CN108062595A - Complex landform region wind energy Forecasting Methodology in short-term based on WRF/CFD/SAHDE-RVM couplings - Google Patents
Complex landform region wind energy Forecasting Methodology in short-term based on WRF/CFD/SAHDE-RVM couplings Download PDFInfo
- Publication number
- CN108062595A CN108062595A CN201711218419.4A CN201711218419A CN108062595A CN 108062595 A CN108062595 A CN 108062595A CN 201711218419 A CN201711218419 A CN 201711218419A CN 108062595 A CN108062595 A CN 108062595A
- Authority
- CN
- China
- Prior art keywords
- wind
- short
- rvm
- sahde
- wrf
- 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
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000008878 coupling Effects 0.000 title claims abstract description 10
- 238000010168 coupling process Methods 0.000 title claims abstract description 10
- 238000005859 coupling reaction Methods 0.000 title claims abstract description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 14
- 238000004088 simulation Methods 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 11
- 230000001373 regressive effect Effects 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 150000001875 compounds Chemical class 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 230000007774 longterm Effects 0.000 claims description 4
- 238000003780 insertion Methods 0.000 claims description 3
- 230000037431 insertion Effects 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 2
- 230000006870 function Effects 0.000 description 19
- 238000010586 diagram Methods 0.000 description 7
- 238000002922 simulated annealing Methods 0.000 description 5
- 230000005611 electricity Effects 0.000 description 4
- 239000012530 fluid Substances 0.000 description 4
- 230000035772 mutation Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011017 operating method Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 239000011229 interlayer Substances 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000010587 phase diagram Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 230000002028 premature Effects 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000004513 sizing Methods 0.000 description 1
- 239000002344 surface layer Substances 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000009423 ventilation 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
- 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
- Y02A30/00—Adapting or protecting infrastructure or their operation
-
- 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
-
- 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
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Invention provides the wind energy Forecasting Methodology in short-term of the complex landform region based on WRF/CFD/SAHDE RVM couplings.The Forecasting Methodology include WRF wind directions prediction, SAHDE RVM short-time wind speeds prediction, tentatively make APG, improve APG and in short-term wind energy prediction and etc..This method can more accurately match the information of corresponding boundary condition by the observation data of airspeedometer.Wind direction and wind speed information in the following short time for the wind tower in complex landform region can be provided, adjusted in real time convenient for each wind tower in region, wind energy power is stung in raising.It has broad application prospects.
Description
Technical field
The present invention relates to wind-powered electricity generation field, more particularly to a kind of wind energy Forecasting Methodology.
Background technology
Predict have in existing method a kind of based on weather research to carry out wind energy in short-term to a complex landform region
The meteorological data provided with Forecast Mode (The Weather Research and Forecasting Model, WRF) exists
Carry out the wind field short-term prediction in region in WAsP softwares, the basic thoughts of WAsP models is will to control the equation of momentum of air motion
Linearisation can so substantially reduce and solve nonlinear partial differential equation (the nonlinear partial
Differential equations) difficulty, while also save substantial amounts of time cost.In addition one kind is carried based on WRF
The meteorological data of confession carries out real-time Simulation prediction, then calculates analog result input CFD software, it is pre- to obtain entire area
The method for surveying wind field information.In addition with a kind of method (Measure-Correlation- for measuring and being contacted with prediction
Prediction, abbreviation MCP).
Inventor has found the prior art, and at least there are following defects：
It is that geostrophic wind is calculated by airspeedometer observational data, it is assumed that turn a certain range ofly first, using WAsP models
Wind speed will not change, and further according to the topographical features of surrounding area, calculate that the surface layer wind speed of any point and wind energy are joined by geostrophic wind
Number.Due to being extrapolated from an observation point, error can be increased with the increase with ventilation measuring point distance, therefore directly use WAsP
It is calculated in the wind energy of large area infeasible；Further, since WAsP is calculated in itself using linear model, can cause with stream
The uncertainty of the result of calculation of complex fluid through landform has its certain limitation；Existing research finds to show simultaneously
The acceleration of air-flow at mountain top can be over-evaluated using linear model and can not solve the problems, such as the gas distribution of massif leeward.
Second is that the method for numerical simulation is carried out in real time using CFD software, it is necessary to extract short-term prediction meteorology from WRF in time
Data, while run that spent computing resource is huge, it is necessary to which certain calculating time, efficiency is difficult to ensure that, is not easy to wind-powered electricity generation
Tower carries out reply adjustment in real time according to wind energy prediction result in short-term.
Third, two necessary conditions are needed using MCP methods：(1) need to obtain one section of complex landform region global and local
Observation data in time, the correlation that (2) global data will have with local data.But due under complex landform region
Fluid behaviour it is non-linear, be very difficult to find a specific equation at present and go to describe their correlation；Meanwhile it carries out global
DATA REASONING on the spot need to set up the observation that multiple airspeedometers carry out whole distract prolonged wind speed and wind direction, this side
Method is built first and maintenance cost is high, and the secondly improper selection of indivedual measurement place positions may influence complex topographic territory
The expression of wind field information.
Therefore it provides the complex landform region wind energy Forecasting Methodology in short-term based on WRF/CFD/SAHDE-RVM couplings, to multiple
The wind energy assessment of miscellaneous shaped area has important meaning.
The content of the invention
The object of the present invention is to provide a kind of bridge up train Control for Speed Limitation that reasonable speed limit can be calculated according to actual state
System, to solve problems of the prior art.
To realize the present invention purpose and the technical solution adopted is that such, based on WRF/CFD/SAHDE-RVM couplings
Complex landform region wind energy Forecasting Methodology in short-term, comprises the following steps：
1) target complex landform region topographic(al) data is obtained, carrying out wind direction in short-term based on the meteorological data that WRF is provided predicts.
Draw the mainstream wind direction in short-term in target complex landform region.
2) SAHDE-RVM short-time wind speeds are predicted.
2.1) regression forecasting is carried out using RVM, establishes RVM regressive prediction models.
2.2) training sample of the historical wind speed wind direction data as SAHDE algorithms at the airspeedometer of target complex landform region is collected
This.By the training to sample data set, adaptive obtaining step 2.1) the RVM regressive prediction models optimized parameter.
2.3) according to airspeedometer real time data, the short-time wind speed prediction result at airspeedometer is returned.
3) in CFD software, the numerical simulation calculation of target complex landform under several wind angles is uniformly carried out.According to
CFD results make initial APG.
4) according to the long-term measured data of airspeedometer, algorithm consummation APG is passed through.
5) the wind direction prediction result of the WRF described in step 1) with APG step 4) described is matched, extracted corresponding
Boundary speed information and information of flow.Obtain the short-term prediction result of entire area wind energy.
Further, the regressive prediction model of RVM described in step 2.1) is：
In formula, wiFor weighting coefficient, K (x, xi) it is kernel function, N is sample size.
Further, the kernel function of the RVM regressive prediction models selects gaussian kernel function to be combined with binomial kernel function
The compound kernel function arrived.Wherein, the compound kernel function is：
K(xi,xj)=λ G (xi,xj)+(1-λ)P(xi,xj) (2)
In formula, G (xi,xj) for gaussian kernel function, P (xi,xj) it is binomial kernel function, λ is weight.Wherein, 0≤λ≤1.
Further, the step 3) specifically includes：
3.1) CFD software obtains the detailed place information in target complex landform region, is corresponded to pre-processing software
Model foundation, establish computational domain and set boundary condition.
3.2) will be divided into after the boundary speed nondimensionalization of input x to and y to.Wherein, x to speed represented with u, y to
Speed is represented with v.Boundary wind speed is denoted as u under different wind angle θB=Vcos θ, vB=Vsin θ, B represent border.Calculate nothing
Information of flow after dimension is represented with f (i, j, k).
3.3) air speed data (u of the corresponding airspeedometer crosswind point under its observed altitude is recordedA,vA) and the corresponding whole audience
The wind speed information on surface.
3.4) by the boundary speed of the nondimensionalization described in step 3.2) and information of flow and the nothing described in step 3.3)
Airspeedometer point (u under the conditions of dimensionA,vA) associated.
3.5) multiple (u obtained according to numerical simulation under different inlet velocitiesA,vA) as a result, its is unitizationIt is denoted as (UA,VA)。
3.6) by (UA,VA) point and it includes boundary speed information and information of flow insertion APG in, draw initial APG.
Wherein, APG only shows the data of airspeedometer actual measurement, but wherein each point contains corresponding border wind speed information and stream
Field information.
The solution have the advantages that unquestionable：
A. the minimum of computing resource needed for realizing；
B. the information of corresponding boundary condition can be more accurately matched by the observation data of airspeedometer；
C. the wind direction and wind speed information in the following short time can be provided for the wind tower in complex landform region, convenient for region
Interior each wind tower is adjusted in real time, and wind energy power is stung in raising；It can also be calculated simultaneously by the relation of wind speed and generated output
Go out generated energy in a short time.
Description of the drawings
Fig. 1 is Forecasting Methodology flow chart；
For WRF, wind direction predicts schematic diagram to Fig. 2 in short-term；
Fig. 3 be APG making, improve and application schematic diagram；
Fig. 4 predicts schematic diagram for SAHDE-RVM short-time wind speeds；
Fig. 5 is to make initial APG schematic diagrames according to CFD results；
Fig. 6 is according to algorithm consummation APG schematic diagrames；
Fig. 7 predicts schematic diagram for SAHDE-RVM short-time wind speeds；
Fig. 8 is WRF cardinal wind prediction result and observation data comparison figure in short-term；
Fig. 9 is SHADE-RVM short-term wind speed forecastings result and measured data comparison diagram；
Figure 10 is to the mesh generation figure of complex landform regional model in CFD software；
Figure 11 schemes for initial APG；
Figure 12 is complex landform region wind energy concentration figure.
Specific embodiment
With reference to embodiment, the invention will be further described, but should not be construed the above-mentioned subject area of the present invention only
It is limited to following embodiments.Without departing from the idea case in the present invention described above, according to ordinary skill knowledge and used
With means, various replacements and change are made, should all be included within the scope of the present invention.
Embodiment 1：
The present embodiment discloses a kind of complex landform region based on WRF/CFD/SAHDE-RVM couplings wind energy prediction side in short-term
Method referring to Fig. 1, comprises the following steps：
1) referring to Fig. 2, target complex landform region topographic(al) data is obtained, and multinest grid is divided to it.Side is set
Interlayer parametric scheme.Wind direction in short-term is carried out based on the meteorological data that WRF is provided to predict.In the following a period of time inner horizontal of simulation
The wind speed of two mutually perpendicular directions exports a data at regular intervals, by the use of the averages of nine mesh points closed on as
The output of wind speed size can obtain being somebody's turn to do for following a period of time as a result, referring to Fig. 8 by the synthesis of two vertical speeds
The mainstream wind direction in short-term in complex landform region.Wherein, the meteorological data include assimilation global circulation model again analysis of data, often
Advise Ground Meteorological data and sounding meteorological data.
2) SAHDE-RVM short-time wind speeds are predicted.
2.1) regression forecasting is carried out using RVM, establishes RVM regressive prediction models.Wherein, the input set of training sample is given
X={ x1,x2,x3,…,xn, with corresponding output collection T={ t1,t2,t3,…,tn, wherein n is number of samples, if tiFor target
It is worth and has ti∈R.Output valve tiFunction model be represented by：
ti=y (xi,w)+εi (1)
In formula, εiRepresent Gauss white noise, and εiObey distribution εi~N (0, σ2), then p (ti|xi)=N (ti|y(xi,
w),σ2).The output of RVM models is represented by the combination of non-thread kernel function, and kernel function need not meet mercer conditions.RVM's returns
The prediction model is returned to be：
In formula, wiFor weighting coefficient, K (x, xi) it is kernel function, N is sample size.
For the output collection t being independently distributediPossibility predication be：
In formula, Φ=(Φ1,Φ2,…,ΦN), ΦI=(1,K(xi,x1),...,K(xi,xn))T(i=1 ..., N).
The Gaussian prior that the weighting parameter defined by sparse Bayesian principle is zero is distributed as：
In formula, αiFor the hyper parameter of priori Gaussian Profile, α=(α0,…,αN)T.Each independent hyper parameter αiIt controls and weighs
Parameter wiPrior distribution, make Method Using Relevance Vector Machine model have it is openness.
According to Bayes principle, by formula (3) and the Posterior distrbutionp of (4) calculating weighted vector w：
In formula,
From formula (5), to determine that weighted vector w need to be to hyper parameter α, σ2It is determined.
The likelihood that hyper parameter is calculated using Bayesian frame is distributed：
p(t|α,σ2)=∫ p (t|w,σ2) p (w | α) dW=N (0, C) (6)
In formula, C is covariance and C=σ2I+ΦA-1ΦT。
To two hyper parameters α, σ in formula (6)2Partial derivative is sought respectively and partial derivative is made to be equal to zero.
It obtains solving optimal hyper parameterMore new iterative algorithm in the process：
In formula, ujIt is j-th of posteriority average weight, γj=1- αjMjj(γj∈ [0,1]), MjjFor weights covariance matrix
The diagonal entry of Σ.
In RVM learning processes, formula (7), two parameters in (8) are initialized first, and pass through that update iterative formula continuous
With newly calculating two hyper parameters, when the parameter in model reaches maximum frequency of training, update, which calculates, to be stopped, and is obtained at this time
α and σ2For optimal value.
If one input value X of given system*, then the probability distribution exported is：
y*=uTΦ(X*)=uMPΦ(X*) (10)
In formula, as a given input value X*, then have corresponding prediction output valve y*。To predict variance, represent uncertain
Property.
2.2) referring to Fig. 7, collect historical wind speed wind direction data at the airspeedometer of target complex landform region and mixed as adaptive
Close differential evolution algorithm (simulated annealing-based hybrid differential evolution, SAHDE)
Training sample.By the training to sample data set, adaptive obtaining step 2.1) the RVM regressive prediction models it is optimal
Parameter.Wherein, in the present embodiment, SAHDE operating procedures are as follows：
A) mutation operation：DE algorithms carry out mutation operation with the following method by a variety of mutation operation methods, choosing.
xm=xbest+F[(x1-x2)+(x3-x4)] (12)
Four parent individualities are picked out at random, are denoted as x respectively1,x2,x3,x4, xbestIt is the optimum individual in parent, xmFor
Make a variation the variation individual generated, and F is aberration rate (F ∈ [0,1.2]).
B) crossover operation：Choose two individual xiAnd xmCrossover operation is carried out, the new individual generated after crossover operation is xc,
Specific operating method such as following formula：
Wherein, rand () represents the random function between [0,1], and randr (i) ∈ { 1,2 ..., D } are whole to randomly generate
Number, D be optimized variable dimension, CRFor crossing-over rate (CR∈(0,1))。
C) selection operation：The offspring individual of DE is generated by greedy back-and-forth method, by above-mentioned intersection with after the operation of variation, producing
Individual x is given birth tocAnd target individualThe two is at war with, and offspring individual, selection operation method such as 11 institute of formula are chosen as compared with the superior
Show：
In formula, xcFor the individual generated after crossover operation and mutation operation,For i-th of individual in filial generation.
SAHDE algorithms carry out global search and local search are carried out in the later stage at the beginning, CRSize by working as evolution
Algebraically gnowWith maximum evolutionary generation gmaxAdaptive adjustment, specific method are as follows：
CR0For crossover operator CRInitial value, CRValue adaptively adjusted according to above formula, initial value CR0It is smaller, then its value
It incrementally increases.
Binary search is carried out to the current optimum individual that SAHDE is generated using simulated annealing.Among SAHDE, choosing
Settled preceding optimum individual is initial individuals, both y0=xbest, initial temperature is chosen to be T0, the mode for generating new individual is as follows：
yr+1,j=yr,j+ηε(xjmax-xjmin), j=1,2 ..., D (16)
In formula, r is the iterations of simulated annealing.yrFor the new individual generated after r iteration.η is disturbed in order to control
Amplitude.ε is to obey average or the stochastic variable of normal distribution.xjmax、xjminThe respectively value range of jth dimension optimized variable.
After the optimum individual that SAHDE is generated is using simulated annealing binary search, fitness variation is Δ F, Δ F=
f(yr+1)-f(yr).Receive new individual if Δ F ＜ 0 and original optimum individual is replaced with new individual.If e(-ΔF/T)＞ rand
() also receives newly generated individual at this time, and a non-optimal individual in population is replaced with it.Otherwise refuse.If selection
Receive new individual, by Tr+1=aTr(0 ＜ a ＜ 1) cools down.Otherwise do not cool down.
In order to accelerate the training speed of sample and improve the precision of prediction of model, normalized method is used in the present embodiment
Wind speed actual measurement sample data is pre-processed：
In formula,For the value of data normalization.xmaxFor the maximum of wind speed sample data.xminFor wind speed sample data
Minimum value.
Select gaussian kernel function with the compound kernel function that binomial kernel function combines as RVM moulds step 2.1) described
The kernel function of type, learns from other's strong points to offset one's weaknesses, and plays respective advantage.Kernel function after combination is as follows：
K(xi,xj)=λ G (xi,xj)+(1-λ)P(xi,xj) (18)
Wherein, G (xi,xj) it is gaussian kernel function, expression formula is：G(xi,xj)=exp (- | | xi-xj||2/σ2)。P(xi,
xj) it is binomial kernel function.λ is weight, 0≤λ≤1.
It collects and passes by training sample of the wind speed and direction data of 1 year as above-mentioned SAHDE-RVM algorithms at airspeedometer, lead to
The training to sample data set is crossed, adaptive optimized parameter α, σ for obtaining model2。
2.3) referring to Fig. 4, according to airspeedometer real time data, the short-time wind speed prediction result at airspeedometer is returned to.Hereafter it is every
The instantaneous wind speed data inputted to it at airspeedometer can return at an airspeedometer in short-term (specified time, such as
15min) forecasting wind speed result.Attached drawing 9 is pair of SHADE-RVM short-term wind speed forecastings result and measured data at 100 measuring points
Than figure.
3) referring to Fig. 5, in CFD software, the numerical simulation calculation of target complex landform under multiple wind angles is uniformly carried out.
Initial airspeedometer phase diagram (Anemometer phase graph, APG) is made according to CFD results.Wherein, it is right in CFD software
The mesh generation figure of complex landform regional model is as shown in Figure 10.Initial APG figures are as shown in figure 11.
3.1) CFD software obtains the detailed place information in target complex landform region, with pre-processing software (such as ICEM)
Corresponding model foundation is carried out, while establishes matched computational domain, FLUENT is imported after grid division, sets side
Condition each border in boundary's carries out numerical simulation calculation after how setting.
3.2) inlet velocity V considers reference altitude zref=10m sentences 10m/s and carries out once the place is evenly dividing 360
Numerical simulation under a wind angle.It records corresponding airspeedometer crosswind point and is denoted as air speed data u of the A points under its observed altitudeA,
vAAnd the wind speed information of corresponding whole audience ground surface.To be divided into after the boundary speed nondimensionalization of input x to and y to.Wherein, x to
Speed represent that y is represented to speed with v with u.Boundary wind speed is denoted as u under different wind angle θB=Vcos θ, vB=Vsin θ,
B represents border.The information of flow after nondimensionalization is calculated, is represented with f (i, j, k).
3.3) air speed data (u of the corresponding airspeedometer crosswind point under its observed altitude is recordedA,vA) and the corresponding whole audience
The wind speed information on surface.
3.4) by the boundary speed of the nondimensionalization described in step 3.2) and information of flow and the nothing described in step 3.3)
Airspeedometer point (u under the conditions of dimensionA,vA) associated.
3.5) the 360 (u obtained according to numerical simulation under different inlet velocitiesA,vA) as a result, its is unitizationIt is denoted as (UA,VA)。
3.6) by (UA,VA) point and it includes boundary speed information and information of flow insertion APG in, draw APG.Wherein,
APG only shows the data of airspeedometer actual measurement, but wherein each point contains corresponding border wind speed information and flow field letter
Breath.
4) according to the long-term measured data of airspeedometer, algorithm consummation APG is passed through.
It is according to the long-term measured data of airspeedometer that APG is perfect referring to Fig. 6.Assuming that u*,v*It is obtained for Anemometer a certain
A data obtain U after unitization*,V*, its point in APG is defined as point P*.If in APG, point P*Close enough certain point
T then by the use of point T as the approximation of P*, proceeds by the flow of next anemometer data.It carries out flowing down if keeping off
Journey：It will point P*It projects in the APG curves of its nearest neighbours, subpoint is set to P, and two endpoints for defining the curved section are T1And T2,
Calculate P*In the relative position of this section：It can be to point P according to r*Border included in APG figures
Wind speed information is calculated：uB'=(1-r) UB1+rUB2, vB'=(1-r) VB1+rVB2, UB1,VB1UB2,VB2Respectively point T1、T2Place
Corresponding boundary speed information.Then pass through uB’,vB' information f ' i, j, the k in flow field are calculated, and insert it into wind speed counting
According to uA’,vA' in.uA’,vA' mould be set to | uA|, it can obtain each above-mentioned amount for acquiring is unitization It will point UA’,VA' and its associated border
Velocity information UB’,VB' and information of flow F ' (i, j, k) be inserted into APG.
Point P is then judged again*Whether close enough certain point T, repeat above-mentioned flow.
As more and more information are collected and typing by airspeedometer, it can become more and more perfect in APG, be capable of providing more
The comprehensively relevant information between anemometer data and boundary speed.
5) referring to Fig. 3, the wind direction prediction result of the WRF described in step 1) with APG step 4) described is matched, is extracted
Go out corresponding boundary speed information and information of flow.Obtain the short-term prediction result of entire area wind energy.It is obtained by prediction result
Complex landform region wind energy concentration figure it is as shown in figure 12.By data at the wind direction prediction result airspeedometer of WRF and draw out
APG is matched, and extracts the boundary speed information and information of flow of corresponding nondimensionalization, according to same wind angle not
With being linearly related this point with reference to the shape of wind conditions Fluid field, using the wind speed size prediction result of SAHDE-RVM, with
Boundary speed information and the information of flow multiplication of nondimensionalization can obtain the short-term prediction result of entire area wind energy.Such as
SAHDE-RVM prediction short-time wind speeds are uC,vC, the boundary speed letter in the corresponding APG of wind direction prediction result of WRF offers
Cease UBC,VBCAnd information of flow FC(i, j, k) calculates corresponding boundary speed It can obtain entire area any point at this time
Wind energy prediction result in short-term.
What deserves to be explained is the present embodiment presents mesoscale numerical weather forecast pattern and small scale numerical model is brand-new
Combination thinking, the method that algorithm consummation APG is used after initial APG is made by CFD, realize needed for computing resource minimum
Change, be not required to carry out numerical simulation calculation to entire area using CFD in real time after WRF prediction data by importing again.Simultaneously because
Site calculated value in mesoscale model be for the average value interior for a period of time of the region in a sizing grid, can not be specific
The wind conditions in a certain definite place are represented, we introduce SAHDE-RVM therefore.By using the single airspeedometer in region
Historical data SAHDE-RVM is trained, realize the forecasting wind speed in opposite mesoscale WRF pattern shorter times.
It is that a regression result made according to previous data is extrapolated difference again with linear WAsP models, the present embodiment includes
There is the numerical simulation calculation that the physical model based on computational fluid dynamics carries out, broader can be answered in complex landform region
With.This example need not only carry out historical wind speed, the wind direction data in the place of wind energy prediction in short-term simultaneously, and only need to obtain full wafer answers
The altitude data of miscellaneous geomorphic province, with reference to WRF and CFD technologies, the sight that can be just obtained by setting up an airspeedometer in the region
Measured data predict entire complex landform region for a period of time after wind speed and direction state.It is right during short-time wind speed prediction
Than traditional DE algorithms, method introduces simulated annealings, carry out binary search to the optimum individual that SAHDE is generated, break away from
The shortcomings that Premature Convergences of traditional DE algorithms.In addition, when the above method has really been applied in wind energy assessment, APG will
A large data sets can be gradually become with the growth of observation time, make it fixed related to traditional MCP methods offer one
Equation is compared, and the information of corresponding boundary condition can be more accurately matched by the observation data of airspeedometer.
The method is applied in wind-powered electricity generation association area, it in the long run, can be to each place in full wafer complex landform region
Often dry to and wind speed size summarized, comprehensive each side's data post analysis can be derived that this area's optimum sets up wind-power electricity generation
The place of tower and the direction of power generation column.On short terms, can be provided for the wind tower in complex landform region in the following short time
Wind direction and wind speed information, adjusted in real time convenient for each wind tower in region, raising sting wind energy power.It can also pass through simultaneously
The relation of wind speed and generated output calculates generated energy in a short time.
Claims (4)
1. the complex landform region wind energy Forecasting Methodology in short-term based on WRF/CFD/SAHDE-RVM couplings, which is characterized in that including
Following steps：
1) target complex landform region topographic(al) data is obtained, carrying out wind direction in short-term based on the meteorological data that WRF is provided predicts；It draws
The mainstream wind direction in short-term in target complex landform region；
2) SAHDE-RVM short-time wind speeds are predicted；
2.1) regression forecasting is carried out using RVM, establishes RVM regressive prediction models；
2.2) training sample of the historical wind speed wind direction data as SAHDE algorithms at the airspeedometer of target complex landform region is collected；
By the training to sample data set, adaptive obtaining step 2.1) the RVM regressive prediction models optimized parameter；
2.3) according to airspeedometer real time data, the short-time wind speed prediction result at airspeedometer is returned；
3) in CFD software, the numerical simulation calculation of target complex landform under several wind angles is uniformly carried out；It is tied according to CFD
Fruit makes initial APG；
4) according to the long-term measured data of airspeedometer, algorithm consummation APG is passed through；
5) the wind direction prediction result of the WRF described in step 1) with APG step 4) described is matched, extracts corresponding border
Velocity information and information of flow；Obtain the short-term prediction result of entire area wind energy.
2. wind energy is predicted in short-term in the complex landform region according to claim 1 based on WRF/CFD/SAHDE-RVM couplings
Method, it is characterised in that：The regressive prediction model of RVM described in step 2.1) is：
<mrow>
<mi>y</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>w</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<msub>
<mi>w</mi>
<mi>i</mi>
</msub>
<mi>K</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>w</mi>
<mn>0</mn>
</msub>
<mo>+</mo>
<msub>
<mi>&epsiv;</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
In formula, wiFor weighting coefficient, K (x, xi) it is kernel function, N is sample size.
3. wind energy is predicted in short-term in the complex landform region according to claim 2 based on WRF/CFD/SAHDE-RVM couplings
Method, it is characterised in that：The kernel function of the RVM regressive prediction models selects gaussian kernel function to be combined with binomial kernel function
The compound kernel function arrived；Wherein, the compound kernel function is：
K(xi,xj)=λ G (xi,xj)+(1-λ)P(xi,xj) (2)
In formula, G (xi,xj) for gaussian kernel function, P (xi,xj) it is binomial kernel function, λ is weight；Wherein, 0≤λ≤1.
4. wind energy is predicted in short-term in the complex landform region according to claim 1 based on WRF/CFD/SAHDE-RVM couplings
Method, which is characterized in that the step 3) specifically includes：
3.1) CFD software obtains the detailed place information in target complex landform region, and corresponding mould is carried out with pre-processing software
Type is established, and is established computational domain and is set boundary condition；
3.2) will be divided into after the boundary speed nondimensionalization of input x to and y to；Wherein, x to speed represent that y is to speed with u
It is represented with v；Boundary wind speed is denoted as u under different wind angle θB=Vcos θ, vB=Vsin θ, B represent border；Calculate dimensionless
Information of flow after change is represented with f (i, j, k)；
3.3) air speed data (u of the corresponding airspeedometer crosswind point under its observed altitude is recordedA,vA) and corresponding whole audience ground surface
Wind speed information；
3.4) by the boundary speed of the nondimensionalization described in step 3.2) and information of flow and the dimensionless described in step 3.3)
Under the conditions of airspeedometer point (uA,vA) associated；
3.5) multiple (u obtained according to numerical simulation under different inlet velocitiesA,vA) as a result, its is unitizationIt is denoted as (UA,VA)；
3.6) by (UA,VA) point and it includes boundary speed information and information of flow insertion APG in, draw initial APG.Wherein,
APG only shows the data of airspeedometer actual measurement, but wherein each point contains corresponding border wind speed information and flow field letter
Breath.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711218419.4A CN108062595B (en) | 2017-11-28 | 2017-11-28 | WRF/CFD/SAHDE-RVM coupling-based short-time wind energy prediction method for complex landform area |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711218419.4A CN108062595B (en) | 2017-11-28 | 2017-11-28 | WRF/CFD/SAHDE-RVM coupling-based short-time wind energy prediction method for complex landform area |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108062595A true CN108062595A (en) | 2018-05-22 |
| CN108062595B CN108062595B (en) | 2021-09-28 |
Family
ID=62135073
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711218419.4A Expired - Fee Related CN108062595B (en) | 2017-11-28 | 2017-11-28 | WRF/CFD/SAHDE-RVM coupling-based short-time wind energy prediction method for complex landform area |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108062595B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108776745A (en) * | 2018-06-27 | 2018-11-09 | 广东电网有限责任公司电力科学研究院 | A kind of complex topography wind field analogy method and device based on WRF and Fluent couplings |
| CN109978275A (en) * | 2019-04-03 | 2019-07-05 | 中南大学 | A kind of extreme wind wind speed forecasting method and system mixing CFD and deep learning |
| CN110009037A (en) * | 2019-04-03 | 2019-07-12 | 中南大学 | A kind of engineering wind speed Forecasting Approach for Short-term and system based on physical message coupling |
| CN110188430A (en) * | 2019-05-20 | 2019-08-30 | 国网陕西省电力公司电力科学研究院 | Exchange the 330kV pendency equal laminated structure optimum design method of tower sectional composite insulator |
| CN110347671A (en) * | 2019-04-09 | 2019-10-18 | 清华大学深圳研究生院 | The method for constructing wind energy on the sea data bank and offshore wind power generation amount database |
| CN111325376A (en) * | 2018-12-14 | 2020-06-23 | 北京金风科创风电设备有限公司 | Wind speed prediction method and device |
| CN112685977A (en) * | 2021-01-14 | 2021-04-20 | 国家气候中心 | Wind power plant wind resource unsteady numerical simulation method and device |
| CN115936474A (en) * | 2022-10-17 | 2023-04-07 | 中南大学 | Method for setting strong wind monitoring points along high-speed railway |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102930177A (en) * | 2012-11-23 | 2013-02-13 | 南京信息工程大学 | Wind speed forecasting method based on fine boundary layer mode for wind farm in complex terrain |
| US20160350453A1 (en) * | 2015-05-29 | 2016-12-01 | Elizabeth Walls | Method of evaluation wind flow based on conservation of momentum and variation in terrain |
| CN106485603A (en) * | 2016-10-21 | 2017-03-08 | 国家电网公司 | A kind of short-term wind speed forecasting method being applied to wind-power electricity generation |
-
2017
- 2017-11-28 CN CN201711218419.4A patent/CN108062595B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102930177A (en) * | 2012-11-23 | 2013-02-13 | 南京信息工程大学 | Wind speed forecasting method based on fine boundary layer mode for wind farm in complex terrain |
| US20160350453A1 (en) * | 2015-05-29 | 2016-12-01 | Elizabeth Walls | Method of evaluation wind flow based on conservation of momentum and variation in terrain |
| CN106485603A (en) * | 2016-10-21 | 2017-03-08 | 国家电网公司 | A kind of short-term wind speed forecasting method being applied to wind-power electricity generation |
Non-Patent Citations (2)
| Title |
|---|
| 方艳莹等: "基于WRF和CFD软件结合的风能资源数值模拟试验研究", 《气象》 * |
| 肖仪清等: "复杂地形风能评估的CFD方法", 《华南理工大学学报(自然科学版)》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108776745A (en) * | 2018-06-27 | 2018-11-09 | 广东电网有限责任公司电力科学研究院 | A kind of complex topography wind field analogy method and device based on WRF and Fluent couplings |
| CN111325376A (en) * | 2018-12-14 | 2020-06-23 | 北京金风科创风电设备有限公司 | Wind speed prediction method and device |
| CN109978275A (en) * | 2019-04-03 | 2019-07-05 | 中南大学 | A kind of extreme wind wind speed forecasting method and system mixing CFD and deep learning |
| CN110009037A (en) * | 2019-04-03 | 2019-07-12 | 中南大学 | A kind of engineering wind speed Forecasting Approach for Short-term and system based on physical message coupling |
| CN110009037B (en) * | 2019-04-03 | 2020-10-27 | 中南大学 | Short-term engineering wind speed prediction method and system based on physical information coupling |
| CN110347671A (en) * | 2019-04-09 | 2019-10-18 | 清华大学深圳研究生院 | The method for constructing wind energy on the sea data bank and offshore wind power generation amount database |
| CN110188430A (en) * | 2019-05-20 | 2019-08-30 | 国网陕西省电力公司电力科学研究院 | Exchange the 330kV pendency equal laminated structure optimum design method of tower sectional composite insulator |
| CN110188430B (en) * | 2019-05-20 | 2022-09-20 | 国网陕西省电力公司电力科学研究院 | Optimized design method for voltage-sharing structure of sectional type composite insulator of alternating-current 330kV suspension tower |
| CN112685977A (en) * | 2021-01-14 | 2021-04-20 | 国家气候中心 | Wind power plant wind resource unsteady numerical simulation method and device |
| CN112685977B (en) * | 2021-01-14 | 2023-12-08 | 国家气候中心 | Wind power plant wind resource unsteady value simulation method and device |
| CN115936474A (en) * | 2022-10-17 | 2023-04-07 | 中南大学 | Method for setting strong wind monitoring points along high-speed railway |
| CN115936474B (en) * | 2022-10-17 | 2023-08-08 | 中南大学 | Method for setting strong wind monitoring points along high-speed railway |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108062595B (en) | 2021-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108062595A (en) | Complex landform region wind energy Forecasting Methodology in short-term based on WRF/CFD/SAHDE-RVM couplings | |
| Wu et al. | Evaluation of stacking and blending ensemble learning methods for estimating daily reference evapotranspiration | |
| Azlan et al. | Review on optimisation methods of wind farm array under three classical wind condition problems | |
| Alessandrini et al. | A novel application of an analog ensemble for short-term wind power forecasting | |
| Labati et al. | A decision support system for wind power production | |
| Penalba et al. | A data-driven long-term metocean data forecasting approach for the design of marine renewable energy systems | |
| CN106251027B (en) | Electric load probability density Forecasting Methodology based on fuzzy support vector quantile estimate | |
| Higgins et al. | Application of artificial intelligence to urban wind energy | |
| CN109325633A (en) | A kind of weather forecast set member choosing method and system | |
| CN105225006B (en) | A kind of short-term wind-electricity power nonparametric probability forecasting method | |
| CN105846425A (en) | Economic dispatching method based on general wind power forecasting error model | |
| Hu et al. | A novel approach for wind farm micro-siting in complex terrain based on an improved genetic algorithm | |
| CN112036658A (en) | Wind power real-time prediction method based on random forgetting long-term and short-term memory | |
| Li et al. | A method of rainfall runoff forecasting based on deep convolution neural networks | |
| CN116805439A (en) | Drought prediction method and system based on artificial intelligence and atmospheric circulation mechanism | |
| Baran et al. | Calibration of wind speed ensemble forecasts for power generation | |
| CN105138729B (en) | Based on PSO GRNN wind power plant wind turbine defect air speed value fill methods | |
| Shin et al. | Effects of spatiotemporal correlations in wind data on neural network-based wind predictions | |
| Cabezon et al. | Comparison of methods for power curve modelling | |
| Khosravi et al. | Wind farm power uncertainty quantification using a mean-variance estimation method | |
| Ramedani et al. | A method based on neural networks for generating solar radiation map | |
| Hassanian et al. | Iceland wind farm assessment case study and development: An empirical data from wind and wind turbine | |
| CN114565136B (en) | Air quality prediction optimization method based on generation countermeasure network | |
| Pan et al. | Probability forecast of storm surge levels in the Changjiang Estuary induced by tropical cyclones based on the Error-Estimation Ensemble method | |
| Wang et al. | A numerical model based on prior distribution fuzzy inference and neural networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210928 |