CN105825289A - Prediction method for wind power time sequence - Google Patents
Prediction method for wind power time sequence Download PDFInfo
- Publication number
- CN105825289A CN105825289A CN201510956093.XA CN201510956093A CN105825289A CN 105825289 A CN105825289 A CN 105825289A CN 201510956093 A CN201510956093 A CN 201510956093A CN 105825289 A CN105825289 A CN 105825289A
- Authority
- CN
- China
- Prior art keywords
- wind power
- power time
- time series
- outcome
- model
- 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 46
- 241001123248 Arma Species 0.000 claims abstract description 27
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 9
- 238000001914 filtration Methods 0.000 claims abstract description 6
- 238000005096 rolling process Methods 0.000 claims abstract description 6
- 238000012360 testing method Methods 0.000 claims abstract description 5
- 230000002159 abnormal effect Effects 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 20
- 230000008569 process Effects 0.000 claims description 10
- 230000006698 induction Effects 0.000 claims description 2
- 238000005259 measurement Methods 0.000 claims 1
- 238000013213 extrapolation Methods 0.000 abstract description 2
- 238000012545 processing Methods 0.000 abstract description 2
- 230000009466 transformation Effects 0.000 description 8
- 230000008859 change Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000005611 electricity Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000003556 assay Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000003412 degenerative effect Effects 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000035800 maturation Effects 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000001172 regenerating effect Effects 0.000 description 1
- 238000011160 research Methods 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a prediction method for a wind power time sequence. The method comprises the steps of obtaining the historical statistical data of the wind power in a wind power plant, processing abnormal historical data to obtain a wind power time sequence, conducting the wavelet decomposition on the wind power time sequence to decompose the wind power time sequence into a low-frequency component and a high-frequency component, conducting the stationary test and the stationary treatment on the wind power time sequence of the high-frequency component, conducting the order and parameter fitting for the ARMA model of the wind power time sequence of the high-frequency component, converting the ARMA model into a Kalman filtering model and making a rolling forward prediction to obtain a high-frequency prediction result; conducting the slope rolling forward extrapolation prediction on the wind power time sequence of the low-frequency component to obtain a low-frequency prediction result; and finally obtaining a final prediction result through the wavelet reconstruction.
Description
Technical field
The invention belongs to power system new energy field, especially relate to a kind of method utilizing Combined model forecast wind power time series.
Background technology
In recent years, lack of energy and environmental pollution states are day by day serious, and regenerative resource gradually receives global concern.Wind energy, solar energy are inexhaustible clean energy resourcies, and wind generating technology has started large-scale application as technology the most ripe in generation of electricity by new energy.
But, undulatory property, randomness and the intermittence having due to wind speed itself, and wind speed is affected by multiple physical factors such as temperature, humidity, air pressure, the state of exerting oneself of wind generator system can be affected by serious, presents passage in time and the state that constantly fluctuates, change, cannot accurately predict.So, after large-scale wind power generating set accesses electrical network, the stability of operation of power networks and reliability can be brought challenge greatly.Therefore, the power sending wind speed or wind-power electricity generation is accurately predicted, is to ensure that one of key factor of power network safety operation under high wind-powered electricity generation permeability, has highly important Research Significance.
Summary of the invention
In sum, the method for more Accurate Prediction wind power of necessary offer a kind of short time, effectively reduces and abandons wind, improve the utilization rate of wind power resources.
The present invention relates to the Forecasting Methodology of a kind of wind power time series, comprise the following steps: obtain the historical statistical data of wind energy turbine set wind power, and history abnormal data is processed, obtain wind power time series;Wind power time series is carried out wavelet decomposition, is decomposed into low frequency component and high fdrequency component;The wind power time series of high fdrequency component is carried out stationary test and tranquilization processes;The wind power time series of high fdrequency component is carried out determination and the parameter fitting of the exponent number of arma modeling;Arma modeling be converted to Kalman filter model and do rolling forward prediction, obtaining high frequency and predict the outcome;The wind power time series of low frequency component is carried out slope and rolls outside forecast forward, obtain low frequency and predict the outcome;And the high frequency obtained predicted the outcome and low frequency predicts the outcome and obtained final predicting the outcome by wavelet reconstruction.
Relative to prior art, Kalman filter model is combined by wind power prediction model of the present invention with arma modeling, adds unidirectional negative feedback links in Discrete stochastic systems, is conducive to improving the precision of prediction of arma modeling;Additionally, by utilizing wavelet analysis method that original series is decomposed into different components, feature according to different components selects Forecasting Methodology prediction targetedly, is finally finally predicted the outcome by wavelet reconstruction and can effectively improve the precision of model prediction, reduces error.
Accompanying drawing explanation
The flow chart of the Forecasting Methodology of the wind power time series that Fig. 1 provides for the embodiment of the present invention.
The FB(flow block) of the Forecasting Methodology of the combined method wind power time series that Fig. 2 provides for the present invention.
The Kalman filtering Discrete stochastic systems figure that Fig. 3 provides for the present invention.
The Kalman filtering that Fig. 4 provides for the present invention optimize after the block diagram of Discrete Linear Random System.
Detailed description of the invention
Below according to Figure of description and in conjunction with specific embodiments technical scheme is stated the most in detail.
Referring to Fig. 1, the built-up pattern that utilizes that the present invention provides carries out the Forecasting Methodology of wind power time series, comprises the following steps:
Step S10, obtains the historical statistical data of wind energy turbine set wind power, and processes history abnormal data, obtain wind power time series;
Step S20, carries out wavelet decomposition to wind power time series, is decomposed into low frequency component and high fdrequency component;
Step S30, carries out stationary test to the wind power time series of high fdrequency component and tranquilization processes;
Step S40, carries out determination and the parameter fitting of the exponent number of arma modeling to the wind power time series of high fdrequency component;
Step S50, is converted to arma modeling Kalman filter model and does rolling forward prediction, obtaining high frequency and predict the outcome;
Step S60, carries out slope to the wind power time series of low frequency component and rolls outside forecast forward, obtain low frequency and predict the outcome;
Step S70, predicts the outcome to the high frequency obtained in step S50 and step S60 and low frequency predicts the outcome and carries out wavelet reconstruction and obtain final predicting the outcome.
In step slo, described wind power time series can obtain to the historical statistical data of wind energy turbine set wind power in one-year age by collecting the first half, can select as required.Further, since wind field equipment, manual operation and wind speed itself are easily subject to the reasons such as the feature of physical factor impact, therefore need historical data is carried out pretreatment, to remove the impact of special situation.Such as, having saltus step the most suddenly in historical data is the phenomenon of zero point, and this is probably by artificially rationing the power supply or running into what accident caused, for the data of similar this situation, directly removes after judging in a program.It is appreciated that the acquisition methods of described wind power time series can need to select according to computational accuracy.
In step S20, described low frequency air quantity and high frequency air quantity can be obtained by described wind power time series is done wavelet transformation.After a signal is done wavelet transformation, generally can be broken down into approximation signal, i.e. low frequency signal and detail signal, i.e. high-frequency signal.Wherein, approximation signal has reacted the trend part of primary signal, and detail signal has then reacted the wave portion of primary signal, plateau.Therefore, after described wind power sequence is carried out wavelet transformation, the sequence of non-stationary can be decomposed into trend component and stable wave component, data decomposition can be come by wavelet transformation, different methods is used to be predicted targetedly according to respective feature different components, to obtain the result of higher precision.
When using wavelet transformation, selecting different basic functions ψ (t) can obtain different results, conventional wavelet basis function includes Haar wavelet basis function, db2 wavelet basis function, db4 wavelet basis function etc..In the present embodiment, use db4 wavelet basis function.Utilizing wavelet decomposition to process described wind power time series, if being broken down into dried layer, the criterion that concrete stopping is decomposed is: until the low frequency trend component of current layer number just can characterize the tendency information of original series.The concrete number of plies decomposed can be to utilize wavelet decomposition 5-6 time, i.e. obtains 6-7 subsequence (5-6 layer high-frequency data, one layer of low frequency tendency information).
In step s 30, described wind power time series needs to carry out stationary test and tranquilization processes.Owing to using ARMA method in the present embodiment, and the stationarity to data of setting up of arma modeling has required.
The definition of described " stationarity " is: to time series { rt, to arbitrary integer m, if rtAnd rt-mCovariance, rtAverage all do not change over, then claim time series { rtMeet weakly stationary.Specifically, if time series { rtIt is weakly stationary, need to meet following condition:
(1) expectation does not changes over time;
(2) variance does not changes over time;
(3) covariance of any two time point does not changes over time.
In order to judge whether described wind power time series has stationarity, it is possible to use autocorrelation coefficient method, i.e. sequence is done autocorrelation analysis, obtain autocorrelation coefficient curve, observe the decrease speed of autocorrelation coefficient curve, if curve declines rapidly, then show that original data sequence is stable;Be otherwise non-smoothly.
If the assay of described wind power sequence is non-stationary series, now need to carry out tranquilization process, by difference method, described wind power non-stationary original series can be converted to stationary sequence.Difference refers to that the data of original series adjacent time point are done difference obtains the process of new sequence, repeats that original series does the sequence obtained after difference n times and is referred to as N jump sub-sequence.Stable sequence is obtained after described wind power time series can be done 1-2 difference processing.
In step s 40, the wind Power x Time after described tranquilization processes needs the matching of determination and the parameter carrying out the exponent number of arma modeling.
ARMA time series models, are again autoregressive moving-average model, are to be combined, by autoregression model (AR model) and moving average model (MA model), the model produced.
Described ARMA (p, q) model can have a form of formula (1):
The exponent number of AR part during wherein p represents model, q represents the exponent number of MA part, γtAnd rt-iRepresent current and the data of history, φ respectivelyiFor constant term, for rt-iCoefficient, i.e. AR part coefficient, atAnd at-iRepresent the residual error of the data of corresponding time point, θ respectivelyiFor constant term, it is at-iCoefficient, i.e. MA part coefficient.When p or q is 0, arma modeling returns to the state of MA or AR model respectively.
Next judge the exponent number of the model of matching, i.e. determine p, q value.The method of judgment models exponent number can be AIC criterion, i.e. Akaike information criterion.Specific practice is the exponent number scope of the most general setting models, the AIC value that then in traversal computer capacity, the model of all exponent numbers is corresponding, and the exponent number of the corresponding model that AIC value is minimum is required.When model parameter estimation mode difference, the calculating standard of AIC value is different, mainly includes following two formulas:
During maximal possibility estimation: AIC=(n-d) log σ2+2(p+q+2)(2)
During least-squares estimation: AIC=(n-d) log σ2+(p+q+1)logn(3)
Wherein n is sample number, σ2For regression criterion quadratic sum, d, p, q are parameter.
After model order determines, coefficient number unknown in model and structure just have determined that, then determine the coefficient of these the unknowns according to historical data.In the present embodiment, determine that the method for unknowm coefficient, for solving system of linear equations, uses least square method.After solving all coefficients, complete the foundation of arma modeling.
In step s 50, the forecast model of described high fdrequency component uses usual linear discrete Kalman filtering system comprise state equation and measure equation, respectively as shown in (4), (5).
Z (k+1)=H (k+1) X (k+1)+v (k+1) (5)
In two formulas, X (k) represents state vector, and ω (k) represents interference vector, and Z (k) represents observation vector, and v (k) represents noise vector,Γ (k+1, k), H (k+1) be constant coefficient matrix.The Discrete stochastic systems determined by (4), (5) formula can represent with Fig. 3.
Further, in order to obtain optimized Kalman prediction model, it is possible to use mathematical induction and orthogonality theorem are derived below equation by (4), (5) two formulas:
K=P (k+1 | k) HT[HP(k+1|k)HT+R]-1(7)
P (k+1 | k+1)=[I-KH] P (k+1 | k) (9)
In formulaRepresent the best estimate in k moment, K represents Kalman gain matrix, P (k+1 | k) represents the error co-variance matrix of the Single-step Prediction from k to the k+1 moment, P (k+1 | k+1) represents the error co-variance matrix of the filter forecasting in k+1 moment, Q is the covariance matrix of ω (k), R is the covariance matrix of v (k+1), and I is unit matrix.Formula (6)~(9) are followed successively by optimal filter respectively and estimate equation, optimum gain matrix equation, Single-step Prediction error covariance equation and filter forecasting error covariance equation.By given correspondenceP (k | k), the initial value of R, Q, can be iterated, according to (6)~(9) four equations, the prediction that rolls forward.(5)~in (9)Γ (k+1, k), H (k+1) be constant coefficient matrix, can be drawn by historical data.
See also Fig. 4, for the block diagram of the Discrete Linear Random System after optimization.
To described Kalman filtering optimization forecast model, it is determined by 3 constant coefficient matrixesГ (k+1, k), H (k+1), and 4 initial valuesP (k | k), R, Q, can be iterated the prediction rolled forward.
For wind power time series is predicted, arma modeling itself has model parameter to determine easily, the advantage of method maturation specification.In contrast, when Kalman filter model is individually predicted, the determination of model parameter is relatively difficult, but has an advantage in that, there is degenerative process in model, it is possible to real-time is adjusted optimal estimation value.Therefore, arma modeling and Kalman filter model just define excellent complementary relationship, both combinations are predicted, beneficially improve the precision predicted the outcome.
As specific embodiment, both concrete associated methods following (to illustrate as a example by ARMA (2,3)):
(1) model form first writing out ARMA (2,3) is:
X (t)=φ0+φ1X(t-1)+φ2X(t-2)+θ1a(t-1)+θ2a(t-2)+θ3a(t-3)+a(t)(10)
In order to be the matrix in Kalman filter model and the form of vector by the parameters in (10) formula and transformation of data, now do with down conversion:
(2) after carrying out the conversion in (1), according to the calculated relationship of (10), in the case of ignoring residual error a (t), there is a following relation:
(5-16) state equation during formula is Kalman filter model, foolproof can see coefficient matrix from (11) formula hereinΓ (k+1, value k).
(3) determinedГ (k+1, after value k), it is considered to measure equation, from Middle taking-up X (t) is exactly optimal direct observation plus the most uncared-for residual error a (t), therefore takes v (k+1)=a (t) and obtains (5-17) formula:
The most also the value of H-matrix has been determined that.
Arrive this, in Kalman filter model all of coefficient matrix all it has been determined that, it is only necessary to select suitable initial value according to practical situation, can bring equation (6) into~(9) are iterated the prediction that rolls forward, obtain a result.
In small wave converting method described in step S20, wind power time series high fdrequency component and low frequency component are resolved into.Its high frequency components (wave component) the most steadily, is suitable for ARMA_Kalman model.Again by above-mentioned iteration rolling forecast, the result that high fdrequency component is predicted can be drawn.
In step S60, the low frequency component (trend component) that described wind power time series is obtained after wavelet transformation.To described low frequency trend component, in a suitable short-term time scale, it is all the regular change dull according to almost fixing slope, therefore for the prediction of this part of component, can directly use and predict according to the method for slope extrapolated back, owing to short term predicted data has similar tendency information with measured data, therefore by the trend component in short-term forecast curve, flex point can be judged during extrapolation, prevent the error of prediction.
In step S70, by the method for wavelet reconstruction, predicting the outcome of the high fdrequency component described in step S50 and the predicting the outcome of low frequency component described in step S60 comprehensively being obtained final predicting the outcome, concrete grammar is as follows:
In order to each data component reduction that decomposition is come, the method that wavelet reconstruction can be used.High fdrequency component described in S50 being predicted the outcome and the low frequency component described in step S60 predicts the outcome and carries out wavelet reconstruction, can obtain last predicting the outcome, can use Mallat algorithm, this algorithm is that wavelet transformation technical field is known.
Kalman filter model is combined by wind power prediction model of the present invention with arma modeling, adds unidirectional negative feedback links in Discrete stochastic systems, is conducive to improving the precision of prediction of arma modeling;Arma modeling model parameter calculation is simple, can provide parameter information for Kalman filter model efficiently again.Original series is decomposed into different components by utilizing wavelet analysis method by this method, feature according to different components selects Forecasting Methodology prediction targetedly, last finally being predicted the outcome by wavelet reconstruction can effectively improve the precision of model prediction, reduce error, than conventional method more accurately comprehensively.
It addition, those skilled in the art also can make other change in spirit of the present invention, all should be included in scope of the present invention.
Claims (8)
1. a Forecasting Methodology for wind power time series, comprises the following steps:
Obtain the historical statistical data of wind energy turbine set wind power, and history abnormal data is processed, obtain wind power time series;
Wind power time series is carried out wavelet decomposition, is decomposed into low frequency component and high fdrequency component;
The wind power time series of high fdrequency component is carried out stationary test and tranquilization processes;
The wind power time series of high fdrequency component is carried out determination and the parameter fitting of the exponent number of arma modeling;
Arma modeling be converted to Kalman filter model and do rolling forward prediction, obtaining high frequency and predict the outcome;
The wind power time series of low frequency component is carried out slope and rolls outside forecast forward, obtain low frequency and predict the outcome;And
The high frequency obtained is predicted the outcome and low frequency predicts the outcome and obtained final predicting the outcome by wavelet reconstruction.
2. the Forecasting Methodology of wind power time series as claimed in claim 1, it is characterised in that the wavelet basis function that wind power time series carries out wavelet decomposition employing is the one in Haar wavelet basis function, db2 wavelet basis function, db4 wavelet basis function.
3. the Forecasting Methodology of wind power time series as claimed in claim 1, it is characterised in that described arma modeling be expressed as ARMA (p, q), form is:
The exponent number of AR part during wherein p represents model, q represents the exponent number of MA part, rtAnd rt-iRepresent the current and data of history respectively,For constant term, for rt-iCoefficient, i.e. AR part coefficient, atAnd at-iRepresent the residual error of the data of corresponding time point, θ respectivelyiFor constant term, it is at-iCoefficient, i.e. MA part coefficient;When p or q is 0, arma modeling returns to the state of MA or AR model respectively.
4. the Forecasting Methodology of wind power time series as claimed in claim 3, it is characterized in that, Akaike information criterion AIC is utilized to determine p, q value, the general exponent number scope of first setting models, then the AIC value that in traversal computer capacity, the model of all exponent numbers is corresponding, the exponent number of the corresponding model that AIC value is minimum is required.
5. the Forecasting Methodology of wind power time series as claimed in claim 4, it is characterised in that the calculating standard of AIC value includes following two formulas:
During maximal possibility estimation: AIC=(n-d) log σ2+2(p+q+2);
During least-squares estimation:
AIC=(n-d) log σ2+(p+q+1)logn;
Wherein n is sample number, σ2For regression criterion quadratic sum, d, p, q are parameter.
6. the Forecasting Methodology of wind power time series as claimed in claim 1, it is characterised in that described high frequency predicts the outcome and uses usual linear discrete Kalman filtering system, comprises state equation and measures equation:
Z (k+1)=H (k+1) X (k+1)+v (k+1);
In two formulas, X (k) represents state vector, and ω (k) represents interference vector, and Z (k) represents observation vector, and v (k) represents noise vector,
Г (k+1, k), H (k+1) be constant coefficient matrix.
7. the Forecasting Methodology of wind power time series as claimed in claim 6, it is characterised in that utilize mathematical induction and orthogonality theorem to derive state equation and measurement equation:
K=P (k+1 | k) HT[HP(k+1|k)HT+R]-1;
P(k|1|k|1)-[IKH]P(k|1|k);
In formulaRepresent the best estimate in k moment, K represents Kalman gain matrix, P (k+1 | k) represents the error co-variance matrix of the Single-step Prediction from k to the k+1 moment, P (k+1 | k+1) represents the error co-variance matrix of the filter forecasting in k+1 moment, Q is the covariance matrix of ω (k), R is the covariance matrix of v (k+1), and I is unit matrix.
8. the Forecasting Methodology of wind power time series as claimed in claim 1, it is characterised in that use Mallat algorithm described high fdrequency component to be predicted the outcome and walks predicting the outcome of low frequency component and carry out wavelet reconstruction.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510956093.XA CN105825289A (en) | 2015-12-17 | 2015-12-17 | Prediction method for wind power time sequence |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510956093.XA CN105825289A (en) | 2015-12-17 | 2015-12-17 | Prediction method for wind power time sequence |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105825289A true CN105825289A (en) | 2016-08-03 |
Family
ID=56514666
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510956093.XA Pending CN105825289A (en) | 2015-12-17 | 2015-12-17 | Prediction method for wind power time sequence |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105825289A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107967542A (en) * | 2017-12-21 | 2018-04-27 | 国网浙江省电力公司丽水供电公司 | A kind of electricity sales amount Forecasting Methodology based on shot and long term memory network |
CN109214575A (en) * | 2018-09-12 | 2019-01-15 | 河海大学 | A kind of super short-period wind power prediction technique based on small wavelength short-term memory network |
CN109904878A (en) * | 2019-02-28 | 2019-06-18 | 西安交通大学 | A kind of windy electric field electricity-generating timing simulation scenario building method |
CN111262873A (en) * | 2020-01-20 | 2020-06-09 | 安徽大学 | User login feature prediction method and device based on wavelet decomposition |
CN111814101A (en) * | 2020-07-10 | 2020-10-23 | 北京无线电测量研究所 | Flight path prediction method and system and electronic equipment |
CN113591382A (en) * | 2021-08-02 | 2021-11-02 | 东北电力大学 | Ultra-short-term rolling prediction method based on WT-TCN wind power |
CN114580192A (en) * | 2022-03-15 | 2022-06-03 | 北京博数智源人工智能科技有限公司 | Method and device for calculating cut-in wind speed and rated wind speed of fan |
CN116930973A (en) * | 2023-09-15 | 2023-10-24 | 湖南赛能环测科技有限公司 | Scene-adaptive sodar wind power accurate prediction method |
CN117725543A (en) * | 2024-02-18 | 2024-03-19 | 中国民航大学 | Multi-element time sequence anomaly prediction method, electronic equipment and storage medium |
CN117803946A (en) * | 2024-02-29 | 2024-04-02 | 山西漳电科学技术研究院(有限公司) | Safety control system and method for low-calorific-value coal burning boiler |
-
2015
- 2015-12-17 CN CN201510956093.XA patent/CN105825289A/en active Pending
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107967542B (en) * | 2017-12-21 | 2021-07-27 | 国网浙江省电力公司丽水供电公司 | Long-short term memory network-based electricity sales amount prediction method |
CN107967542A (en) * | 2017-12-21 | 2018-04-27 | 国网浙江省电力公司丽水供电公司 | A kind of electricity sales amount Forecasting Methodology based on shot and long term memory network |
CN109214575A (en) * | 2018-09-12 | 2019-01-15 | 河海大学 | A kind of super short-period wind power prediction technique based on small wavelength short-term memory network |
CN109214575B (en) * | 2018-09-12 | 2021-08-31 | 河海大学 | Ultrashort-term wind power prediction method based on small-wavelength short-term memory network |
CN109904878A (en) * | 2019-02-28 | 2019-06-18 | 西安交通大学 | A kind of windy electric field electricity-generating timing simulation scenario building method |
CN109904878B (en) * | 2019-02-28 | 2020-07-28 | 西安交通大学 | Multi-wind-field power generation time sequence simulation scene construction method |
CN111262873B (en) * | 2020-01-20 | 2022-06-03 | 安徽大学 | User login feature prediction method and device based on wavelet decomposition |
CN111262873A (en) * | 2020-01-20 | 2020-06-09 | 安徽大学 | User login feature prediction method and device based on wavelet decomposition |
CN111814101A (en) * | 2020-07-10 | 2020-10-23 | 北京无线电测量研究所 | Flight path prediction method and system and electronic equipment |
CN113591382A (en) * | 2021-08-02 | 2021-11-02 | 东北电力大学 | Ultra-short-term rolling prediction method based on WT-TCN wind power |
CN113591382B (en) * | 2021-08-02 | 2023-08-04 | 东北电力大学 | Ultra-short-term rolling prediction method based on WT-TCN wind power |
CN114580192A (en) * | 2022-03-15 | 2022-06-03 | 北京博数智源人工智能科技有限公司 | Method and device for calculating cut-in wind speed and rated wind speed of fan |
CN116930973A (en) * | 2023-09-15 | 2023-10-24 | 湖南赛能环测科技有限公司 | Scene-adaptive sodar wind power accurate prediction method |
CN116930973B (en) * | 2023-09-15 | 2023-12-08 | 湖南赛能环测科技有限公司 | Scene-adaptive sodar wind power accurate prediction method |
CN117725543A (en) * | 2024-02-18 | 2024-03-19 | 中国民航大学 | Multi-element time sequence anomaly prediction method, electronic equipment and storage medium |
CN117725543B (en) * | 2024-02-18 | 2024-05-03 | 中国民航大学 | Multi-element time sequence anomaly prediction method, electronic equipment and storage medium |
CN117803946A (en) * | 2024-02-29 | 2024-04-02 | 山西漳电科学技术研究院(有限公司) | Safety control system and method for low-calorific-value coal burning boiler |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105825289A (en) | Prediction method for wind power time sequence | |
CN103279804B (en) | The Forecasting Methodology of super short-period wind power | |
Paternina et al. | Identification of electromechanical oscillatory modes based on variational mode decomposition | |
CN103577710B (en) | Aviation Power Converter faults Forecasting Methodology based on fractional order UPF | |
CN102867132B (en) | The online fault combination forecasting method of aviation DC converter based on fractional wavelet transform | |
CN110309603B (en) | Short-term wind speed prediction method and system based on wind speed characteristics | |
CN111191191B (en) | Construction method of combined model for accurately predicting deformation effect of concrete dam | |
CN111582551B (en) | Wind power plant short-term wind speed prediction method and system and electronic equipment | |
CN102880810B (en) | Wind power prediction method based on time sequence and neural network method | |
CN104573876A (en) | Wind power plant short-period wind speed prediction method based on time sequence long memory model | |
CN107704953A (en) | The short-term wind-electricity power probability density Forecasting Methodology of EWT quantile estimate forests | |
CN108717579B (en) | Short-term wind power interval prediction method | |
CN107358060A (en) | A kind of method estimated wind power prediction error burst based on HMM | |
Ling-ling et al. | The use of wavelet theory and ARMA model in wind speed prediction | |
CN106384170A (en) | Wavelet decomposition and reconstruction-based time sequence wind speed prediction method | |
CN105354636A (en) | Wind power fluctuation probability density modeling method based on nonparametric kernel density estimation | |
Lv et al. | Short-term wind speed forecasting based on non-stationary time series analysis and ARCH model | |
CN104201700A (en) | Regional power grid thermal power frequency modulation unit configuration method considering wind power uncertainty fluctuation | |
CN104462861A (en) | Reservoir regulation decision-making method based on reservoir regulation rule synthesis | |
CN106655190A (en) | Method for solving P-OPF (Probabilistic-Optimal Power Flow) of wind power stations | |
CN106447063A (en) | Combined prediction method for short-period wind speed of grid-connected wind power station | |
CN108074015A (en) | A kind of ultrashort-term wind power prediction method and system | |
CN110942170A (en) | Short-term wind speed prediction method and system based on information processing | |
CN104537233B (en) | A kind of power distribution network puppet based on Density Estimator measures generation method | |
CN104951654A (en) | Method for evaluating reliability of large-scale wind power plant based on control variable sampling |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160803 |
|
RJ01 | Rejection of invention patent application after publication |