CN109918692A - A kind of statistical model method for building up and device based on numerical simulation - Google Patents
A kind of statistical model method for building up and device based on numerical simulation Download PDFInfo
- Publication number
- CN109918692A CN109918692A CN201811323208.1A CN201811323208A CN109918692A CN 109918692 A CN109918692 A CN 109918692A CN 201811323208 A CN201811323208 A CN 201811323208A CN 109918692 A CN109918692 A CN 109918692A
- Authority
- CN
- China
- Prior art keywords
- data
- acquisition system
- model
- numerical simulation
- data acquisition
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000004088 simulation Methods 0.000 title claims abstract description 48
- 238000013179 statistical model Methods 0.000 title claims abstract description 36
- 238000012549 training Methods 0.000 claims abstract description 26
- 239000012530 fluid Substances 0.000 claims abstract description 10
- 238000012546 transfer Methods 0.000 claims description 13
- 238000013519 translation Methods 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 210000000746 body region Anatomy 0.000 claims description 4
- 238000012512 characterization method Methods 0.000 claims description 2
- 238000013461 design Methods 0.000 claims description 2
- 238000004458 analytical method Methods 0.000 abstract description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000005457 optimization Methods 0.000 description 4
- 230000005611 electricity Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 238000011144 upstream manufacturing Methods 0.000 description 1
Landscapes
- Wind Motors (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of title of the present invention are as follows: statistical model method for building up and device based on numerical simulation.Technical field is related to modeling field, in particular to a kind of statistical model method for building up and device based on numerical simulation.Include: that multiple Fluid Mechanics Computation (CFD) numerical simulation is carried out to same simulated domain using different boundary parameter combination producings different model parameter and boundary condition value, generates multiple analog result；The set of multiple analog result is extracted into a subset and is combined together as a verifying sample and training sample with the statistical model of self-learning capability from different boundary parameters, and statistical model is trained, implementation model is established.Invention, which is realized, replaces luid mechanics computation model with the statistical model based on numerical simulation, method for numerical simulation is replaced with analytic method, and guarantee that the calculated result of two methods is almost the same, it solves the problems, such as largely to cause operand excessive using Computational fluid mechanics numerical simulation method in problem solving.
Description
Technical field
The present invention relates to modeling fields, in particular to a kind of statistical model method for building up based on numerical simulation and
Device.
Background technique
For the complex flowfield that can not be solved with analytic method, Fluid Mechanics Computation (CFD) Numerical-Mode is generally used
Quasi- method.But the method operand of CFD numerical simulation is big, calculation resources require height, and operation time is long, are needing to same
Often due to operand is excessive and exceeds the operational capability of calculation resources in the case that one simulated domain solves repeatedly.
Particularly, in wind power plant planning and designing, need to consider wind-power electricity generation group wake flow to leeward direction wind-driven generator group
Influence, and then the influence to wind power plant total power generation, so establishing high-precision wake model as Accurate Prediction generated energy
An important factor for.In currently used a variety of wake models, actuation disc wake model is generally considered to be a kind of high-precision
Wake model, its modeling method is the air dynamic behaviour according to wind regime and earth's surface, and according to wind power generating set
The influence of mechanical property and its stream field carries out numerical simulation.In addition, maximum most in order to automatically derive Wind turbines generated energy
Excellent layout, generallys use the algorithm based on route searching, its feature be constantly generate new placement scheme and existing program into
Row compares, that is, to carry out a large amount of tentative calculations.If using traditional actuation disc in the optimization of wind-driven generator autoplacement
Wake model, it is necessary to numerical simulation be carried out to same wind field region in each tentative calculation, thus cause excessively high overhead
So that exceeding the operational capability of calculation resources.
There are no appearance in the world at present replaces luid mechanics computation model with the statistical model based on numerical simulation, uses base
The case of method for numerical simulation is replaced in the analytic method of statistical model.Particularly, Automatic Optimal is laid out in wind power generating set
In there are no using actuation disc wake model case.
The present invention, which realizes, replaces luid mechanics computation model with the statistical model based on numerical simulation under certain condition,
Method for numerical simulation is replaced with the analytic method based on statistical model, and guarantees that the calculated result of two methods is almost the same, is solved
Determined certain flow field problems solution in largely cause operand excessive using Computational fluid mechanics numerical simulation method problem.
Particularly, it when the modeling method is modeled for wind power generating set wake flow, realizes the actuating circle based on analytic method
Disk wake model optimizes for wind power generating set autoplacement, substantially increases layout optimization speed.
Summary of the invention
The embodiment of the invention provides a kind of statistical model method for building up and device based on numerical simulation.Particularly, also
It includes how for the modeling method to be used for the modeling of wind power generating set wake flow, solve at least in the solution of certain flow field problems
A large amount of problems for causing operand excessive using Computational fluid mechanics numerical simulation method, particularly, when the modeling method is used
When the modeling of wind power generating set wake flow, realizes and the actuation disc wake model based on analytic method is used for wind-driven generator
Group autoplacement optimization, substantially increases layout optimization speed.
According to the one aspect of inventive embodiments, a kind of statistical model method for building up based on numerical simulation, the party are provided
Method includes: the different boundary parameter combination of design, and the parameter group is combined into the first data acquisition system；It is generated according to the first data acquisition system
The different boundary condition of multiple groups and boundary parameter carry out multiple Fluid Mechanics Computation (CFD) numerical simulation to same simulated domain,
Multiple analog result data are generated, these data are the second data set；A subset conduct is extracted in the second data set
Third data acquisition system；One statistical model with self-learning capability of setting；According to the first data acquisition system and the second data set
Data generate the training of statistical model and verify data sample carries out statistical model training, determine model parameter, complete modeling.
Further, different boundary parameter combinations is designed, the parameter group is combined into the first data acquisition system, comprising: works as institute
When stating modeling method for the modeling of wind power generating set wake flow, every group of data include at least following data in first set data
: simulation roughness of ground surface z0, simulation entrance hub height vertical wind speed u0, all groups of data are by different z0And u0Traversal group
Symphysis at.
Further, multiple groups different boundary condition and boundary parameter are generated to same simulation region according to the first data acquisition system
Domain carries out multiple Fluid Mechanics Computation (CFD) numerical simulation, generates multiple analog result data, these data are the second data set
Close, comprising: when the modeling method is modeled for wind power generating set wake flow, the same simulated domain by first area and
Second area is constituted, and the first area is the region for characterizing wind-driven generator surrounding air flow field, and the second area is embedding
Cover the region that wind-driven generator group object air dynamic behaviour is characterized in first area.First area is rectangular body region,
Including 4 sides interface, a following interface and top interface.Centered on the line at following interface the right and left midpoint
Line；Second area is collar plate shape region, and disc centre point is located at wind-power electricity generation wheel above the following interface centerlines in first area
At hub height, disk normal and centerline parallel, disk diameter are wind power generator impeller diameter, and one perpendicular to center line
Side interface is entrance boundary face, and following interface is without sliding wall surface, and top interface is sliding wall surface, and other boundary faces are outlet
Boundary face；The size of mesh opening of second area is smaller than the size of mesh opening of first area；In the movement of the CFD governing equation of second area
Source item is increased according to the mechanical property of wind-driven generator and air dynamic behaviour in equation；According to the first data acquisition system according to big
Gas boundary layer theory generates each group CFD numerical simulation entrance boundary condition, and the direction of wind speed is given birth to perpendicular to entrance boundary face
At the boundary condition of other boundary faces；Multiple CFD numerical simulation is carried out to the simulated domain with each group boundary condition, it is raw
At multiple numerical simulation result, the second data set is constituted, the second data set is made of multi-group data, wherein every group of data are extremely
It less include speed, position coordinates, turbulence transfer kinetic energy.
Further, a subset is extracted in the second data set as third data acquisition system, comprising: when the modeling
When method is modeled for wind power generating set wake flow, coordinate origin is moved to immediately below hub of wind power generator on center line,
For third data acquisition system by including that the data group of different representative locations coordinates is constituted, the representative locations coordinate is position coordinates
A part, wherein every group of data include at least velocity component u 'x、u′y、u′z, representative locations coordinate x, y, z, turbulence transfer
Kinetic energy k '.
Further, the statistical model with self-learning capability is set, comprising: the statistical model is neural network
Back propagation model, model structure include input layer, middle layer and output layer, and mode input data include at least: hub height
Vertical wind speed u0, roughness of ground surface z0, representative locations coordinate x, y, z, model output data at least includes: wind speed component ux、
uy、uz, turbulence transfer kinetic energy k, decay factor a.
Further, the training and verifying of statistical model are generated according to the data of the first data acquisition system and third data acquisition system
Data sample carries out statistical model training, determines model parameter, completes modeling, comprising:
Training input sample is by the representative locations coordinate traversal group symphysis in the first data acquisition system and third data acquisition system
At every group of data include at least u0、z0, each data item of x, y, z；Verifying sample is generated by third data acquisition system, and every group of data include
u′x、u′y、u′z, k ', a ', whereinU is exported to modelx、uy、uz, k, a verified, carry out
Training input sample is consistent with the representative locations coordinate for verifying sample when verifying.Using the model in other coordinate systems
When being predicted, in mode input data, vertical wind speed u0With roughness of ground surface z0It is set as predicted condition value, and by changing coordinates
Position coordinates of the position coordinates of system by translation and when being converted to modeling, make vertical wind speed be oriented parallel to plate way
Line, wind power generating set position are in coordinate origin；After the completion of prediction, position coordinates are converted to by former seat by coordinate inverse transform
Mark system coordinate.
According to the other side of inventive embodiments, a kind of statistical model based on numerical simulation is provided and establishes device, it should
Device includes: the first generation unit, and for designing and generating different boundary parameter combinations, the parameter group is combined into the first data
Set；Second generation unit, for generating multiple groups different boundary condition and boundary parameter to same according to the first data acquisition system
Simulated domain carries out multiple Fluid Mechanics Computation (CFD) numerical simulation, generates multiple analog result data, these data are second
Data acquisition system；Third generation unit, for extracting a subset in the second data set as third data acquisition system；Modeling is single
Member, for establishing that model structure is determining and the uncertain statistical model with learning ability of model parameter；Model training
Unit, for setting the statistical model with self-learning capability, according to the number of the first data acquisition system and the second data set
Statistical model training is carried out according to the training and verify data sample that generate statistical model, model parameter is determined, completes modeling.
Further, when the modeling method is modeled for wind power generating set wake flow, every group in first set data
Data include at least following data item: simulation roughness of ground surface z0, simulation entrance hub height vertical wind speed u0, all groups of data
By different z0And u0Traverse combination producing.
Further, when the modeling method is modeled for wind power generating set wake flow, the second generation unit includes: 1,
First generates subelement, and for generating simulated domain, simulated domain is made of first area and second area, the first area
For the region for characterizing wind-driven generator surrounding air flow field, the second area is to be nested in characterization wind-power electricity generation in first area
The region of machine entity air dynamic behaviour, first area are rectangular body region, including 4 sides interface, a following interface
With top interface.The line at following interface the right and left midpoint is center line；Second area is collar plate shape region, in disk
Heart point is located above the following interface centerlines in first area at hub of wind power generator height, disk normal and centerline parallel,
Disk diameter is wind power generator impeller diameter, and a side interface perpendicular to center line is entrance boundary face, following interface
For wall surface, other boundary faces are outlet border face；The size of mesh opening of second area is smaller than the size of mesh opening of first area；2, simulation
Subelement is controlled, for carrying out derivation, the root in the equation of motion of the CFD governing equation of second area according to governing equation
Increase source item according to the mechanical property and air dynamic behaviour of wind-driven generator；3, boundary condition generates subelement, is used for basis
First data acquisition system generates each group CFD numerical simulation entrance boundary condition according to the neutral line knot theory of atmosphere boundary theory,
And generate the boundary condition of other boundary faces.Multiple CFD Numerical-Mode is carried out to the simulated domain with each group boundary condition
It is quasi-, multiple numerical simulation result is generated, constitutes the second data set, the second data set is made of multi-group data, wherein every group
Data include at least speed, position coordinates, turbulence transfer kinetic energy.
Further, when the modeling method is modeled for wind power generating set wake flow, third generation unit is by coordinate
In origin translation to center line immediately below hub of wind power generator, third data acquisition system is by including different representative locations coordinates
Data group is constituted, and the representative locations coordinate is a part of position coordinates, wherein every group of data include at least velocity component
u′x、u′y、u′z, representative locations coordinate x, y, z, turbulence transfer kinetic energy k '.
Further, when the modeling method is modeled for wind power generating set wake flow, modeling unit further include: setting
One statistical model with self-learning capability, the statistical model are neural network back propagation model, and model structure includes
Input layer, middle layer and output layer, mode input data include at least: hub height vertical wind speed u0, roughness of ground surface z0, generation
Table position coordinates x, y, z, model output data at least include: wind speed component ux、uy、uz, turbulence transfer kinetic energy k, decay factor
a。
Further, when the modeling method is modeled for wind power generating set wake flow, model training unit includes: 1,
Training input sample is established, training input sample is by the representative locations coordinate time in the first data acquisition system and third data acquisition system
Combination producing is gone through, every group of data include at least u0、z0, each data item of x, y, z；2, verifying sample is generated, verifies sample by third number
It is generated according to set, every group of data include u 'x、u′y、u′z, k ', a ', whereinModel is exported
ux、uy、uz, k, a verified, when being verified training input sample with verifying sample representative locations coordinate be consistent；
3, model training.
Further, it is modeled when the modeling method for wind power generating set wake flow and uses the model in other seats
When mark system is predicted, in mode input data, vertical wind speed u0With roughness of ground surface z0It is set as predicted condition value, and will be current
Position coordinates of the position coordinates of coordinate system by translation and when being converted to modeling, make vertical wind speed be oriented parallel to disk
Normal, wind power generating set position are in coordinate origin；After the completion of prediction, position coordinates are converted to by original by coordinate inverse transform
Coordinate system coordinate.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present invention, constitutes part of this application, this hair
Bright illustrative embodiments and their description are used to explain the present invention, and are not constituted improper limitations of the present invention.In the accompanying drawings:
Fig. 1 is the flow chart of modeling and application method according to an embodiment of the present invention；
Fig. 2 is the schematic diagram of simulated domain according to an embodiment of the present invention.
Specific embodiment
It in order to enable those skilled in the art to better understand the solution of the present invention, below will be to the skill in the embodiment of the present invention
Art scheme is clearly and completely described, it is clear that and the described embodiment is only a part of the embodiment of the present invention, without
It is whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not before making creative work
Every other embodiment obtained is put, should fall within the scope of the present invention.
It should be noted that term " first " in description and claims of this specification and Figure of description, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that using in this way
Data be interchangeable under appropriate circumstances, so as to the embodiment of the present invention described herein can in addition to illustrating herein or
Sequence other than those of description is implemented.In addition, term " includes " and " having " and their any deformation, it is intended that cover
Cover it is non-exclusive include, for example, the process, method, system, product or equipment for containing a series of steps or units are not necessarily limited to
Step or unit those of is clearly listed, but may include be not clearly listed or for these process, methods, product
Or other step or units that equipment is intrinsic.
Term is explained
Roughness of ground surface is to indicate earth surface degree of roughness and the characteristic parameter with length dimension.
Wind-driven generator refers to the impeller type generating set for converting wind energy into electric energy, by wheel hub, blade and other parts
It is formed.
Wheel hub, refers to the part of wind-driven generator support blade and other wind energy transforming devices, and hub height is exactly wheel hub
The height of height and impeller center away from ground away from ground.
Impeller diameter is the diameter of the border circular areas scanned when blade rotation.
A specific embodiment of the invention is made below with reference to the embodiment of wind power generating set wake flow modeling further
It is described in detail.
Step 101, different boundary parameter combinations is designed, the parameter group is combined into the first data acquisition system.First set number
Every group of data include at least following data item in: simulation roughness of ground surface z0, simulation entrance hub height vertical wind speed u0, institute
There are group data by different z0And u0Traverse combination producing.Optionally, z0Value can include: 0.001,0.003,0.006,0.01,
0.03,0.06,0.1,0.3,0.6,0.9,1.2,1.5,2.0 meter；u0Value can include: 1,3,5,8,11,13,15,18,
21,25 meter per second.
Step 102, simulated domain is established, the first area is the region for characterizing wind-driven generator surrounding air flow field,
The second area is to be nested in the region that wind-driven generator entity air dynamic behaviour is characterized in first area.First area
For rectangular body region, including 4 sides interface, a following interface and top interface.Following interface the right and left midpoint
Line be center line；Second area is collar plate shape region, and disc centre point is located above the following interface centerlines in first area
At hub of wind power generator height, disk normal and centerline parallel, disk diameter are wind power generator impeller diameter, and one hangs down
It is directly entrance boundary face in the side interface of center line, following interface is without sliding wall surface, and top interface is sliding wall surface, other
Boundary face is outlet border face, and optionally, first area width is 10 to 15 impeller diameters, and length is that 20 to 30 impellers are straight
Diameter is highly 5 to 8 impeller diameters, and second area is 4 to 6 impeller diameters apart from entrance boundary；The grid ruler of second area
Very little smaller than the size of mesh opening of first area, optionally, the size of mesh opening of first area is 0.25 impeller diameter, second area
Size of mesh opening is 0.1 impeller diameter, and the disc height of second area is 1 grid；
Step 103, in the equation of motion of the CFD governing equation of second area according to the mechanical property of wind-driven generator and
Air dynamic behaviour increases source item, optionally, added source item are as follows:Wherein FiFor second area
The source item of i-th of grid, i.e. wind are come power from i-th of grid of second area to component that generate on；CTIt is U for upstream wind speed∞
When thrust coefficient, by wind-driven generator manufacture producer provide；ρ is atmospheric density；ΔAiIt is met for i-th of grid of second area
The area in wind face,Wherein A is the second area windward side gross area,Wherein D is wind-driven generator
Impeller diameter；V is the total volume of second area；ΔViIt is the volume of i-th of grid；
Step 104, each group CFD numerical simulation entrance boundary is generated according to atmosphere boundary theory according to the first data acquisition system
Condition, and generate the boundary condition of other boundary faces.Optionally, top interface is set as sliding wall surface, and following interface is set as
Without sliding wall surface, using k-e turbulence model, the speed and turbulence transfer kinetic energy and turbulent diffusivity for exporting side boundaries are set as zero
Gradient organizes boundary parameter z to Mr. Yu0And u0, entrance velocity u are as follows:Wherein z is ordinate, u*For friction speed
Degree,zhFor hub height；Turbulence transfer kinetic energy k isTurbulence dissipation rate ε are as follows:
Step 105, multiple CFD numerical simulation is carried out to the simulated domain with each group boundary condition, generated multiple
Numerical simulation result constitutes the second data set.The second data set is made of multi-group data, wherein every group of data include at least
Speed, position coordinates, turbulence transfer kinetic energy.Optionally, solver can not press solver using Simple stable state, and Equations of Turbulence is adopted
With k-e Equations of Turbulence；
Step 106, a subset is extracted in the second data set to translate as third data acquisition system, and by coordinate origin
Immediately below to hub of wind power generator on center line.Third data acquisition system is by the data group structure including different representative locations coordinates
At the representative locations coordinate is a part of position coordinates, wherein every group of data include at least speed ux0、uy0、uz0, generation
Table position coordinates x0、y0、z0, turbulence transfer kinetic energy k0.Optionally, representative locations coordinate may include following data combination: x
Coordinate includes 1D, 2D, 3D, 5D, 7D, 10D, 13D, 16D, 20D, y-coordinate include 0,0.2D, 0.5D, 1D, 1.5D, 2D, 2.5D,
3D, 4D, 5D, z coordinate include: zh、zh±0.2D、zh±0.5D、zh±0.8D、zh±1.2D；
Step 107, neural network back propagation model is set up.Model structure includes input layer, middle layer and output layer, can
Selection of land, input layer include 6 nodes, and middle layer includes 12 nodes, and output layer includes 5 nodes.Mode input data are the
The combination of one data acquisition system and representative locations coordinate, group model input data every in this way include: hub height vertical wind speed u,
Roughness of ground surface z0, representative locations coordinate x, y, z, model output data includes wind speed ux、uy、uz, turbulence transfer kinetic energy k, decline
Subtracting coefficient a.
Step 108, model training is carried out, comprising: 1, training input sample is established, training input sample is by the first data set
It closes and is generated with the representative locations in third data acquisition system, every group of data include at least u0、z0, each data item of x, y, z；2, it generates
Sample is verified, verifying sample is generated by third data acquisition system, and every group of data include u 'x、u′y、u′z, k ', a ', whereinU is exported to modelx、uy、uz, k, a verified, when being verified training input sample and verifying
The representative locations coordinate of sample is consistent；3, model training.
Step 109, when using the model when other coordinate systems are predicted, in mode input data, vertical wind speed
u0With roughness of ground surface z0It is set as predicted condition value, and the position coordinates of current coordinate system are converted to and are built by translation and rotation
Position coordinates when mould, make vertical wind speed be oriented parallel to disk normal, and wind power generating set position is in coordinate origin；Prediction
After the completion, position coordinates are converted to by former coordinate system coordinate by coordinate inverse transform.
Claims (6)
1. a kind of statistical model method for building up and device based on numerical simulation characterized by comprising
Different boundary parameter combinations is designed, the parameter group is combined into the first data acquisition system；
Multiple groups different model parameter and boundary condition is generated according to the first data acquisition system repeatedly to count same simulated domain
Fluid operator mechanics (CFD) numerical simulation, generates multiple analog result data, these data are the second data set；
A subset is extracted in the second data set as third data acquisition system；
One statistical model with self-learning capability of setting；
The training of statistical model is generated according to the data of the first data acquisition system and the second data set and verify data sample carries out
Statistical model training, determines model parameter, completes modeling.
2. according to the method for claim 1, it is characterised in that: design different boundary parameter combinations, the parameter combination
For the first data acquisition system, comprising:
When the modeling method is modeled for wind power generating set wake flow, in the first data acquisition system every group of data include at least with
Lower data item: simulation roughness of ground surface z0, simulation entrance hub height vertical wind speed u0, all groups of data are by different z0And u0
Traverse combination producing.
3. according to the method for claim 1, it is characterised in that: generate the different model of multiple groups according to the first data acquisition system and join
Several and boundary condition carries out multiple Fluid Mechanics Computation (CFD) numerical simulation to same simulated domain, generates multiple analog result
Data, these data are the second data set, comprising: when the modeling method is modeled for wind power generating set wake flow:
The same simulated domain is made of first area and second area, and the first area is around characterization wind-driven generator
The region of air flow field, the second area are to be nested in first area to characterize wind-driven generator group object aerodynamics spy
Property region, first area is rectangular body region, including 4 sides interface, a following interface and top interface；Below
The line at interface the right and left midpoint is center line；Second area is collar plate shape region, and disc centre point is located under first area
Above boundary face center line at hub of wind power generator height, disc face normal and centerline parallel, disk diameter are wind-force hair
Electric motor with vane wheel diameter, a side interface perpendicular to center line are entrance boundary face, and following interface is without sliding wall surface, top
Interface is sliding wall surface, and other boundary faces are outlet border face；
The size of mesh opening of second area is smaller than the size of mesh opening of first area；
It is special according to the mechanical property and aerodynamics of wind-driven generator in the equation of momentum of the CFD governing equation of second area
Property increases source item；
Each group CFD numerical simulation entrance boundary condition is generated according to atmosphere boundary theory according to the first data acquisition system, wind speed
Direction generates the boundary condition of other boundary faces perpendicular to entrance boundary face；
Multiple CFD numerical simulation is carried out to the simulated domain with each group boundary condition, generates multiple numerical simulation result,
The second data set is constituted, the second data set is made of multi-group data, wherein every group of data include at least speed, position is sat
Mark, each data item of turbulence transfer kinetic energy.
4. according to the method for claim 1, it is characterised in that: extract a subset in the second data set as third
Data acquisition system, comprising:
When the modeling method is modeled for wind power generating set wake flow, coordinate origin is being moved into hub of wind power generator just
On lower central line, third data acquisition system is by including that the data group of different representative locations coordinates is constituted, the representative locations
Coordinate is a part of position coordinates, wherein every group of data include at least velocity component u 'x、u′y、u′z, representative locations coordinate
X, y, z, turbulence transfer kinetic energy k '.
5. according to the method for claim 1, it is characterised in that: one statistical model with self-learning capability of setting, packet
It includes:
When the modeling method is modeled for wind power generating set wake flow, the statistical model is neural network backpropagation mould
Type, model structure include input layer, middle layer and output layer, and mode input data include at least: hub height vertical wind speed u0、
Roughness of ground surface z0, representative locations coordinate x, y, z, model output data at least includes: wind speed component ux、uy、uz, turbulent flow pass
Defeated kinetic energy k, decay factor a.
6. according to the method for claim 1, it is characterised in that: according to the data of the first data acquisition system and third data acquisition system
The training and verify data sample for generating statistical model carry out statistical model training, determine model parameter, complete modeling, comprising:
When the modeling method is modeled for wind power generating set wake flow:
Training input sample traverses combination producing by the representative locations coordinate in the first data acquisition system and third data acquisition system, often
Group data include at least u0、z0, each data item of x, y, z；
Verifying sample is generated by third data acquisition system, and every group of data include u 'x、u′y、u′z, k ', a ', whereinU is exported to modelx、uy、uz, k, a verified, when being verified training input sample and verifying sample
This representative locations coordinate is consistent；
Using the model when other coordinate systems are predicted, in mode input data, vertical wind speed u0With roughness of ground surface z0
Predicted condition value, and the position coordinates by the position coordinates of current coordinate system by translation and when being converted to modeling are set as,
Vertical wind speed is set to be oriented parallel to disk normal, wind power generating set position is in coordinate origin；After the completion of prediction, pass through coordinate
Position coordinates are converted to former coordinate system coordinate by inverse transform.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811323208.1A CN109918692A (en) | 2018-11-08 | 2018-11-08 | A kind of statistical model method for building up and device based on numerical simulation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811323208.1A CN109918692A (en) | 2018-11-08 | 2018-11-08 | A kind of statistical model method for building up and device based on numerical simulation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109918692A true CN109918692A (en) | 2019-06-21 |
Family
ID=66959667
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811323208.1A Pending CN109918692A (en) | 2018-11-08 | 2018-11-08 | A kind of statistical model method for building up and device based on numerical simulation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109918692A (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080177518A1 (en) * | 2007-01-18 | 2008-07-24 | Cfd Research Corporation | Integrated Microfluidic System Design Using Mixed Methodology Simulations |
| CN102663251A (en) * | 2012-04-09 | 2012-09-12 | 华北电力大学 | Physical prediction method for wind power station power based on computational fluid mechanics model |
| CN104699936A (en) * | 2014-08-18 | 2015-06-10 | 沈阳工业大学 | Sector management method based on CFD short-term wind speed forecasting wind power plant |
| CN106770928A (en) * | 2016-11-11 | 2017-05-31 | 陕西省地质环境监测总站 | The assay method of the lower WATER LEVEL CHANGES threshold of ecology fragility mining area vegetation constraint |
| CN107066731A (en) * | 2017-04-13 | 2017-08-18 | 中南大学 | The method that gas distributional pattern in two phase flow is recognized according to numerical simulation result |
| CN107273747A (en) * | 2017-05-22 | 2017-10-20 | 中国人民公安大学 | The method for extorting software detection |
| US20180174068A1 (en) * | 2016-12-15 | 2018-06-21 | Sintef Tto As | Method and process for providing a subject-specific computational model used for treatment of cardiovascular diseases |
| CN108732931A (en) * | 2018-05-17 | 2018-11-02 | 北京化工大学 | A kind of multi-modal batch process modeling method based on JIT-RVM |
-
2018
- 2018-11-08 CN CN201811323208.1A patent/CN109918692A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080177518A1 (en) * | 2007-01-18 | 2008-07-24 | Cfd Research Corporation | Integrated Microfluidic System Design Using Mixed Methodology Simulations |
| CN102663251A (en) * | 2012-04-09 | 2012-09-12 | 华北电力大学 | Physical prediction method for wind power station power based on computational fluid mechanics model |
| CN104699936A (en) * | 2014-08-18 | 2015-06-10 | 沈阳工业大学 | Sector management method based on CFD short-term wind speed forecasting wind power plant |
| CN106770928A (en) * | 2016-11-11 | 2017-05-31 | 陕西省地质环境监测总站 | The assay method of the lower WATER LEVEL CHANGES threshold of ecology fragility mining area vegetation constraint |
| US20180174068A1 (en) * | 2016-12-15 | 2018-06-21 | Sintef Tto As | Method and process for providing a subject-specific computational model used for treatment of cardiovascular diseases |
| CN107066731A (en) * | 2017-04-13 | 2017-08-18 | 中南大学 | The method that gas distributional pattern in two phase flow is recognized according to numerical simulation result |
| CN107273747A (en) * | 2017-05-22 | 2017-10-20 | 中国人民公安大学 | The method for extorting software detection |
| CN108732931A (en) * | 2018-05-17 | 2018-11-02 | 北京化工大学 | A kind of multi-modal batch process modeling method based on JIT-RVM |
Non-Patent Citations (3)
| Title |
|---|
| QIDUN MAULANA B.S. ET AL.: ""Numerical simulation of the effect of flow direction angle at inlet boundary condition on flow characteristics and performance of axial flow hydraulic propeller turbine runner "", 《2015 ICSEEA》 * |
| 蔡少鹏: ""树林地貌对高速公路路堤区域流场的影响"", 《中国优秀硕士学位论文全文数据库 (基础科学辑)》, pages 034 - 150 * |
| 马申: ""基于CFD风环境模拟长春市居住小区布局优化研究"", 《中国优秀硕士学位论文全文数据库 (工程科技Ⅱ辑)》, pages 038 - 23 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Crespo et al. | Survey of modelling methods for wind turbine wakes and wind farms | |
| US9399981B2 (en) | Methods and systems for comparing vertical axis turbine arrays and providing configurations thereof | |
| CN108416075A (en) | Wind energy conversion system aerodynamics model modeling method based on CFD technologies | |
| CN102663251A (en) | Physical prediction method for wind power station power based on computational fluid mechanics model | |
| Wang et al. | Mesoscale simulations of a real onshore wind power base in complex terrain: Wind farm wake behavior and power production | |
| CN106407577B (en) | A kind of improvement actuator surface method for establishing model of emulation wind turbine wake flow | |
| Elsakka et al. | Comparison of the computational fluid dynamics predictions of vertical axis wind turbine performance against detailed pressure measurements | |
| CN113627101A (en) | Wind turbine wake flow simulation method based on improved AD/RSM model | |
| Liu et al. | Mathematical model for the analysis of wind-turbine wakes | |
| SnehalNarendrabhai et al. | Numerical simulation of flow through INVELOX wind turbine system | |
| CN106919730B (en) | Wind power plant wake flow determination method adopting wind speed attenuation factor | |
| CN109255361B (en) | Tidal current energy power generation field unit layout method considering infeasible area | |
| Wu et al. | A synchronization methodology for 3D offshore wind farm layout optimization with multi-type wind turbines and obstacle-avoiding cable network | |
| Hameed et al. | Shape optimization of a shrouded Archimedean-spiral type wind turbine for small-scale applications | |
| Putra et al. | Multilevel Diffuser Augmented for Horizontal Axis Wind Turbine | |
| Wang et al. | Impeller modeling and analysis based on UG NX/KF and Fluent | |
| Li et al. | Equilateral-triangle mesh for optimal micrositing of wind farms | |
| Belloni et al. | A numerical analysis of bidirectional ducted tidal turbines in yawed flow | |
| CN108536907A (en) | A kind of Wind turbines far field wake flow Analytic modeling method based on simplified momentum theorem | |
| CN109918692A (en) | A kind of statistical model method for building up and device based on numerical simulation | |
| CN109992890B (en) | Coupling model and coupling calculation method for high-speed train-tornado | |
| Zhang et al. | Numerical investigation and wind tunnel validation on near-wake vortical structures of wind turbine blades | |
| Riva et al. | Implementation and application of the actuator line model by OpenFOAM for a vertical axis wind turbine | |
| Deskos | Numerical simulations of wind turbine wakes | |
| da Silva | CFD Analysis of the Air Flow Through Horizontal Axis Wind Turbines |
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 |