CN106875033A - A kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting - Google Patents
A kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting Download PDFInfo
- Publication number
- CN106875033A CN106875033A CN201611215714.XA CN201611215714A CN106875033A CN 106875033 A CN106875033 A CN 106875033A CN 201611215714 A CN201611215714 A CN 201611215714A CN 106875033 A CN106875033 A CN 106875033A
- Authority
- CN
- China
- Prior art keywords
- wind
- power
- cluster
- electricity generation
- powered electricity
- 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.)
- Granted
Links
- 230000005611 electricity Effects 0.000 title claims abstract description 56
- 238000013277 forecasting method Methods 0.000 title claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 26
- 238000000034 method Methods 0.000 claims abstract description 25
- 238000005259 measurement Methods 0.000 claims abstract description 15
- 238000011156 evaluation Methods 0.000 claims abstract description 7
- 230000000694 effects Effects 0.000 claims description 13
- 241001123248 Arma Species 0.000 claims description 12
- 238000010586 diagram Methods 0.000 claims description 11
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000000205 computational method Methods 0.000 claims description 6
- 238000011144 upstream manufacturing Methods 0.000 claims description 6
- 206010008190 Cerebrovascular accident Diseases 0.000 claims description 3
- 208000006011 Stroke Diseases 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 230000005684 electric field Effects 0.000 claims description 3
- 238000012821 model calculation Methods 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000000714 time series forecasting Methods 0.000 claims description 3
- 238000006467 substitution reaction Methods 0.000 claims description 2
- 238000012804 iterative process Methods 0.000 description 4
- 241000208340 Araliaceae Species 0.000 description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 235000008434 ginseng Nutrition 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development 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"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Evolutionary Computation (AREA)
- Entrepreneurship & Innovation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Wind Motors (AREA)
Abstract
Invention provides a kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting, carries out according to the following steps:Step 1:Historical data is collected, wind-powered electricity generation cluster is divided;Step 2:According to the wind-powered electricity generation cluster for dividing, setup time sequential forecasting models, numerical weather forecast forecast model, space resources matching three forecast models of forecast model, and train three power predictions of forecast model of wind-powered electricity generation cluster;Step 3:The optimal forecast model of training error evaluation result is selected according to three kinds of training error evaluation results of models;Step 4:Collect real time value weather forecast NWP data and realtime power measurement data;Step 5:According to the forecast model selected in training process, real-time NWP data and realtime power measurement data are substituted into, obtain sub-cluster and predict the outcome, by the power prediction results added of sub-cluster, obtain cluster macro-forecast result.The present invention chooses optimal forecast model for the wind-powered electricity generation cluster of different operating modes, lifts precision of prediction.
Description
Technical field
The present invention relates to technical field of wind power generation, and in particular to a kind of wind-powered electricity generation cluster power based on dynamic self-adapting is pre-
Survey method, it is adaptable to the power prediction of large-scale wind power cluster.
Background technology
In recent years, as global energy problem is increasingly serious, Renewable Energy Development generating, especially wind-power electricity generation are more
It is important.But wind energy has intrinsic fluctuation, unstability and intermittence so that wind-powered electricity generation is exerted oneself with the change of wind speed
Fluctuation.If exerting oneself for the correctly predicted wind-powered electricity generation future time instance of energy, will can all bring positive shadow to the safe and stable operation of power network
Ring.By predicting the wind power generation capacity of future time instance, grid side can in advance adjust operation plan so as to avoid electric energy it is unstable,
Lack the problems such as supplying.The power generating value of wind power plant day can be in advance obtained so as to scientific arrangement overhaul of the equipments and failure in wind farm side
Safeguard.
Wind power forecasting system both domestic and external is directed to single wind power plant mostly, and the method for use has Physical, time sequence
Row method, artificial intelligence method etc..But the power prediction of single wind power plant can not meet the demand of dispatching of power netwoks.To dispatching of power netwoks
For, the fluctuation meaning of the wind-powered electricity generation cluster overall power that multiple wind power plants are formed is even more important.Wind-powered electricity generation cluster power both domestic and external
Forecasting system mainly rises method of scales using the addition method and statistics.The addition method adds up the power prediction result of single wind power plant, shape
Into the overall power of wind-powered electricity generation cluster.Statistics rises method of scales and first selects benchmark wind power plant, and predicts the power of benchmark wind power plant, then leads to
The power prediction result for crossing benchmark wind power plant rises yardstick, obtains the power of wind-powered electricity generation cluster.Power prediction of these methods to cluster
With certain effect, but there is a problem of that the model training time is long, precision is not high.
The content of the invention
The present invention lifts the power prediction precision of wind-powered electricity generation cluster, there is provided one kind is based on to overcome the deficiencies in the prior art
The wind-powered electricity generation cluster power forecasting method of dynamic self-adapting, the wind-powered electricity generation cluster for different operating modes chooses optimal forecast model, carries
Rise precision of prediction.
The technical scheme that the present invention takes is as follows:
A kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting, it is characterised in that carry out according to the following steps:
Step 1:Wind power plant historical data is collected, wind-powered electricity generation cluster is entered according to local geographical position and topological structure of electric
Row is divided;
Step 2:According to the wind-powered electricity generation cluster for dividing, setup time sequential forecasting models, numerical weather forecast forecast model, sky
Between three forecast models of resource matched forecast model, and train three power predictions of forecast model of wind-powered electricity generation cluster;
Step 3:The optimal prediction mould of training error evaluation result is selected according to three kinds of training error evaluation results of models
Type;
Step 4:Collect real time value weather forecast NWP data and realtime power measurement data;
Step 5:According to the forecast model selected in training process, real-time NWP data and realtime power measurement data are substituted into,
Obtain sub-cluster to predict the outcome, by the power prediction results added of sub-cluster, obtain cluster macro-forecast result.
The step 1 specifically includes following steps:
Step 1-1:Wind power plant weather history forecast data is collected, weather history forecast data contains wind speed, wind direction, temperature
Degree, humidity and pneumatic parameter;
Step 1-2:Wind power plant geographic position data is collected, principle is closed on geographical position wind-powered electricity generation cluster is divided;
Step 1-3:Collect each wind power plant historical power data.
The step 2 specifically includes following steps:
Step 2-1:Setup time sequential forecasting models:With autoregressive moving-average model ARMA as time series forecasting
Model, parameter identification is carried out to arma modeling using the power data of history wind-powered electricity generation cluster, forms upstream and downstream effect forecast model;
I.e.
Wherein xtThe power of prediction time t, x are wanted in representativet-jRepresent the measured power at t-j moment;εt-kIt is pre- for the t-k moment
Error is surveyed, m, n are respectively arma modeling exponent number,θkAnd arma modeling exponent number m, n is obtained by long auto-regression method;For
Autoregression model coefficient, θkIt is moving average model(MA model) coefficient;
Step 2-2:Set up numerical weather forecast forecast model:The forecast model based on BP neural network, with the collection
The internal all NWP of group forecast that wind speed, wind direction and the cluster of point predict that preceding 12 one-hour rating is |input paramete, the actual work(of cluster
Rate is trained for output parameter;In training process, BP neural network node in hidden layer is obtained by traveling through optimization;
Step 2-3:Set up space resources matching forecast model:The computational methods of the forecast model are shown in formula (2);
Wherein,It is the wind-powered electricity generation cluster power prediction value after h hours;L is represented by calculating weight coefficient, is found altogether
The L weight coefficient highest for matching set and t+h moment to be predicted;piIt is the measurement of the wind-powered electricity generation cluster power in matching set
Value;ωi,t+hIt is weight coefficient, weight coefficient value is bigger, and the weighted value represented shared by the set is bigger;L is really in formula (2)
It is fixed, with weight coefficient ωi,t+hComputational methods it is relevant;For the prediction of wind-powered electricity generation cluster, the essence of weight coefficient is calculating two
The distance of space resources parameter between individual cluster;This is apart from di,t+hComputing formula (3) shown in;
M in formula (3) represents the number of cluster apoplexy electric field;ηkIt is certain space resources parameter for overall metering
The weight coefficient of significance level, such as wind speed are the most important parameter of wind power prediction, and weight coefficient could be arranged to highest,
The wind power plant weight coefficient that the corresponding weight coefficient of the big wind power plant of capacity answers specific capacity small is high;vk,t+hFor the moment to be predicted certain
One space resources parameter, vk,iIt is some space resources parameter of history match object;β is shared by power distance in formula
Weight coefficient, Pi,Pt+h-1Represent the power measurement values at i moment and t+h-1 moment;According to the distance that formula (3) is calculated, draw
Historical power and space resources apart from scatter diagram an example;For historical power and space resources apart from scatter diagram and
Speech, need to set a threshold value δs;Less than δsThe corresponding historical power of matching set will be used for the prediction of realtime power, it is and big
In δsSet be then considered as unrelated with power to be predicted, therefore can be excluded.Threshold value δsCalculating such as formula (4) shown in,
Wherein dminIt is lowest distance value;dmedIt is the median apart from scatter diagram;prIt is from dminAnd dmedInterception is close in interval
dminData percentage;
δs=dmin+pr·(dmed-dmin) (4)
For model calculation formula (2), it is necessary to further determine that the weight system of each set after matching set determines
Number ωi,t+h, it is calculated as shown in formula (5), whereinIt is distance weighting coefficient,It is time weighting coefficient;
Distance weighting coefficientCalculate as shown in formula (6), wherein di,t+hIt is the distance that formula (3) is calculated, μ
It is the median in range distribution scatter diagram, α is undetermined coefficient, will be in optimized selection in training.
Time weighting coefficientRemarkable effect of the time factor in wind power prediction is reflected, closer to current
Its effect of the historical data of predicted time point is more important, time weighting coefficientτiIt is time gap, τi=t+
H-i, λ are time factor, 0<λ<1 need to be in optimized selection in the training process.For different predicted time yardsticks, model pair
Answer different optimized parameters.
The step 4 specifically includes following steps:
Step 4-1:Collect SCADA system the inside realtime power and go out force data;
Step 4-2:Collect the real-time NWP data at numerical weather forecast center.
The step 5 specifically includes herein below:
The minimum forecast model of training error is selected according to step 3, the data in step 4 are substituted into selected prediction mould
In type, obtain sub-cluster and predict the outcome, by the power prediction results added of sub-cluster, obtain cluster macro-forecast result.
The prediction process of three kinds of forecast models is different, divides three kinds of situations to launch to discuss below:
If selection upstream and downstream effect forecast model, the power data in step 4-1 is substituted into formula (1) and is obtained 12 hours
Wind-powered electricity generation cluster predicts the outcome.
If selection weather forecast forecast model, carries out NWP data corrections with formula (7) first.
yt=x0,t+x1,tvt+x2,tvt 2+x3,tvt 3+qt (7)
Wherein vtIt is that NWP models are exported in the wind speed of t, ytIt is the forecasting wind speed error of t.xi,t(i=0,1,2,
3) be using Kalman filter estimate coefficient.Then the power data and revised weather forecast number for step 4 being obtained
Obtain predicting the outcome for the 1st hour according to BP neural network model is substituted into.Needed in the |input paramete of the 2nd hour pre- by the 1st hour
Power scale is substituted into, the like.
If one-hour rating data before NWP data and prediction are substituted into formula by selection space resources matching forecast model
(2)-(6) are predicted.It is worth noting that, in preceding 4 hours of prediction, future position previous hour is contained in |input paramete
Power, the power without previous hour in rear several hours |input parametes of prediction.In preceding 4 hours of prediction, input ginseng
Several iteration.
Compared with prior art, the beneficial effect that reaches of the present invention is:
The present invention is capable of achieving the wind-powered electricity generation cluster power prediction based on dynamic self-adapting technology, further improves power prediction essence
Degree.It is specific as follows:
(1) present invention selects optimal forecast model type according to the predicated error of training stage, it is to avoid because random choosing
Precision is not high caused by selecting forecast model.
(2) present invention proposes a kind of effective space resources Matching Model, and the model modeling is simple, computation complexity
Low, high precision is practical.
(3) space resources matching forecast model proposed by the present invention, contains in preceding four hour |input parametes of prediction
The measurement power of future position previous moment, the measurement power without previous moment, carries in rear 8 hour |input parametes of prediction
The precision of prediction of preceding 4 hours high, but the precision of prediction after 4 hours is not influenceed.
Brief description of the drawings
Fig. 1 is the sample range data that the present invention is provided;
Fig. 2 is the iterative process figure of BP neural network |input paramete;
Fig. 3 is the iterative process figure of the resource matched method |input paramete of wind-powered electricity generation cluster power space.
Fig. 4 is the overall prediction flow chart that the present invention is provided.
Specific embodiment
Below in conjunction with the accompanying drawings, pre- flow gauge of the invention is further elaborated, following instance is used to illustrate the present invention, but
Can not be used for limiting the scope of the present invention.
As shown in figure 4, a kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting, it is characterised in that by following step
Suddenly carry out:
Step 1:The weather history forecast data of each wind farm wind velocity, wind direction, temperature, humidity and air pressure is collected, collects each
Wind power plant geographic position data, is divided with power network topology to wind-powered electricity generation cluster, collects each wind power plant historical power data;
Step 2:According to the wind-powered electricity generation cluster for dividing, setup time sequential forecasting models, numerical weather forecast forecast model, sky
Between three forecast models of resource matched forecast model, and train three power predictions of forecast model of wind-powered electricity generation cluster;
Concretely comprise the following steps:
Step 2-1:Setup time sequential forecasting models:With autoregressive moving-average model ARMA as time series forecasting
Model, parameter identification is carried out to arma modeling using the power data of history wind-powered electricity generation cluster, forms upstream and downstream effect forecast model;
I.e.
Wherein xtThe power of prediction time t, x are wanted in representativet-jRepresent the measured power at t-j moment;εt-kIt is pre- for the t-k moment
Error is surveyed, m, n are respectively arma modeling exponent number,θkAnd arma modeling exponent number m, n is obtained by long auto-regression method;For
Autoregression model coefficient, θkIt is moving average model(MA model) coefficient;
Step 2-2:Set up numerical weather forecast forecast model:The forecast model based on BP neural network, with the collection
The internal all NWP of group forecast that wind speed, wind direction and the cluster of point predict that preceding 12 one-hour rating is |input paramete, the actual work(of cluster
Rate is trained for output parameter;In training process, BP neural network node in hidden layer is obtained by traveling through optimization;
Step 2-3:Set up space resources matching forecast model:Shown in the computational methods of the forecast model such as formula (2);
Wherein,It is the wind-powered electricity generation cluster power prediction value after h hours;L is represented by calculating weight coefficient, is found altogether
The L weight coefficient highest for matching set and t+h moment to be predicted;piIt is the measurement of the wind-powered electricity generation cluster power in matching set
Value;ωi,t+hIt is weight coefficient, weight coefficient value is bigger, and the weighted value represented shared by the set is bigger;L is really in formula (2)
It is fixed, with weight coefficient ωi,t+hComputational methods it is relevant;For the prediction of wind-powered electricity generation cluster, the essence of weight coefficient is calculating two
The distance of space resources parameter between individual cluster;This is apart from di,t+hCalculating such as formula (3) shown in;
M in formula (3) represents the number of cluster apoplexy electric field;ηkFor certain space resources parameter is important for overall metering
The weight coefficient of degree, such as wind speed are the most important parameter of wind power prediction, and its weight coefficient could be arranged to highest, are held
The wind power plant weight coefficient that the big corresponding weight coefficient of wind power plant of amount answers specific capacity small is high;vk,t+hIt is a certain for the moment to be predicted
Individual space resources parameter, vk,iIt is some space resources parameter of history match object;β is the power shared by power distance in formula
Weight coefficient, Pi,Pt+h-1Represent the power measurement values at i moment and t+h-1 moment;According to the distance that formula (3) is calculated, draw
Historical power and space resources apart from scatter diagram an example, as shown in Figure 2.For the figure, a threshold need to be set
Value δs.Less than δsThe corresponding historical power of matching set will be used for the prediction of realtime power, and be more than δsSet be then considered as with
Power to be predicted is unrelated, therefore can be excluded.Dotted line is threshold value in accompanying drawing 1, and the square frame of solid line is the collection chosen
Close.Threshold value δsCalculating such as formula (4) shown in, wherein dminIt is lowest distance value;dmedIt is the middle position apart from scatter diagram
Number;prIt is from dminAnd dmedInterception is near d in intervalminData percentage.
δs=dmin+pr·(dmed-dmin) (4)
For model calculation formula (2), it is necessary to further determine that the weight system of each set after matching set determines
Number ωi,t+h, shown in its computing formula (5), whereinIt is distance weighting coefficient,It is time weighting coefficient;
Distance weighting coefficientCalculate as shown in formula (6), wherein di,t+hIt is the distance that formula (3) is calculated, μ
It is the median in range distribution scatter diagram, α is undetermined coefficient, will be in optimized selection in training;
Time weighting coefficientRemarkable effect of the time factor in wind power prediction is reflected, closer to current pre-
Its effect of the historical data at survey time point is more important;Time weighting coefficientτiIt is time gap, τi=t+h-
I, λ are time factor, 0<λ<1, need to be in optimized selection in the training process.For different predicted time yardsticks, model pair
Answer different optimized parameters.
Step 3:According to three kinds of forecast models of the training error evaluation result selection training error minimum of models;
Step 4:Collect the real-time NWP numbers that SCADA system the inside realtime power goes out force data sum value weather forecast center
According to.
Step 5:According to the forecast model selected in training process, real-time NWP data and realtime power measurement data are substituted into,
Obtain sub-cluster to predict the outcome, by the power prediction results added of sub-cluster, obtain cluster macro-forecast result;Specially:Root
According to the forecast model that step 3 selection training error is minimum, during the data in step 4 are substituted into selected forecast model, three kinds
The prediction process of forecast model is different, divides three kinds of situations to launch to discuss below:
If selection upstream and downstream effect forecast model, SCADA system the inside realtime power will be collected in step 4 and goes out force data
The wind-powered electricity generation cluster that power data substitution formula (1) obtains 12 hours predicts the outcome;
If selection weather forecast forecast model, carries out NWP data corrections with formula (7) first;
yt=x0,t+x1,tvt+x2,tvt 2+x3,tvt 3+qt (7)
Wherein vtIt is that NWP models are exported in the wind speed of t, ytIt is the forecasting wind speed error of t.xi,t(i=0,1,2,
3) be using Kalman filter estimate coefficient.Then the power data and revised weather forecast number for step 4 being obtained
Obtain predicting the outcome for the 1st hour according to BP neural network model is substituted into.Needed in the |input paramete of the 2nd hour pre- by the 1st hour
Power scale is substituted into, the like.The detailed iterative process of |input paramete is shown in accompanying drawing 2.
If one-hour rating data before NWP data and prediction are substituted into formula by selection space resources matching forecast model
(2)-(6) are predicted.It is worth noting that, in preceding 4 hours of prediction, future position previous hour is contained in |input paramete
Power, the power without previous hour in rear several hours |input parametes of prediction.In preceding 4 hours of prediction, input ginseng
Several iterative process are as shown in Figure 3.
Claims (8)
1. a kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting, it is characterised in that carry out according to the following steps:
Step 1:Wind power plant historical data is collected, wind-powered electricity generation cluster is drawn according to local geographical position and topological structure of electric
Point;
Step 2:According to the wind-powered electricity generation cluster for dividing, setup time sequential forecasting models, numerical weather forecast forecast model, space money
Source matches three forecast models of forecast model, and trains three power predictions of forecast model of wind-powered electricity generation cluster;
Step 3:The optimal forecast model of training error evaluation result is selected according to three kinds of training error evaluation results of models;
Step 4:Collect real time value weather forecast NWP data and realtime power measurement data;
Step 5:According to the forecast model selected in training process, real-time NWP data and realtime power measurement data are substituted into, obtained
Sub-cluster predicts the outcome, and by the power prediction results added of sub-cluster, obtains cluster macro-forecast result.
2. the wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting according to claim 1, it is characterised in that described
Step 1 specifically includes following steps:
Step 1-1:Collect wind power plant weather history forecast data, weather history forecast data contains wind speed, wind direction, temperature, wet
Degree and pneumatic parameter;
Step 1-2:Wind power plant geographic position data is collected, principle is closed on geographical position wind-powered electricity generation cluster is divided;
Step 1-3:Collect each wind power plant historical power data.
3. the wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting according to claim 1, it is characterised in that described
Step 2 specifically includes following steps:
Step 2-1:Setup time sequential forecasting models:Using autoregressive moving-average model ARMA as time series forecasting mould
Type, parameter identification is carried out to arma modeling using the power data of history wind-powered electricity generation cluster, forms upstream and downstream effect forecast model;I.e.
Wherein xtThe power of prediction time t, x are wanted in representativet-jRepresent the measured power at t-j moment;εt-kFor the prediction at t-k moment is missed
Difference, m, n are respectively arma modeling exponent number,θkAnd arma modeling exponent number m, n is obtained by long auto-regression method;It is to return certainly
Return model coefficient, θkIt is moving average model(MA model) coefficient;
Step 2-2:Set up numerical weather forecast forecast model:The forecast model based on BP neural network, with the cluster
The all NWP in portion forecast that wind speed, wind direction and the cluster of point predict that preceding 12 one-hour ratings are |input paramete, the actual power of cluster
For output parameter is trained;In training process, BP neural network node in hidden layer is obtained by traveling through optimization;
Step 2-3:Set up space resources matching forecast model:The computational methods of the forecast model are shown in formula (2);
Wherein,It is the wind-powered electricity generation cluster power prediction value after h hours;L is represented by calculating weight coefficient, and L is found altogether
Weight coefficient highest with set with t+h moment to be predicted;piIt is the measured value of the wind-powered electricity generation cluster power in matching set;
ωi,t+hIt is weight coefficient, weight coefficient value is bigger, and the weighted value represented shared by the set is bigger;The determination of L in formula (2), with
Weight coefficient ωi,t+hComputational methods it is relevant;For the prediction of wind-powered electricity generation cluster, the essence of weight coefficient is to calculate two clusters
Between space resources parameter distance;This is apart from di,t+hComputing formula (3) shown in;
M represents the number of cluster apoplexy electric field in formula (3);ηkIt is that certain space resources parameter measures significance level for overall
Weight coefficient, such as wind speed are the most important parameter of wind power prediction, and weight coefficient is set to highest, the big wind power plant of capacity
The wind power plant weight coefficient that corresponding weight coefficient answers specific capacity small is high;vk,t+hFor some space resources at moment to be predicted is joined
Number, vk,iIt is some space resources parameter of history match object;β is the weight coefficient shared by power distance, P in formulai,
Pt+h-1Represent the power measurement values at i moment and t+h-1 moment;According to the distance that formula (3) is calculated, the historical power for drawing and
An example of the space resources apart from scatter diagram;For historical power and space resources are apart from scatter diagram, one need to be set
Threshold value δs;Less than δsThe corresponding historical power of matching set will be used for the prediction of realtime power, and be more than δsSet then regard
It is unrelated with power to be predicted and excludes;Threshold value δsComputing formula (4) shown in, wherein dminIt is lowest distance value;dmedFor
Apart from the median of scatter diagram;prIt is from dminAnd dmedInterception is near d in intervalminData percentage;
δs=dmin+pr·(dmed-dmin) (4)
For model calculation formula (2), it is necessary to further determine that the weight coefficient of each set after matching set determines
ωi,t+h, shown in its computing formula (5), whereinIt is distance weighting coefficient,It is time weighting coefficient;
Distance weighting coefficientCalculate as shown in formula (6), wherein di,t+hFor the distance that formula (3) is calculated, μ is distance
Median in distribution scatter diagram, α is undetermined coefficient, will be in optimized selection in training;
Time weighting coefficientRemarkable effect of the time factor in wind power prediction is reflected, during closer to current predictive
Between historical data its effect for putting it is more important;Time weighting coefficientτiIt is time gap, τi=t+h-i, λ are
Time factor, 0<λ<1, need to be in optimized selection in the training process;For different predicted time yardsticks, model correspond to not
Same optimized parameter.
4. the wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting according to claim 1, it is characterised in that described
Step 4 specifically includes following steps:
Step 4-1:Collect SCADA system the inside realtime power and go out force data;
Step 4-2:Collect the real-time NWP data at numerical weather forecast center.
5. the wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting according to claim 1, it is characterised in that described
Step 5 is specially:The minimum forecast model of training error is selected according to step 3, the data in step 4 is substituted into selected pre-
Survey in model, obtain sub-cluster and predict the outcome, by the power prediction results added of sub-cluster, obtain cluster macro-forecast result.
6. the wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting according to claim 5, it is characterised in that if choosing
Upstream and downstream effect forecast model is selected, the power data in step 4-1 is substituted into the wind-powered electricity generation cluster prediction that formula (1) obtains 12 hours
As a result.
7. the wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting according to claim 5, it is characterised in that if choosing
Weather forecast forecast model is selected, NWP data corrections is carried out with formula (7) first.
yt=x0,t+x1,tvt+x2,tvt 2+x3,tvt 3+qt (7)
Wherein vtIt is that NWP models are exported in the wind speed of t, ytIt is the forecasting wind speed error of t;xi,t(i=0,1,2,3) it is
The coefficient estimated using Kalman filter, the power data for then obtaining step 4 and revised data of weather forecast generation
Enter BP neural network model to obtain predicting the outcome for the 1st hour;Needed the pre- measurement of power of the 1st hour in the |input paramete of the 2nd hour
Rate is substituted into, the like.
8. the wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting according to claim 5, it is characterised in that if choosing
Space resources matching forecast model is selected, power data substitution formula (2)-(6) of NWP data and prediction previous hour is carried out pre-
Survey;It is worth noting that, in preceding 4 hours of prediction, containing the power of future position previous hour in |input paramete, in prediction
The power of previous hour is free of in several hours |input parametes afterwards, in preceding 4 hours of prediction, the iteration mistake of |input paramete.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611215714.XA CN106875033B (en) | 2016-12-26 | 2016-12-26 | Wind power cluster power prediction method based on dynamic self-adaption |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611215714.XA CN106875033B (en) | 2016-12-26 | 2016-12-26 | Wind power cluster power prediction method based on dynamic self-adaption |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106875033A true CN106875033A (en) | 2017-06-20 |
CN106875033B CN106875033B (en) | 2020-06-02 |
Family
ID=59164185
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611215714.XA Active CN106875033B (en) | 2016-12-26 | 2016-12-26 | Wind power cluster power prediction method based on dynamic self-adaption |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106875033B (en) |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108256693A (en) * | 2018-02-11 | 2018-07-06 | 阳光电源股份有限公司 | A kind of photovoltaic power generation power prediction method, apparatus and system |
CN108428019A (en) * | 2018-05-15 | 2018-08-21 | 阳光电源股份有限公司 | The method for building up and photovoltaic power prediction technique of component battery temperature computation model |
CN109934374A (en) * | 2017-12-18 | 2019-06-25 | 斗山重工业建设有限公司 | Power consumption forecasting system and its method |
CN110570030A (en) * | 2019-08-22 | 2019-12-13 | 国网山东省电力公司经济技术研究院 | Wind power cluster power interval prediction method and system based on deep learning |
CN111191815A (en) * | 2019-11-25 | 2020-05-22 | 清华大学 | Ultra-short-term output prediction method and system for wind power cluster |
CN111222738A (en) * | 2019-10-18 | 2020-06-02 | 华中科技大学 | Method for predicting power and optimizing parameters of wind power cluster |
CN111323847A (en) * | 2018-12-13 | 2020-06-23 | 北京金风慧能技术有限公司 | Method and apparatus for determining weight ratios for analog integration algorithms |
CN112633632A (en) * | 2020-11-26 | 2021-04-09 | 华中科技大学 | Integrated short-term wind power cluster power prediction method based on signal decomposition technology |
CN112906928A (en) * | 2019-12-03 | 2021-06-04 | 国网山西省电力公司电力科学研究院 | Wind power plant cluster active power prediction method and system |
CN112990533A (en) * | 2021-01-19 | 2021-06-18 | 中国农业大学 | Wind power cluster power prediction method based on sparse constraint and dynamic weight distribution |
WO2022057427A1 (en) * | 2020-09-15 | 2022-03-24 | 华为技术有限公司 | Method, apparatus and device for power prediction |
WO2023093774A1 (en) * | 2021-11-26 | 2023-06-01 | 中国华能集团清洁能源技术研究院有限公司 | Deep learning-based wind power cluster power prediction method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103699944A (en) * | 2013-12-31 | 2014-04-02 | 国电南京自动化股份有限公司 | Wind and photovoltaic generation power prediction system with multiple prediction modes |
WO2015159356A1 (en) * | 2014-04-15 | 2015-10-22 | 株式会社日立製作所 | Power control system and power control method |
CN106251242A (en) * | 2016-08-08 | 2016-12-21 | 东南大学 | A kind of wind power output interval combinations Forecasting Methodology |
-
2016
- 2016-12-26 CN CN201611215714.XA patent/CN106875033B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103699944A (en) * | 2013-12-31 | 2014-04-02 | 国电南京自动化股份有限公司 | Wind and photovoltaic generation power prediction system with multiple prediction modes |
WO2015159356A1 (en) * | 2014-04-15 | 2015-10-22 | 株式会社日立製作所 | Power control system and power control method |
CN106251242A (en) * | 2016-08-08 | 2016-12-21 | 东南大学 | A kind of wind power output interval combinations Forecasting Methodology |
Non-Patent Citations (1)
Title |
---|
彭小圣 等: ""风电集群短期及超短期功率预测精度改进方法综述"", 《中国电机工程学报》 * |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109934374A (en) * | 2017-12-18 | 2019-06-25 | 斗山重工业建设有限公司 | Power consumption forecasting system and its method |
CN108256693A (en) * | 2018-02-11 | 2018-07-06 | 阳光电源股份有限公司 | A kind of photovoltaic power generation power prediction method, apparatus and system |
CN108256693B (en) * | 2018-02-11 | 2024-02-13 | 阳光电源股份有限公司 | Photovoltaic power generation power prediction method, device and system |
CN108428019B (en) * | 2018-05-15 | 2021-09-03 | 阳光电源股份有限公司 | Method for establishing component battery temperature calculation model and photovoltaic power prediction method |
CN108428019A (en) * | 2018-05-15 | 2018-08-21 | 阳光电源股份有限公司 | The method for building up and photovoltaic power prediction technique of component battery temperature computation model |
CN111323847A (en) * | 2018-12-13 | 2020-06-23 | 北京金风慧能技术有限公司 | Method and apparatus for determining weight ratios for analog integration algorithms |
CN110570030A (en) * | 2019-08-22 | 2019-12-13 | 国网山东省电力公司经济技术研究院 | Wind power cluster power interval prediction method and system based on deep learning |
CN111222738A (en) * | 2019-10-18 | 2020-06-02 | 华中科技大学 | Method for predicting power and optimizing parameters of wind power cluster |
CN111222738B (en) * | 2019-10-18 | 2022-04-15 | 华中科技大学 | Method for predicting power and optimizing parameters of wind power cluster |
CN111191815A (en) * | 2019-11-25 | 2020-05-22 | 清华大学 | Ultra-short-term output prediction method and system for wind power cluster |
CN111191815B (en) * | 2019-11-25 | 2022-08-16 | 清华大学 | Ultra-short-term output prediction method and system for wind power cluster |
CN112906928A (en) * | 2019-12-03 | 2021-06-04 | 国网山西省电力公司电力科学研究院 | Wind power plant cluster active power prediction method and system |
CN112906928B (en) * | 2019-12-03 | 2022-09-16 | 国网山西省电力公司电力科学研究院 | Wind power plant cluster active power prediction method and system |
WO2022057427A1 (en) * | 2020-09-15 | 2022-03-24 | 华为技术有限公司 | Method, apparatus and device for power prediction |
CN112633632A (en) * | 2020-11-26 | 2021-04-09 | 华中科技大学 | Integrated short-term wind power cluster power prediction method based on signal decomposition technology |
CN112990533A (en) * | 2021-01-19 | 2021-06-18 | 中国农业大学 | Wind power cluster power prediction method based on sparse constraint and dynamic weight distribution |
CN112990533B (en) * | 2021-01-19 | 2024-01-05 | 中国农业大学 | Wind power cluster power prediction method adopting sparse constraint and dynamic weight distribution |
WO2023093774A1 (en) * | 2021-11-26 | 2023-06-01 | 中国华能集团清洁能源技术研究院有限公司 | Deep learning-based wind power cluster power prediction method |
Also Published As
Publication number | Publication date |
---|---|
CN106875033B (en) | 2020-06-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106875033A (en) | A kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting | |
CN103279804B (en) | The Forecasting Methodology of super short-period wind power | |
Liu et al. | Wind power plant prediction by using neural networks | |
CN102880810B (en) | Wind power prediction method based on time sequence and neural network method | |
CN103683274B (en) | Regional long-term wind power generation capacity probability prediction method | |
CN107704953A (en) | The short-term wind-electricity power probability density Forecasting Methodology of EWT quantile estimate forests | |
CN105654207A (en) | Wind power prediction method based on wind speed information and wind direction information | |
CN111353652B (en) | Wind power output short-term interval prediction method | |
CN104899665A (en) | Wind power short-term prediction method | |
CN102938562B (en) | Prediction method of total wind electricity power in area | |
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 | |
CN107516145A (en) | A kind of multichannel photovoltaic power generation output forecasting method based on weighted euclidean distance pattern classification | |
CN105069521A (en) | Photovoltaic power plant output power prediction method based on weighted FCM clustering algorithm | |
CN109086928A (en) | Photovoltaic plant realtime power prediction technique based on SAGA-FCM-LSSVM model | |
CN102682207A (en) | Ultrashort combined predicting method for wind speed of wind power plant | |
CN111612244B (en) | QRA-LSTM-based method for predicting nonparametric probability of photovoltaic power before day | |
CN105303250A (en) | Wind power combination prediction method based on optimal weight coefficient | |
Xu et al. | Correlation based neuro-fuzzy Wiener type wind power forecasting model by using special separate signals | |
CN106295857A (en) | A kind of ultrashort-term wind power prediction method | |
Yang et al. | Photovoltaic power forecasting with a rough set combination method | |
CN113505909B (en) | Error compensation method for short-term wind power trend prediction | |
CN105741192A (en) | Short-term wind speed combined forecasting method for wind turbine cabin of wind power plant | |
CN105844350A (en) | Short period wind power prediction system based on covariance preferable combination model | |
CN102750542A (en) | Support vector regression machine wind speed combination forecast method with interpolation being smoothed and optimized |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |