CN104820869B - A kind of wind power climbing event prediction method of the mechanism containing models switching - Google Patents
A kind of wind power climbing event prediction method of the mechanism containing models switching Download PDFInfo
- Publication number
- CN104820869B CN104820869B CN201510219455.7A CN201510219455A CN104820869B CN 104820869 B CN104820869 B CN 104820869B CN 201510219455 A CN201510219455 A CN 201510219455A CN 104820869 B CN104820869 B CN 104820869B
- Authority
- CN
- China
- Prior art keywords
- prediction
- model
- wind power
- climbing
- event
- 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.)
- Active
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The present invention relates to a kind of wind power forecasting methods, more particularly, to a kind of wind power climbing event prediction method of handover mechanism containing prediction model.Method is adhered in the climbing of wind power forecasting method of this method based on the mechanism containing models switching and consideration operation of power networks state, mainly includes the following steps: step 1, completes the pretreatment of wind power data and seeking for partial model prove-in length；Step 2 completes the identification of optimum prediction model on the basis of local segment divides, and the training of models switching mechanism is completed according to recognition result；Step 3 completes long-term wind power prediction in conjunction with forecasting model database and models switching mechanism；Step 4 carries out climbing event differentiation to the wind power result of prediction in conjunction with the definition of climbing event and power grid actual motion state.Method provided by the invention guarantees the accuracy of climbing event prediction by the wind power prediction of degree of precision, formulates correctly climbing control measure for electric system and provides reliable guarantee.
Description
Technical field
The invention patent relates to a kind of wind power forecasting methods, more particularly, to a kind of handover mechanism containing prediction model
Wind power climbing event prediction method.
Background technique
With getting worse for energy crisis, all renewable energy is being greatly developed in worldwide.Currently, wind-powered electricity generation is made
For a kind of abundant, resource that can be developed and used, stepped up in the permeability of power grid.However since wind energy is by air motion
Influence have randomness and fluctuation, especially the Wind Power Development mode in extensive, high concentration degree, in the short time substantially
It is huge to power grid harm to be worth wind-powered electricity generation fluctuation, to climb event most very.In order to instruct traffic department to formulate effective power generation meter in advance
It draws, the wind-powered electricity generation event of the significant threats such as prevention climbing, and takes control hazard mitigation measure immediately, guarantee the safe, steady of electric system
Fixed operation, it is particularly significant to the look-ahead of wind power climbing event.
Foreign scholar is more early to the research starting of wind power climbing event.One was had occurred in texas,U.S due to 2008
Field is extensive, endangers biggish power descending event, causes the electric reliability committee (Electric Reliability
Council of Texas, ERCOT) serious economic loss, therefore the problem is obtained extensive attention.At present to wind-powered electricity generation function
The research of rate climbing event is concentrated mainly on the definition, classification and forecasting research of climbing event.Wherein there are still not for forecasting research
Foot, the method mainly used is that climbing detection and differentiation is carried out on the basis of wind power prediction, therefore for event of climbing
Definition research be carry out climbing event prediction premise.Currently used several numerical value definition, relate generally to climbing event
Change the important features such as amplitude, duration and climbing rate.Doctor Sevlian of Stanford University proposes to use on this basis
The climbing incident Detection Algorithm of marking mode.And the purpose for event category of climbing is to determine climbing thing according to climbing prediction result
Part type, thus selection suitable effectively control, control measures.
Climbing event is defined in the changed power event in a period of time.In order to realize that the prediction needs of climbing event are examined
Problem of both considering, first is that the realization of prolonged event prediction, second is that how effectively to extract wind-powered electricity generation from prediction result
Power climbing affair character amount.It mainly include at present two class of physical model and statistical model about wind power method, wherein object
Reason model can realize longer prediction, but precision of prediction is not high；And statistical model has short-term wind power prediction
Higher precision of prediction, but error can increase with the increase of prediction duration.In order to solve the forecasting problem of climbing event, need
The power prediction result of long period is wanted to meet the requirement of climbing event prediction, while to guarantee effectively to capture climbing event
The precision of prediction of feature, wind power prediction need to be guaranteed, to guarantee to effectively realize climbing event prediction research.
Summary of the invention
The purpose of the invention patent is to propose that one kind can be by long-term wind power prediction by climbing for above-mentioned status
The demand of event is converted into continuous local wind power prediction, and more high-precision by choosing optimal Short-term Forecasting Model completion
The wind power prediction of degree completes the new method of wind power climbing event prediction in conjunction with climbing definition and electric network state.
What above-mentioned technical problem of the invention was achieved through the following technical solutions:
A kind of wind power climbing event prediction method of the handover mechanism containing prediction model, it is characterised in that: including following
Step,
Step 1: using the wind power data of history as research object, detecting the historied climbing event of institute, and to it
Duration is counted, and time window being applicable in as local increment that can satisfy most climbing durations is found out
Duration；Meanwhile initial data is analyzed, it completes to the work such as data prediction and phase space reconfiguration, including following sub-step
Suddenly,
Step 1.1, the processing such as bad value, missing value is carried out to original wind power data sequence first, to guarantee data
Completeness and efficiency, while unnecessary noise is eliminated by filtering processing, obtain clean sequence；On the other hand, pass through
To the statistical analysis of history climbing incident duration, the applicable duration of local increment is provided；
Step 1.2, in order to facilitate the foundation of prediction model, Space Reconstruction is carried out to data first；If original data sequence
For { xn, then take two observed quantity x of interval time interval τnAnd xn+τ, association relationship is calculated by formula one, takes Mutual information entropy the
Delay time of one minimum point as reconstruct；
Formula one
Step 1.3, it is assumed that the reconstruct Embedded dimensions of original series are m, then calculate phase according to formula two to the sequence after reconstruct
Distance between point takes m as the Embedded dimensions of reconstruct if meeting formula three, four；
Formula two
Formula three
Formula four
Step 1.4, according to the delay time and Embedded dimensions found out in step 1.2 and step 1.3, sequence is completed according to formula five
The phase space reconfiguration of column；Using the data after reconstructing as object, the partial model prove-in length that is acquired according to step 1.1 is by history
Data are divided into continuous event section one by one, to be studied when carrying out model training and prediction as basic unit；
xn=(xn,xn+τ,...xn+(m-1)τ)∈Rm, n=1,2..., N0=N- (m-1) τ formula five
Step 2: based on multivariable chaotic prediction model, in order to realize longer power prediction, in conjunction with number
The meteorological data that value weather forecast provides establishes the chaotic prediction model of multivariable；The physics of different climbing events is considered simultaneously
Background is different, and to further increase the precision of prediction of model, different kernel function structures are added on the basis of Classical forecast model
At forecasting model database, complete to identify the model of historical events section according to given model library, it is pre- that training obtains optimal part
Survey model；Including following sub-step,
Step 2.1, according to the Local-region Linear Prediction model of traditional chaos time sequence, Nonlinear Mapping, construction are introduced
The fundamental forecasting model as shown in formula six；
Formula six
In formula, Φ indicates nonlinear transformation, and e indicates residual error item；
Step 2.2, in order to reflect under different physical backgrounds climb event changed power particularity, to nonlinear transformation Φ
It is indicated using different kernel functions, chooses 3 kinds of common kernel functions: Polynomial kernel function, Radial basis kernel function and Sigmoid core
Function, such as formula seven, to constitute the model library of local increment；
Formula seven
Step 3: according to model library obtained above, model training being carried out to each independent event section respectively, obtained every
Optimum prediction model in a section, is then trained according to the corresponding relationship of initial data and optimal models, difference can be obtained
The handover mechanism of prediction model；
Step 4: the models switching mechanism in conjunction with the model library in step 2 and in step 3 completes longer wind power
Prediction prediction；The climbing event detection for completing prediction data is finally defined according to the numerical value of climbing event, it is contemplated that practical hair is not
It is related with the state of power grid that climbing harm occurs, therefore climb to wind power further combined with the completion of the actual motion state of power grid
The prediction and judgement of slope event；
Formula 11.
In the wind power climbing event prediction method of the above-mentioned handover mechanism containing prediction model, the step 3 is mainly wrapped
Following sub-step is included,
Step 3.1, the model in step 2 in model library is numbered, when carrying out model training to continuous events section,
Numbered sequence { the c of optimal models can be obtainedn, in conjunction with original event segment data, if the training set of formula eight can be established；
Formula eight
Step 3.2, above-mentioned training set can be regarded as containing multi-class categorized data set, and one of classification indicates a kind of pre-
Model is surveyed, multi-class support vector cassification model training classifier can be established to this；By method of classifying, for containing k
The other data of type need to establish k-1 basic SVM models, such as following formula:
s.t.yi(<ω·xi>+b)≥1-ξi
ξi>=0, i=1,2...N0Formula nine
In formula, ω, b are classifier parameters；ξ is slack variable；ynThe class variable redefined is indicated, when a certain classification
When for main study subject, then yn=1, remaining class label yn=-1；3 classification in step 2 can be obtained by the solution of SVM model
The classifier of model library, the classifier are the handover mechanism of local increment.
In the wind power climbing event prediction method of the above-mentioned handover mechanism containing prediction model, the step 4 is mainly wrapped
Following sub-step is included,
Step 4.1, obtained in the step 1 as unit of partial model prove-in length, in conjunction with numerical weather forecast as a result,
Complete the local wind power prediction in the short time；Simultaneously according to models switching mechanism, different event section prediction model is carried out
Switching, the prediction of comprehensive multiple continuous events sections, completes long-term wind power prediction；
Step 4.2, it detects to meet in prediction result and climb such as formula ten according to given wind power climbing event definition
The climbing event that slope numerical value defines；Operation of power networks shape is being added in situations such as considering the plan of power grid actual schedule, such as cutting machine-cut load
On the basis of state, detects the climbing event with harmfulness, be scheduled and control so as to guide electric system；
Formula ten
This definition thinks the Error Absolute Value and time interval Δ of the wind power amplitude at two moment for being separated by length Δ t
The ratio of t is greater than a certain given maximum power climbing rate Rval, then a referred to as wind power climbing.
Therefore, the present invention can reach it is following the utility model has the advantages that 1, compared with traditional prediction method, due to climb event spy
Different property, during considering event prediction, the structure feature of present invention combination climbing event is proposed as unit of event section
Prediction technique.Simultaneously by the method for the wind power variation analysis climbing event in predicted events section, by event of climbing
Predictive conversion solved at traditional data point prediction；2, it during realizing wind power long-term forecast, not only examines
The short-term forecast precision for having considered statistical model is high, while the advantages of Integrative Logistic System Model practicable long-term forecast, mentioning in conjunction with the two
Out the long-term mechanism containing models switching prediction technique, it can be achieved that degree of precision wind power prediction；3, in order to improve to electricity
Force system is effectively climbed event information, and the present invention not only needs the numerical value in conjunction with climbing event to define, while considering actual electric network
Operating status, it is comprehensive to provide following a possibility that climbing event occurs.
Detailed description of the invention
Fig. 1 is the wind power climbing prediction Integral Thought figure of the mechanism provided by the invention containing models switching.
Fig. 2 a is history provided by the invention climbing incident duration probability distribution graph.
Fig. 2 b is history provided by the invention climbing event accumulated time probability distribution graph.
Fig. 3 is the long-term wind power prediction result figure of the mechanism provided by the invention containing models switching.
Fig. 4 is wind power provided by the invention climbing event prediction result figure.
Specific embodiment
Below with reference to the embodiments and with reference to the accompanying drawing further description of the technical solution of the present invention.
The present invention relates to a kind of wind power of handover mechanism containing prediction model climbing prediction techniques.Since wind power is climbed
Slope event brings serious threat to the safety and stability of electric system, climbs to prevent in advance and effectively control harmfulness
Slope event, it is particularly significant to the look-ahead of climbing event.
Embodiment
Technical solution of the present invention mainly includes two parts, that is, the prediction technique of the mechanism containing models switching is utilized to complete to wind
The long-term forecast of electrical power and combination climbing definition, operation of power networks state carry out the prediction of climbing event.
One, the principle of the present invention is introduced first.
Step 1: using the wind power data of history as research object, detecting the historied climbing event of institute, and to it
Duration is counted, and time window being applicable in as local increment that can satisfy most climbing durations is found out
Duration.Meanwhile initial data is analyzed, it completes to work such as data prediction and phase space reconfigurations.
During carrying out historical data analysis, above-mentioned steps 1 mainly include following sub-step,
Step 1.1, the processing such as bad value, missing value is carried out to original wind power data sequence first, to guarantee data
Completeness and efficiency, while unnecessary noise is eliminated by filtering processing, obtain clean sequence.On the other hand, pass through
To the statistical analysis of history climbing incident duration, the applicable duration of local increment is provided.
Step 1.2, in order to facilitate the foundation of prediction model, Space Reconstruction is carried out to data first.If original data sequence
For { xn, then take two observed quantity x of interval time interval τnAnd xn+τ, association relationship is calculated by formula (1), takes Mutual information entropy the
Delay time of one minimum point as reconstruct.
Step 1.3, it is assumed that the reconstruct Embedded dimensions of original series are m, then calculate phase according to formula (2) to the sequence after reconstruct
Distance between point takes m as the Embedded dimensions of reconstruct if meeting formula (3-4).
Step 1.4, it according to the delay time and Embedded dimensions found out in step 1.2 and step 1.3, is completed according to formula (5)
The phase space reconfiguration of sequence.Using the data after reconstructing as object, the partial model prove-in length that is acquired according to step 1.1 is by history
Data be divided into continuous event section one by one, so as to carry out model training and prediction when to be ground as basic unit
Study carefully.
xn=(xn,xn+τ,...xn+(m-1)τ)∈Rm, n=1,2..., N0=N- (m-1) τ (5)
Step 2: based on multivariable chaotic prediction model, in order to realize longer power prediction, in conjunction with number
The meteorological data that value weather forecast provides establishes the chaotic prediction model of multivariable.The physics of different climbing events is considered simultaneously
Background is different, and to further increase the precision of prediction of model, different kernel function structures are added on the basis of Classical forecast model
At forecasting model database, complete to identify the model of historical events section according to given model library, it is pre- that training obtains optimal part
Survey model.
In the establishment process of prediction model, the step 2 includes following sub-step,
Step 2.1, according to the Local-region Linear Prediction model of traditional chaos time sequence, Nonlinear Mapping, construction are introduced
The fundamental forecasting model as shown in formula (6).
In formula, Φ indicates nonlinear transformation, and e indicates residual error item.
Step 2.2, in order to reflect under different physical backgrounds climb event changed power particularity, to nonlinear transformation Φ
Indicated using different kernel function, choose 3 kinds of common kernel functions: Polynomial kernel function, Radial basis kernel function (RBF) and
Sigmoid kernel function, such as formula (7), to constitute the model library of local increment.
Step 3: according to model library obtained above, model training being carried out to each independent event section respectively, obtained every
Optimum prediction model in a section, is then trained according to the corresponding relationship of initial data and optimal models, difference can be obtained
The handover mechanism of prediction model.
In the finding process of handover mechanism, above-mentioned steps 3 mainly include following sub-step,
Step 3.1, the model in step 2 in model library is numbered, when carrying out model training to continuous events section,
Numbered sequence { the c of optimal models can be obtainedn, in conjunction with original event segment data, if the training set of formula (8) can be established.
Step 3.2, above-mentioned training set can be regarded as containing multi-class categorized data set, and one of classification indicates a kind of pre-
Model is surveyed, multi-class support vector machines (SVM) disaggregated model training classifier can be established to this.It is right by method of classifying
K-1 basic SVM models need to be established in the other data of type containing k, such as following formula:
s.t.yi(<ω·xi>+b)≥1-ξi
ξi>=0, i=1,2...N0 (9)
In formula, ω, b are classifier parameters；ξ is slack variable；ynThe class variable redefined is indicated, when a certain classification
When for main study subject, then yn=1, remaining class label yn=-1.3 classification in step 2 can be obtained by the solution of SVM model
The classifier of model library, the classifier are the handover mechanism of local increment.
Step 4: the models switching mechanism in conjunction with the model library in step 2 and in step 3 completes longer wind power
Prediction prediction.The climbing event detection for completing prediction data is finally defined according to the numerical value of climbing event, it is contemplated that practical hair is not
It is related with the state of power grid that climbing harm occurs, therefore climb to wind power further combined with the completion of the actual motion state of power grid
The prediction and judgement of slope event.
During above-mentioned wind power climbs prediction, step 4 mainly includes following sub-step,
Step 4.1, obtained in the step 1 as unit of partial model prove-in length, in conjunction with numerical weather forecast as a result,
Complete the local wind power prediction in the short time.Simultaneously according to models switching mechanism, different event section prediction model is carried out
Switching, the prediction of comprehensive multiple continuous events sections, completes long-term wind power prediction.
Step 4.2, it according to given wind power climbing event definition, such as formula (10), detects to meet in prediction result
The climbing event that climbing numerical value defines.Operation of power networks is being added in situations such as considering the plan of power grid actual schedule, such as cutting machine-cut load
On the basis of state, detects the climbing event with harmfulness, be scheduled and control so as to guide electric system.
This definition thinks the Error Absolute Value and time interval Δ of the wind power amplitude at two moment for being separated by length Δ t
The ratio of t is greater than a certain given maximum power climbing rate Rval, then a referred to as wind power climbing.
Two, choosing the control area U.S. BPA annual wind power output data in 2013 below is sample set, wherein wind-powered electricity generation function
The temporal resolution of rate data is every 5 minutes sampled points, in conjunction with attached drawing and case study on implementation the present invention will be described in detail technical side
Case.
Using original wind power data as object, the pretreatment work of complete paired data, such as bad value, missing value and denoising etc.
Reason then to clean effective selection for carrying out completing Parameters for Phase Space Reconstruction according to formula (1-4), and completes phase space reconfiguration.
On the other hand, the climbing event occurred in historical process is detected, and for statistical analysis to time of climb, such as attached drawing 2.
Take the time window of the time of climb requirement of satisfaction 95% as local increment prove-in length.
According to traditional nonlinear prediction model of chaotic prediction model foundation, and take 3 kinds of typical kernel functions as non-thread
Property transformation base configuration model library.Initial data is divided according to model prove-in length, obtains continuous event Duan Xu
Column, identify optimum prediction model to each data segment.Regard the corresponding relationship of prediction model and data as assorting process, then
The handover mechanism of model can be obtained by SVM model training.Collective model library and models switching mechanism carry out test set data
Prediction, as a result as shown in Fig. 3.
According to the prediction result of wind power, foundation is defined as with the climbing of formula (10), due to total appearance of sample set data
Amount is 4000MW, then the threshold value for taking the changed power of 50% installed capacity in 4 hours to determine as climbing is completed to climbing event
Preliminary detection.In order to adapt to the demand of practical power systems, consider that operation of power networks state is further completed to sentence climbing event
It is disconnected, to complete the prediction to wind power climbing event.
The above embodiments are merely illustrative of the technical scheme of the present invention and are not intended to be limiting thereof, the technical field of the invention
Technical staff can do various modifications or compensation to the specific example example of description is said, but without departing from appended claims of the present invention
Defined range.
Claims (3)
- The event prediction method 1. a kind of wind power of handover mechanism containing prediction model is climbed, it is characterised in that: including following step Suddenly,Step 1: using the wind power data of history as research object, detecting the historied climbing event of institute, and continue to it Time is counted, and where applicable of the time window that can satisfy most climbing durations as local increment is found out It is long；Meanwhile initial data is analyzed, it completes to work to data prediction and phase space reconfiguration, including following sub-step,Step 1.1, bad value, missing value processing are carried out to original wind power data sequence first, to guarantee the integrality of data And validity, while unnecessary noise is eliminated by filtering processing, obtain clean sequence；On the other hand, by history The statistical analysis of climbing incident duration, provides the applicable duration of local increment；Step 1.2, in order to facilitate the foundation of prediction model, Space Reconstruction is carried out to data first；If original data sequence is {xn, then take two observed quantity x of interval time interval τnAnd xn+τ, association relationship is calculated by formula one, takes Mutual information entropy first Delay time of a minimum point as reconstruct；Step 1.3, it is assumed that the reconstruct Embedded dimensions of original series are m, then calculating phase point according to formula two to the sequence after reconstruct Distance take m as the Embedded dimensions of reconstruct if meeting formula three to formula four；Step 1.4, according to the delay time and Embedded dimensions found out in step 1.2 and step 1.3, sequence is completed according to formula five Phase space reconfiguration；Using the data after reconstructing as object, the partial model prove-in length that is acquired according to step 1.1 is by the data of history It is divided into continuous event section one by one, to be studied when carrying out model training and prediction as basic unit；xn=(xn,xn+τ,…xn+(m-1)τ)∈Rm, n=1,2 ..., N0=N- (m-1) τ formula fiveStep 2: based on multivariable chaotic prediction model, in order to realize longer power prediction, in conjunction with numerical value day The meteorological data that gas forecast provides establishes the chaotic prediction model of multivariable；The physical background of different climbing events is considered simultaneously It is different, and to further increase the precision of prediction of model, different kernel functions are added on the basis of Classical forecast model and constitute in advance Model library is surveyed, is completed to identify that training obtains optimal local prediction mould to the model of historical events section according to given model library Type；Including following sub-step,Step 2.1, according to the Local-region Linear Prediction model of traditional chaos time sequence, Nonlinear Mapping is introduced, is constructed such as formula Fundamental forecasting model shown in six；In formula, Φ indicates nonlinear transformation, and e indicates residual error item；Step 2.2, in order to reflect under different physical backgrounds climb event changed power particularity, to nonlinear transformation Φ use Different kernel functions indicates, chooses 3 kinds of common kernel functions: Polynomial kernel function, Radial basis kernel function and Sigmoid core letter Number, such as formula seven, to constitute the model library of local increment；Step 3: according to model library obtained above, model training being carried out to each independent event section respectively, obtains each section Interior optimum prediction model, is then trained according to the corresponding relationship of initial data and optimal models, and different predictions can be obtained The handover mechanism of model；Step 4: the models switching mechanism in conjunction with the model library in step 2 and in step 3 completes longer wind power prediction Prediction；The climbing event detection for completing prediction data is finally defined according to the numerical value of climbing event, it is contemplated that practical to occur Climbing harm is related with the state of power grid, therefore completes thing of climbing to wind power further combined with the actual motion state of power grid The prediction and judgement of part；
- The event prediction method 2. a kind of wind power of handover mechanism containing prediction model according to claim 1 is climbed, Be characterized in that: the step 3 mainly includes following sub-step,Step 3.1, the model in step 2 in model library is numbered, when carrying out model training to continuous events section, can be obtained To the numbered sequence { c of optimal modelsn, in conjunction with original event segment data, if the training set of formula eight can be established；Step 3.2, above-mentioned training set can be regarded as containing multi-class categorized data set, and one of classification indicates a kind of prediction mould Type can establish multi-class support vector cassification model training classifier to this；By method of classifying, for type containing k Other data need to establish k-1 basic SVM models, such as following formula:In formula, ω, b are classifier parameters；ξ is slack variable；ynThe class variable redefined is indicated, based on a certain classification When wanting research object, then yn=1, remaining class label yn=-1；3 class models in step 2 can be obtained by the solution of SVM model The classifier in library, the classifier are the handover mechanism of local increment.
- The event prediction method 3. a kind of wind power of handover mechanism containing prediction model according to claim 1 is climbed, Be characterized in that: the step 4 mainly includes following sub-step,Step 4.1, obtained in the step 1 as unit of partial model prove-in length, in conjunction with numerical weather forecast as a result, completing Local wind power prediction in short time；Simultaneously according to models switching mechanism, the switching of different event section prediction model is carried out, The prediction of comprehensive multiple continuous events sections, completes long-term wind power prediction；Step 4.2, it detects to meet climbing number in prediction result such as formula ten according to given wind power climbing event definition It is worth the climbing event of definition；Consider the plan of power grid actual schedule, such as cut machine-cut load condition, in the base that operation of power networks state is added On plinth, detects the climbing event with harmfulness, be scheduled and control so as to guide electric system；This definition thinks the Error Absolute Value of the wind power amplitude at two moment for being separated by length Δ t with time interval Δ t's Ratio is greater than a certain given maximum power climbing rate Rval, then a referred to as wind power climbing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510219455.7A CN104820869B (en) | 2015-04-29 | 2015-04-29 | A kind of wind power climbing event prediction method of the mechanism containing models switching |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510219455.7A CN104820869B (en) | 2015-04-29 | 2015-04-29 | A kind of wind power climbing event prediction method of the mechanism containing models switching |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104820869A CN104820869A (en) | 2015-08-05 |
| CN104820869B true CN104820869B (en) | 2019-01-29 |
Family
ID=53731155
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510219455.7A Active CN104820869B (en) | 2015-04-29 | 2015-04-29 | A kind of wind power climbing event prediction method of the mechanism containing models switching |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104820869B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105160434A (en) * | 2015-09-15 | 2015-12-16 | 武汉大学 | Wind power ramp event prediction method by adopting SVM to select forecasting model |
| CN107909212B (en) * | 2017-11-20 | 2021-12-17 | 武汉大学 | Climbing event prediction method based on mesoscale wind power development process serialization |
| CN110245866A (en) * | 2019-06-14 | 2019-09-17 | 南京信息工程大学 | A kind of wind power climbing event detecting method |
| CN112529275B (en) * | 2020-12-02 | 2023-09-22 | 中国矿业大学 | Wind power climbing event prediction method based on feature extraction and deep learning |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102479347A (en) * | 2010-11-24 | 2012-05-30 | 香港理工大学 | Method and system for forecasting short-term wind speed of wind farm based on data driving |
| CN103400039A (en) * | 2013-07-31 | 2013-11-20 | 武汉大学 | Wind power climbing prediction model switching method based on gale weather classification |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080133438A1 (en) * | 2006-11-30 | 2008-06-05 | International Business Machines Corporation | System and method for feature based load shedding in classification |
| WO2011130297A2 (en) * | 2010-04-13 | 2011-10-20 | The Regents Of The University Of California | Methods of using generalized order differentiation and integration of input variables to forecast trends |
-
2015
- 2015-04-29 CN CN201510219455.7A patent/CN104820869B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102479347A (en) * | 2010-11-24 | 2012-05-30 | 香港理工大学 | Method and system for forecasting short-term wind speed of wind farm based on data driving |
| CN103400039A (en) * | 2013-07-31 | 2013-11-20 | 武汉大学 | Wind power climbing prediction model switching method based on gale weather classification |
Non-Patent Citations (3)
| Title |
|---|
| Forecasting Uncertainty Related to Ramps of Wind Power Production;Arthur Bossavy 等;《European Wind Energy Conference and Exhibition》;20100430;第5-7页 |
| 基于原子稀疏分解和BP神经网络的风电功率爬坡事件预测;崔明建 等;《电力系统自动化》;20140625;第38卷(第12期);第6-10页 |
| 风电功率爬坡事件预测时间窗口的选取;欧阳庭辉 等;《电网技术》;20150205;第39卷(第2期);第414-418页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104820869A (en) | 2015-08-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xie et al. | A review of machine learning applications in power system resilience | |
| US11204591B2 (en) | Modeling and calculating normalized aggregate power of renewable energy source stations | |
| KR102159692B1 (en) | solar photovoltatic power generation forecasting apparatus and method based on big data analysis | |
| Sharma et al. | An application of IOT to develop concept of smart remote monitoring system | |
| Zou et al. | An inter type-2 FCR algorithm based T–S fuzzy model for short-term wind power interval prediction | |
| CN102663513B (en) | Utilize the wind power combined prediction modeling method of grey relational grade analysis | |
| Cali et al. | Short-term wind power forecasting using long-short term memory based recurrent neural network model and variable selection | |
| CN105337575B (en) | Photovoltaic plant status predication and method for diagnosing faults and system | |
| CN104820869B (en) | A kind of wind power climbing event prediction method of the mechanism containing models switching | |
| CN109522599A (en) | Transmission line of electricity catastrophic failure method for early warning caused by a kind of typhoon | |
| CN110265996A (en) | It is a kind of suitable for photovoltaic/wind power prediction time characteristic scale modeling method | |
| CN105160434A (en) | Wind power ramp event prediction method by adopting SVM to select forecasting model | |
| Hou et al. | Spatial distribution assessment of power outage under typhoon disasters | |
| He et al. | Day-ahead peak load probability density forecasting based on QRLSTM-DF considering exogenous factors | |
| Sun et al. | A multi-model-integration-based prediction methodology for the spatiotemporal distribution of vulnerabilities in integrated energy systems under the multi-type, imbalanced, and dependent input data scenarios | |
| Hu et al. | Operational reliability evaluation method based on big data technology | |
| CN105958474A (en) | Power transmission line dynamic capacity increasing method and system used for power grid regulation and control system | |
| Yokoyama et al. | Optimal hyperparameters for random forest to predict leakage current alarm on premises | |
| Abaas et al. | Long short-term memory customer-centric power outage prediction models for weather-related power outages | |
| Wei et al. | Ultra-short-term/short-term wind power continuous prediction based on fuzzy clustering analysis | |
| Femin et al. | Modeling the temporal variations in the output of large solar PV power plants | |
| CN113052250A (en) | Decision support method, system, device and medium based on meteorological disaster | |
| Li et al. | Convolutional Neural Networks and Bidirectional Gated Recurrent Unit Model Based Photovoltaic Power Prediction | |
| Xie et al. | Heavy overload forecasting of distribution transformers based on neural network | |
| Galphade et al. | Intelligent multiperiod wind power forecast model using statistical and machine learning model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |