
蒋金良等: "基于ARIMA模型的自动站风速预测", 《控制理论与应用》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103514488A (en) * | 2013-10-08 | 2014-01-15 | 哈尔滨工程大学 | Electrical power system short-term load forecasting device and method based on combination forecasting model |
CN103514488B (en) * | 2013-10-08 | 2016-06-29 | 哈尔滨工程大学 | Power-system short-term load forecasting device and method based on combination forecasting |
CN105354763A (en) * | 2015-11-16 | 2016-02-24 | 华北电力科学研究院有限责任公司 | Method and device for measuring wind speed of incoming flow of wind turbine generator |
CN110210343A (en) * | 2019-05-20 | 2019-09-06 | 深圳供电局有限公司 | Big data face recognition method and system and readable storage medium |
CN113291055A (en) * | 2021-04-14 | 2021-08-24 | 西安理工大学 | Artificial intelligent flexographic printing pressure prediction method |
CN113408071A (en) * | 2021-06-22 | 2021-09-17 | 鲁能集团有限公司 | Wind turbine generator tower attitude prediction method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Kusiak et al. | Short-term prediction of wind farm power: A data mining approach | |
CN102184453A (en) | Wind power combination predicting method based on fuzzy neural network and support vector machine | |
CN105046374A (en) | Power interval predication method based on nucleus limit learning machine model | |
CN105426956A (en) | Ultra-short-period photovoltaic prediction method | |
CN104077632A (en) | Wind power field power prediction method based on deep neural network | |
CN109086928A (en) | Photovoltaic plant realtime power prediction technique based on SAGA-FCM-LSSVM model | |
CN106875033A (en) | A kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting | |
CN105139264A (en) | Photovoltaic generation capacity prediction method based on particle swarm algorithm wavelet neural network | |
CN104835018B (en) | The reliability estimation method and device of medium voltage distribution network rack | |
CN105160423A (en) | Photovoltaic power generation prediction method based on Markov residual error correction | |
CN102750542A (en) | Support vector regression machine wind speed combination forecast method with interpolation being smoothed and optimized | |
CN104376371B (en) | A kind of distribution based on topology is layered load forecasting method | |
CN102738792B (en) | Wind power predicting method | |
CN103927460A (en) | Wind power plant short-term wind speed prediction method based on RBF | |
CN106295857A (en) | A kind of ultrashort-term wind power prediction method | |
CN104809532A (en) | Method for predicting generating capacity of photovoltaic system | |
CN102938562A (en) | Prediction method of total wind electricity power in area | |
CN104978605A (en) | Large-scale wind power prediction system and method based on deep learning network | |
CN116599050A (en) | Photovoltaic prediction method and related device based on self-attention mechanism | |
CN105844350B (en) | Short-term wind power prediction system based on covariance optimal combination model | |
Kim et al. | Transformer based prediction method for solar power generation data | |
CN109426901A (en) | Long-term power consumption prediction method and device in one kind | |
CN106485603A (en) | A kind of short-term wind speed forecasting method being applied to wind-power electricity generation | |
CN105303319A (en) | Wind power plant dynamic clustering equivalence method based on operating data | |
CN105207255B (en) | A kind of power system peak regulation computational methods suitable for wind power output |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C12 | Rejection of a patent application after its publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20121024 |