CN107391871A - A kind of space lattice deformation method based on parallelization RBF - Google Patents

A kind of space lattice deformation method based on parallelization RBF Download PDF

Info

Publication number
CN107391871A
CN107391871A CN201710655495.5A CN201710655495A CN107391871A CN 107391871 A CN107391871 A CN 107391871A CN 201710655495 A CN201710655495 A CN 201710655495A CN 107391871 A CN107391871 A CN 107391871A
Authority
CN
China
Prior art keywords
rbf
parallelization
object plane
method based
point
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
Application number
CN201710655495.5A
Other languages
Chinese (zh)
Inventor
赵钟
常兴华
马戎
何磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Computational Aerodynamics Institute of China Aerodynamics Research and Development Center
Original Assignee
Computational Aerodynamics Institute of China Aerodynamics Research and Development Center
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Computational Aerodynamics Institute of China Aerodynamics Research and Development Center filed Critical Computational Aerodynamics Institute of China Aerodynamics Research and Development Center
Priority to CN201710655495.5A priority Critical patent/CN107391871A/en
Publication of CN107391871A publication Critical patent/CN107391871A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00Computer-aided design [CAD]
    • G06F30/20Design optimisation, verification or simulation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00Computer-aided design [CAD]
    • G06F30/10Geometric CAD
    • G06F30/15Vehicle, aircraft or watercraft design

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Geometry (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Hardware Design (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a kind of space lattice deformation method based on parallelization RBF, comprise the following steps:Step 1: establish parallel architecture, the grid of computational fields is subjected to subregion, the corresponding process of each blockette, and each process is had the backup of a global object plane unit;Step 2: traveling through all spatial points in each process, parallel RBF deformations are carried out, wherein, each point that each spatial point is concentrated using its control point performs serial RBF deformations as control point, and solution obtains the deflection of spatial point.Compared with prior art, the positive effect of the present invention is:Versatility is good, and the parallelization RBF that the present invention uses has stronger versatility, is suitable for the application fields such as data interpolating, distortion of the mesh;Method is simple, it is only necessary to a small amount of change, it is possible to by the RBF method parallelizations of existing serialization.

Description

A kind of space lattice deformation method based on parallelization RBF
Technical field
The present invention relates to one kind based on parallelization RBF (Parallel Radial Basis Function, with Space lattice deformation method down referred to as " PRBF ").
Background technology
With the continuous development of modern numerical analogy method, the physical problem increasingly approaching to reality profile of numerical simulation, Multiple physical field, unsteady simulation are developed into from former single physical field, permanent simulation.Fluid Mechanics Computation (CFD), as A kind of effective flight vehicle aerodynamic simulation means, have also obtained more and more extensive application, have been mainly shown as unsteady simulation side Method has obtained increasing engineer applied.Real physical world is essentially all the UNSTEADY FLOW under dynamic boundary condition. In the problems such as Multi-bodies Separation of aircraft and boost motor, fluid structurecoupling problem, wing deformation, object plane profile can constantly become with the time Change, it is necessary to the mesh point in space also be changed therewith, i.e. distortion of the mesh.
Currently conventional space lattice deformation method is mainly based upon SPRING METHOD OF STONE and Radial Basis Function Method, former approach letter It is single, but poor universality;The latter's versatility is good, there is larger technical advantage in distortion of the mesh field.
RBF (Radial Basis Function, hereinafter referred to as " RBF ") method, it is that one kind utilizes distance The method that basic function carries out data interpolating, is widely used in engineering fields such as computational mathematics, numerical simulations.For example, entering Promoting the circulation of qi is dynamic/heat/structure multi- scenarios method simulation in, for the flow field variable interpolation between space cell and object plane unit;In bullet Property the Moving Boundary Problem such as deformation, Multi-bodies Separation in, by the way that the deformation of the point on object plane is interpolated into space, realize that space lattice becomes Shape.
The general principle of the existing space lattice deformation method based on RBF is:
Here, N is that (control point subset is by the multiple object plane deformation points nearest from spatial point for the number at control point subset midpoint Form), f (r) is space midpoint Pfield,iDisplacement, r Pfield,iPosition, riFor i-th of object plane control point Psurface,i Initial position, | | r-ri| | represent Pfield,iAnd Psurface,iBetween Euclidean distance,Selected RBF is represented, wiRepresent i-th of object plane control point Psurface,iThe weight coefficient at place, by taking x directions as an example, the coefficient meets following relational expression:
Wherein, Δ xiFor i-th of object plane control point Psurface,iDeflection,Basic functionGeneral choosing It is defined as with C2 functions:
By formula (1), by the positional displacement interpolation of object plane point to space, the deflection of spatial point is obtained.
It is so far, several although Radial Basis Function Method has obtained a large amount of uses in mesh deformation technique The application of the overwhelming majority is all based on serial method, and this is due to the matrix operation being related in RBF methods in formula (2), And the matrix in the formula right-hand vector is made up of the object plane point in global calculation domain so that RBF methods are difficult to parallelization.Although at present There are some preliminary Parallel Algorithms in academia, but be typically all to use third party's math libraries such as PETs, LAPACK, , the problem of robustness is poor, efficiency is low, poor universality be present in the matrix progress parallelization in formula (2) inversion operation.
Due to being limited by computer resource, the Radial Basis Function Method of serialization make it that it is solid in Multi-bodies Separation, stream It is very restricted in the numerical simulation problems such as coupling.
The content of the invention
In order to overcome the disadvantages mentioned above of prior art, the present invention proposes a kind of space based on parallelization RBF Grid deforming method.
The technical solution adopted for the present invention to solve the technical problems is:A kind of space based on parallelization RBF Grid deforming method, comprise the following steps:
Step 1: establishing parallel architecture, the grid of computational fields is subjected to subregion, each blockette corresponds to a process, and Each process is set to have the backup of a global object plane unit;
Step 2: traveling through all spatial points in each process, parallel RBF deformations are carried out, wherein, each spatial point is with from it Nearest multiple object plane deformation points are control point subset, perform serial RBF deformations, and solution obtains the deflection of spatial point.
Compared with prior art, the positive effect of the present invention is:
1st, versatility is good.The parallelization RBF that the present invention uses has stronger versatility, is suitable for data and inserts The application fields such as value, distortion of the mesh;
2nd, method is simple.Only need to change on a small quantity, it is possible to by the RBF method parallelizations of existing serialization.
Brief description of the drawings
Examples of the present invention will be described by way of reference to the accompanying drawings, wherein:
Fig. 1 is the flow chart of the inventive method.
Embodiment
A kind of space lattice deformation method based on parallelization RBF, as shown in figure 1, comprising the following steps:
Step 1, parallel architecture is established, the parallel architecture of establishing comprises the following steps:
Step 101, the partitioning algorithm (as called METIS) using routine, are n blockette by the mesh generation of computational fields Z={ Z1、Z2、Z3······Zn, the corresponding process P={ P of each block1、P2、P3······Pn};Here Subregion block number n determined by initial mesh scale and calculator memory, typically make the grid cell in each subregion after scoring area Number is less than 1,000,000.
Step 102, by each blockette ZiOn object plane unit send to corresponding server processes Pi
All object plane cell S={ s that step 103, server processes will be collected into1、s2、s3······slAssembling For whole object plane S;Here, l<=n.
Step 104, server processes will include all object plane cell S={ s1、s2、s3······slIt is whole Object plane is sent to each process PiSo that each process PiThere is the backup of a global object plane cell S;
Step 2, the institute for traveling through space a little, carry out parallel RBF deformations.For each process PiIn each spatial point Pfield,iDeformation:
In step 201, the object plane control point on whole object plane unit collection S, search from spatial point Pfield,iNearest is N number of Control point (N typically takes 10~30), form Pfield,iControl point subset { ci};
Step 202, to spatial point Pfield,i, the control point subset { c that is possessed based on itiAnd these control points deflection, According to formula (1), formula (2), solution obtains spatial point Pfield,iDeformation.
Table 1 be using the present invention space lattice deformation method based on parallelization RBF with it is traditional based on string The efficiency comparative of row RBF space lattice deformation method.It can be seen that either for simple profile, or for as transporting A kind of wing deformation of machine, calculating time of the invention all significantly shorten, and obtain good effect.
The efficiency comparative of the present invention of table 1 and conventional method
Serial RBF Parallel RBF
Cube rotates 78 minutes 4.2 minutes (8 core)
JSM transporters wing deforms 8.5 hour 17 minutes (2048 core)

Claims (6)

  1. A kind of 1. space lattice deformation method based on parallelization RBF, it is characterised in that:Comprise the following steps:
    Step 1: establishing parallel architecture, the grid of computational fields is subjected to subregion, the corresponding process of each blockette, and make every Individual process has the backup of a global object plane unit;
    Step 2: traveling through all spatial points in each process, parallel RBF deformations are carried out, wherein, each spatial point is with its nearest neighbours Multiple object plane deformation points be control point subset, perform serial RBF deformation, solution obtains the deflection of spatial point.
  2. 2. a kind of space lattice deformation method based on parallelization RBF according to claim 1, its feature exist In:1,000,000 are less than to the number of meshes in each subregion after grid progress subregion described in step 1.
  3. 3. a kind of space lattice deformation method based on parallelization RBF according to claim 1, its feature exist In:The object plane unit on corresponding blockette that global object plane unit described in step 1 is collected into by each server processes forms.
  4. 4. a kind of space lattice deformation method based on parallelization RBF according to claim 1, its feature exist In:The number at the control point subset midpoint is 10~30.
  5. 5. a kind of space lattice deformation method based on parallelization RBF according to claim 1, its feature exist In:The method of solution room point deformation amount is as follows:The displacement of spatial point is calculated as follows:
    Wherein:N is the number at control point subset midpoint, and f (r) is spatial point Pfield,iDisplacement, r is spatial point Pfield,i's Position, riFor i-th of object plane control point Psurface,iInitial position, | | r-ri| | represent Pfield,iAnd Psurface,iBetween Europe Formula distance,Represent selected RBF, wiRepresent i-th of object plane control point Psurface,iThe weight coefficient at place.
  6. 6. a kind of space lattice deformation method based on parallelization RBF according to claim 5, its feature exist In:The weight coefficient wiMeet following relational expression:
    Wherein, △ xiFor i-th of object plane control point Psurface,iDeflection,
CN201710655495.5A 2017-08-03 2017-08-03 A kind of space lattice deformation method based on parallelization RBF Pending CN107391871A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710655495.5A CN107391871A (en) 2017-08-03 2017-08-03 A kind of space lattice deformation method based on parallelization RBF

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710655495.5A CN107391871A (en) 2017-08-03 2017-08-03 A kind of space lattice deformation method based on parallelization RBF

Publications (1)

Publication Number Publication Date
CN107391871A true CN107391871A (en) 2017-11-24

Family

ID=60343547

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710655495.5A Pending CN107391871A (en) 2017-08-03 2017-08-03 A kind of space lattice deformation method based on parallelization RBF

Country Status (1)

Country Link
CN (1) CN107391871A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112989497A (en) * 2021-04-21 2021-06-18 中国空气动力研究与发展中心计算空气动力研究所 Tight branch radial basis function data transfer method based on geometric space main feature extraction
CN113255230A (en) * 2021-06-16 2021-08-13 中国地质科学院 Gravity model forward modeling method and system based on MQ radial basis function

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150112910A1 (en) * 2013-10-22 2015-04-23 In2H2 Hardware enhancements to radial basis function with restricted coulomb energy learning and/or k-nearest neighbor based neural network classifiers
CN104765589A (en) * 2014-01-02 2015-07-08 广州中国科学院软件应用技术研究所 Grid parallel preprocessing method based on MPI
CN106067064A (en) * 2016-06-03 2016-11-02 福州大学 Function approximator system based on RBF

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150112910A1 (en) * 2013-10-22 2015-04-23 In2H2 Hardware enhancements to radial basis function with restricted coulomb energy learning and/or k-nearest neighbor based neural network classifiers
CN104765589A (en) * 2014-01-02 2015-07-08 广州中国科学院软件应用技术研究所 Grid parallel preprocessing method based on MPI
CN106067064A (en) * 2016-06-03 2016-11-02 福州大学 Function approximator system based on RBF

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
PIETER COULIER 等: "Efficient mesh deformation based on radial basis function interpolation by means of the inverse fast multipole method", 《COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING》 *
孙岩 等: "基于径向基函数与混合背景网格的动态网格变形方法", 《航空学报》 *
马文鹏 等: "径向基函数在动网格中的应用及可并行性研究", 《科研信息化技术与应用》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112989497A (en) * 2021-04-21 2021-06-18 中国空气动力研究与发展中心计算空气动力研究所 Tight branch radial basis function data transfer method based on geometric space main feature extraction
CN113255230A (en) * 2021-06-16 2021-08-13 中国地质科学院 Gravity model forward modeling method and system based on MQ radial basis function
CN113255230B (en) * 2021-06-16 2024-02-20 中国地质科学院 Gravity model forward modeling method and system based on MQ radial basis function

Similar Documents

Publication Publication Date Title
CN108446445B (en) Composite material wing optimization design method based on aerodynamic reduced order model
CN105843073B (en) A kind of wing structure aeroelastic stability analysis method not knowing depression of order based on aerodynamic force
CN110705029B (en) Flow field prediction method of oscillating flapping wing energy acquisition system based on transfer learning
CN110110413A (en) A kind of structural topological optimization method based on yard of material reduction series expansion
CN106777482A (en) A kind of structure Multidisciplinary design optimization method based on mesh parameterization
WO2019056887A1 (en) Method for performing probabilistic modeling of large-scale renewable-energy data
CN108241777B (en) Method for calculating seepage velocity field in hydrate sediment based on unstructured grid finite element method
CN108228970B (en) Structural dynamics analysis explicit different step length parallel computing method
CN103326351B (en) Distribution type parallel load flow calculation method based on hierarchical clustering automatic partition
CN108108559B (en) Structure response obtaining method and sensitivity obtaining method based on substructure
CN112597610B (en) Optimization method, device and equipment for lightweight design of mechanical arm structure
CN111028238A (en) Robot vision-based three-dimensional segmentation method and system for complex special-shaped curved surface
CN113159287B (en) Distributed deep learning method based on gradient sparsity
Bourgault-Côté et al. Multi-layer icing methodologies for conservative ice growth
CN107391871A (en) A kind of space lattice deformation method based on parallelization RBF
Deng et al. Prediction of transonic flow over supercritical airfoils using geometric-encoding and deep-learning strategies
Chiba et al. High-fidelity multidisciplinary design optimization of aerostructural wing shape for regional jet
Chang et al. A parallel implicit hole-cutting method based on background mesh for unstructured Chimera grid
CN107391950A (en) A kind of turbulent flow wall is apart from quick calculation method
CN112580239A (en) Substructure response reconstruction method and system, and storage medium
Durasov et al. Debosh: Deep bayesian shape optimization
CN103065015B (en) A kind of bearing structure low-carbon (LC) material-saving method for designing based on internal force path geometry form
CN102496033A (en) Image SIFT feature matching method based on MR computation framework
CN105718619A (en) Method for determining fuel quality characteristics of aircraft based on finite element method
CN110175372B (en) Envelope surface characterization method based on mother surface characteristic parameters

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