CN107609273B - Engineering product design method - Google Patents
Engineering product design method Download PDFInfo
- Publication number
- CN107609273B CN107609273B CN201710823081.9A CN201710823081A CN107609273B CN 107609273 B CN107609273 B CN 107609273B CN 201710823081 A CN201710823081 A CN 201710823081A CN 107609273 B CN107609273 B CN 107609273B
- Authority
- CN
- China
- Prior art keywords
- design
- target
- value
- values
- scheme
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000013461 design Methods 0.000 title claims abstract description 214
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000005457 optimization Methods 0.000 claims abstract description 35
- 239000002245 particle Substances 0.000 claims abstract description 35
- 230000014509 gene expression Effects 0.000 claims abstract description 28
- 238000012937 correction Methods 0.000 claims abstract description 26
- 238000004364 calculation method Methods 0.000 claims abstract description 13
- 238000009499 grossing Methods 0.000 claims abstract description 13
- 238000010521 absorption reaction Methods 0.000 claims description 20
- 230000008859 change Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 5
- 238000012938 design process Methods 0.000 description 3
- 230000002452 interceptive effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a design method of an engineering product, which comprises the following steps: reading design variables, design constraints and design targets; expressing the design target into a single target expression by adopting a smoothing calculation formula; taking the maximum value of the smooth single-target expression as a target, and carrying out heuristic search by adopting a particle swarm optimization algorithm; correcting all target values which do not meet the design target in the scheme; outputting the serial numbers of all the design constraints which do not meet the scheme and the optimal values found by the corresponding design constraints; or outputting values of design variables corresponding to all design schemes capable of satisfying the design. The invention adopts the particle swarm optimization algorithm to perform heuristic search, automatically correct design constraints and interactively select a design scheme, is a creative design method, can process the design problems of modern engineering products containing large-scale design variables, design constraints and design targets, and enables designers to efficiently obtain new products meeting the subjective consciousness of the designers in limited time because the optimization and the automatic design constraint correction are not involved.
Description
Technical Field
The invention relates to the technical field of product design, in particular to a design method of an engineering product.
Background
At present, the design of engineering products usually aims at searching an optimal design scheme, and after designers formulate design variables, design constraints and design targets according to experience or existing similar products, a computer adopts various optimization algorithms to search the optimal design targets under the design constraints by changing the design variables.
However, with the increasing complexity of engineering product design targets and the rapid development of information technology, the data volume of design variables, design constraints and design targets has increased explosively, and such design method based on optimization idea will face the challenges of computational efficiency and optimization capability. There are mainly the following problems:
1) for the problem of high-dimensional nonlinearity that the number of design variables, design constraints and design targets in the design process of modern practical engineering products is huge, an optimization algorithm capable of quickly obtaining an optimal value is rarely seen at present; in fact, for the high-dimensional nonlinear problem in practical engineering, the existing optimization algorithm can not obtain the optimal solution generally;
2) for the optimization problem which can meet the solution does not exist, the automatic correction of the design constraint can not be carried out in the optimization design process, and a designer can only modify the design target by depending on experience after the optimization fails;
3) the existing optimization design method completely isolates designers from design constraints, is a design process lacking interaction, and the designers can only passively accept an optimal solution, but are difficult to integrate subjective consciousness into engineering products.
Disclosure of Invention
In view of the above, the present invention is directed to a design method of engineering products, so as to solve the problem of completely isolating designers from design targets.
Based on the above purpose, the present invention provides a design method of engineering products, comprising:
step 101, reading design variables, design constraints and design targets;
102, expressing a design target into a single target expression by adopting a smoothing calculation formula;
103, searching a heuristic search by adopting a particle swarm optimization algorithm by taking the maximum value of the smooth single-target expression as a target;
step 104, correcting all target values which do not meet the design target in the scheme;
step 105, outputting all numbers which do not meet the design constraint of the scheme and the optimal values found by the corresponding design constraint; or,
and 106, outputting values of design variables corresponding to all design schemes capable of meeting the satisfaction.
In some embodiments of the present invention, if there are multiple design targets, the multiple design targets are expressed as a single target expression by using a smoothing calculation formula.
In some embodiments of the invention, the smoothing calculation formula is as follows:
in the formula, beta is a smooth factor, the value is more than 0, the smoothness is reduced along with the increase of the beta value, and the fitting degree of the fitting curved surface and the actual curved surface is higher;
x1,x2,...,xnis a design variable;
y1,y2,...,yka current value of a design objective;
g1,g2,...,gmthe method is specifically as follows:
in the formula, V is a target value of a design target.
In some embodiments of the present invention, the performing a heuristic search using a particle swarm optimization algorithm includes:
step 1031, finding the current optimal value of the single target expression through a particle swarm optimization algorithm;
step 1032, judging whether the design scheme corresponding to the current optimal value is a satisfiable scheme;
step 1033, recording all current design constraints and design targets not meeting the design, and finding the best target value Vbest；
Step 1034, update the iteration times by the following formula:
N＝N+1
if N is smaller than the set maximum iteration number, finding the current optimal value of the single target expression through a particle swarm optimization algorithm; if N reaches the maximum number of iterations, then step 1035 is performed;
step 1035, number n of scenarios if it can be satisfiedokIf equal to 0, go to step 104; if the number of schemes n can be satisfiedokIf greater than 0, go to step 106.
In some embodiments of the invention, the maximum number of iterations is 2000.
In some embodiments of the present invention, the finding the current optimal value of the single target expression by the particle swarm optimization algorithm includes:
1) each particle updates the value of the corresponding design variable and calculates the current value of the design target according to the value of the design variable;
2) each particle calculates the value of a single target expression according to the current value of the design target corresponding to the particle;
3) and finding the maximum value in the values of the single target expressions corresponding to all the particles as the current optimal value.
In some embodiments of the present invention, the correcting all target values that do not meet the design goal in the solution includes:
step 1041, if the number of times of correction NmfIf the number of times of correction is greater than the set maximum number of times of correction, go to step 105; if the number of corrections NmfIf the number of times of correction is less than the set maximum number of times, go to step 1042;
step 1042, the correction formula is as follows:
in the formula, VnewIs the corrected target value;
step 1043, update correction times Nmf
Nmf＝Nmf+1
Step 1044, change V to VbestStep 102 is performed.
The design method of the engineering product provided by the embodiment of the invention gets rid of the design idea of taking the optimum as the target, after designers customize the design variables, the design constraints and the design targets, the smooth calculation formula is adopted to express a plurality of design targets into a simple formula with high robustness, meanwhile, the optimum value of the design target is not searched, only the particle swarm optimization algorithm is adopted to carry out heuristic search, and a series of design variable values and design target values which can meet the design constraints are searched in a certain iteration step to serve as a design scheme library. Meanwhile, the satisfied conditions of each design constraint are recorded in the searching process, and if a feasible design scheme cannot be found in a certain iteration step, the design constraint is corrected step by step according to the real-time satisfied conditions of the design constraint until the design scheme can be generated. And finally, the engineering personnel select a design scheme which accords with the subjective consciousness of the engineering personnel through an interactive image interface.
Therefore, the design method of the engineering product provided by the invention adopts the particle swarm optimization algorithm to perform heuristic search, automatically correct design constraints and select the design scheme with interactivity, is a creative design method, can process the design problems of modern engineering products containing large-scale design variables, design constraints and design targets, and enables designers to efficiently obtain new products meeting the subjective consciousness of the designers in a limited time because the optimization is not involved and the automatic design constraint correction is contained.
Drawings
FIG. 1 is a flow chart of a method for designing an engineered product according to an embodiment of the invention;
FIG. 2 is a schematic structural view of an energy absorption box according to an embodiment of the present invention;
FIG. 3 is a schematic structural view of a crash box designed with a set of possible solutions according to an embodiment of the present invention;
FIG. 4 is a schematic structural view of an energy absorption box according to another possible design set of embodiments of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to specific embodiments and the accompanying drawings.
Fig. 1 is a flow chart of a design method of an engineering product according to an embodiment of the present invention. As an embodiment of the present invention, the design method includes the steps of:
step 101, reading design variables, design constraints and design goals.
In this step, design variables, design constraints, and design targets specified by the designer are read. Wherein the design constraint is a value range of the design variable. The design target may be plural.
As an embodiment of the present invention, taking the crash box as an example, the design variables may be a thickness x2-x25 of a honeycomb structure inside the crash box and a thickness x1 of an outer wall of the crash box, the design constraint may be a value range of the thicknesses, and the design target may be at least one of energy absorption performance, weight, maximum impact force and/or specific energy absorption.
And 102, expressing the design target into a single target expression by adopting a smoothing calculation formula.
If the design targets are multiple, in the step, expressing the multiple design targets into a single target expression by adopting a smoothing calculation formula.
As another embodiment of the present invention, to improve the efficiency of searching for the satisfactory solution, the smoothing calculation formula is as follows:
in the formula, beta is a smooth factor, the value is more than 0, the smoothness is reduced along with the increase of the beta value, and the fitting degree of the fitting curved surface and the actual curved surface is higher;
x1,x2,...,xnis a design variable;
y1,y2,...,ykthe current value of the design objective.
g1,g2,...,gmThe method is specifically as follows:
in the formula, V is a target value of a design target.
The target value V is a target value of a design target.
And 103, searching a heuristic search by adopting a particle swarm optimization algorithm by taking the maximum value of the smoothed single-target expression as a target.
Specifically, the step of performing heuristic search by using a particle swarm optimization algorithm includes:
and step 1031, finding the current optimal value of the single target expression through a particle swarm optimization algorithm. In this step, the current best value of the single target expression in step 102 is found by the particle swarm optimization algorithm.
As another embodiment of the present invention, the finding the current optimal value of the single target expression by the particle swarm optimization algorithm includes:
1) each particle updates the value of the corresponding design variable and calculates the current value of the design target according to the value of the design variable;
2) each particle calculates the value of a single target expression according to the current value of the design target corresponding to the particle;
3) and finding the maximum value in the values of the single target expressions corresponding to all the particles as the current optimal value.
Step 1032, determine whether the design solution corresponding to the current optimal value is a satisfiable solution.
In particular, if pminIf the number of the design targets is more than or equal to 0, the current scheme is a satisfiable scheme, the values of all current design targets, design constraints and design targets are recorded, and the number n of satisfiable schemes is recordedokThen go to step 1034; if p isminIf < 0, the current solution is not the satisfactory solution, and step 1033 is performed.
It should be noted that the satisfiable solution refers to a solution in which the design constraints and the design targets meet the requirements, because all the design constraints are satisfied, pminCan be greater than or equal to zero.
Step 1033, recording all the design constraints and design targets in the unsatisfied solution, and finding the best target value Vbest。
Specifically, step 1033 includes:if the current value of the design target is less than the target value, i.e., y ≦ V, then VbestFor the minimum value of the currently found design target, if the design target is that y is more than or equal to V, then VbestThe maximum value of the currently found design objective.
Step 1034, update the iteration number by the following formula:
N＝N+1
if N is less than the set maximum iteration number, executing step 1031; if N reaches the maximum number of iterations, step 1035 is performed.
Alternatively, the maximum number of iterations may be 2000.
In the iterative process, the values of the design constraints are continuously updated by adopting a particle swarm optimization algorithm, and the values of the design targets are calculated according to the design constraints.
Step 1035, number n of scenarios if it can be satisfiedokIf equal to 0, go to step 104; if the number of schemes n can be satisfiedokIf greater than 0, go to step 106.
It should be noted that if a satisfiable solution is found in the previous step, then nokIs not zero, if no satisfiable solution is found, nokEqual to zero.
Step 104, all target values that do not meet the design objective in the solution are corrected.
In this step, the target values of the design constraints in the unsatisfied project may be corrected according to the number of all the design constraints of the unsatisfied project recorded in step 1032.
Specifically, the step 104 includes:
step 1041, if the number of times of correction NmfIf the number of times of correction is greater than the set maximum number of times of correction, go to step 105; if the number of corrections NmfIf the number of corrections is less than the set maximum number of corrections, go to step 1042.
Step 1042, the correction formula is as follows:
in the formula, VnewIs the corrected target value.
Step 1043, update correction times Nmf
Nmf＝Nmf+1
Step 1044, change V to VbestStep 102 is performed.
And 105, prompting design failure, and outputting all numbers which do not meet the design constraints of the scheme and the optimal values found by the corresponding design constraints for reference of designers.
And 106, outputting design parameters corresponding to all design schemes capable of meeting the satisfaction for the designer to refer. The design parameters include values of various design variables.
As shown in FIG. 2, the energy absorption box of an electric automobile is designed, the energy absorption box is a square pipe, the length of the pipe is 0.2 m, the inside of the pipe is of a honeycomb structure, and the requirements of energy absorption, maximum impact force and specific energy absorption are met. The design method of the energy absorption box comprises the following steps:
(1) read design variables, design constraints, and design goals.
The design variables are the thickness x2-x25 of the honeycomb structure in the energy absorption box and the thickness x1 of the outer wall of the energy absorption box; the design constraint is that the value range of all the thicknesses (x1-x25) can be 0.1-0.2; the design targets are energy absorption (y1) > 45000, maximum impact force (y2) < 1950000, and specific energy absorption (y3) < 5000000.
(2) And expressing the design target into a single target expression by adopting a smoothing calculation formula. The smoothing formula used is as follows:
wherein the smoothing factor β is 1.5.
(3) And searching the maximum value of the smooth single-target expression as a target, and performing heuristic search by adopting a particle swarm optimization algorithm. The maximum iteration number is set to 2000 times, and PSO iterative search is executed.
(4) After the PSO iterates the search result, n is foundokEqual to 0. Therefore, the limit range of the absorbed energy is modified according to the number recorded in step (3) that does not satisfy the design constraint.
(5) According to the step (4), the value range of specific energy absorption (y3) in the design target is modified to be less than 5600000, and the limit value ranges of the energy absorption (y1) and the maximum collision force (y2) are unchanged.
(6) The maximum iteration number is set to 2000 times, and PSO iterative search is executed.
(7) After the PSO iterates the search result, n is foundokEqual to 2, two sets of feasible solutions were found:
the first set of solutions:
the values of each design variable are:
x1＝0.1094，x2＝0.1000，x3＝0.1006，x4＝0.1992，x5＝0.1875，x6＝0.1992，x7＝0.1999，x8＝0.2000，x9＝0.1999，x10＝0.1406，x11＝0.2000，x12＝0.1999，x13＝0.2000，x14＝0.1999
x15＝0.2000，x16＝0.1992，x17＝0.2000，x18＝0.1992，x19＝0.1999，x20＝0.1909，
x21＝0.1875，x22＝0.1999，x23＝0.1919，x24＝0.2000，x25＝0.1875。
the values of each design objective are:
y1＝149630.015900，y2＝1671448.844800，y3＝5596077.575500。
the graphical display is shown in figure 3.
The second set of solutions:
the values of each design variable are:
x1＝0.1094，x2＝0.1000，x3＝0.1001，x4＝0.1992，x5＝0.1875，x6＝0.1992，x7＝0.2000，
x8＝0.2000，x9＝0.2000，x10＝0.1380，x11＝0.2000，x12＝0.2000，x13＝0.1997，x14＝0.2000，
x15＝0.2000，x16＝0.1999，x17＝0.2000，x18＝0.1774，x19＝0.2000，x20＝0.2000，
x21＝0.1875，x22＝0.2000，x23＝0.1996，x24＝0.2000，x25＝0.1953。
the values of each design objective are:
y1＝145723.081900，y2＝1619194.594600，y3＝5498431.161400。
the graphical display is shown in fig. 4.
(8) And (4) handing the two groups of feasible solutions and the corresponding graphs to a designer, and selecting a final solution by the designer.
The design method of the engineering product provided by the embodiment of the invention gets rid of the design idea of taking the optimum as the target, after designers customize the design variables, the design constraints and the design targets, the smooth calculation formula is adopted to express a plurality of design targets into a simple formula with high robustness, meanwhile, the optimum value of the design target is not searched, only the particle swarm optimization algorithm is adopted to carry out heuristic search, and a series of design variable values and design target values which can meet the design constraints are searched in a certain iteration step to serve as a design scheme library. Meanwhile, the satisfied conditions of each design constraint are recorded in the searching process, and if a feasible design scheme cannot be found in a certain iteration step, the design constraint is corrected step by step according to the real-time satisfied conditions of the design constraint until the design scheme can be generated. And finally, the engineering personnel select a design scheme which accords with the subjective consciousness of the engineering personnel through an interactive image interface.
Therefore, the design method of the engineering product provided by the invention adopts the particle swarm optimization algorithm to perform heuristic search, automatically correct design constraints and select the design scheme with interactivity, is a creative design method, can process the design problems of modern engineering products containing large-scale design variables, design constraints and design targets, and enables designers to efficiently obtain new products meeting the subjective consciousness of the designers in a limited time because the optimization is not involved and the automatic design constraint correction is contained.
Those of ordinary skill in the art will understand that: the discussion of any embodiment above is meant to be exemplary only, and is not intended to intimate that the scope of the disclosure, including the claims, is limited to these examples; within the idea of the invention, also technical features in the above embodiments or in different embodiments may be combined and there are many other variations of the different aspects of the invention as described above, which are not provided in detail for the sake of brevity. Therefore, any omissions, modifications, substitutions, improvements and the like that may be made without departing from the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (5)
1. A method for designing a crash box, comprising:
step 101, reading design variables, design constraints and design targets;
102, expressing a design target into a single target expression by adopting a smoothing calculation formula;
103, searching a heuristic search by adopting a particle swarm optimization algorithm by taking the maximum value of the smooth single-target expression as a target;
step 104, correcting all target values which do not meet the design target in the scheme;
step 105, outputting all numbers which do not meet the design constraint of the scheme and the optimal values found by the corresponding design constraint; or,
step 106, outputting values of design variables corresponding to all design schemes capable of meeting the satisfaction;
the design variables are the thickness of a honeycomb structure inside the energy absorption box and the thickness of the outer wall of the energy absorption box, the design constraint is the value range of the thickness, and the design target is at least one of energy absorption performance, weight, maximum impact force and/or specific energy absorption;
the smoothing calculation formula is as follows:
in the formula, beta is a smooth factor, the value is more than 0, the smoothness is reduced along with the increase of the beta value, and the fitting degree of the fitting curved surface and the actual curved surface is higher;
x1,x2,...,xnis a design variable;
y1,y2,...,yka current value of a design objective;
g1,g2,...,gmthe method is specifically as follows:
wherein V is a target value of a design target;
the heuristic search by adopting the particle swarm optimization algorithm comprises the following steps:
step 1031, finding the current optimal value of the single target expression through a particle swarm optimization algorithm;
step 1032, judging whether the design scheme corresponding to the current optimal value is a satisfiable scheme;
step 1033, recording all current design constraints and design targets not meeting the design, and finding the best target value Vbest；
Step 1034, update the iteration times by the following formula:
N＝N+1
if N is smaller than the set maximum iteration number, finding the current optimal value of the single target expression through a particle swarm optimization algorithm; if N reaches the maximum number of iterations, then step 1035 is performed;
step 1035, number n of scenarios if it can be satisfiedokIf equal to 0, go to step 104; if the number of schemes n can be satisfiedokIf greater than 0, go to step 106.
2. The method of claim 1, wherein if there are multiple design targets, the smoothing calculation formula is used to express the multiple design targets as a single target expression.
3. The method of claim 1, wherein the maximum number of iterations is 2000.
4. The method for designing an energy absorption box according to claim 1, wherein the finding the current optimal value of the single target expression by a particle swarm optimization algorithm comprises:
1) each particle updates the value of the corresponding design variable and calculates the current value of the design target according to the value of the design variable;
2) each particle calculates the value of a single target expression according to the current value of the design target corresponding to the particle;
3) and finding the maximum value in the values of the single target expressions corresponding to all the particles as the current optimal value.
5. The method of designing a crash box according to claim 1, wherein said modifying all target values that do not meet the design objective in the solution comprises:
step 1041, if the number of times of correction NmfIf the number of times of correction is greater than the set maximum number of times of correction, go to step 105; if the number of corrections NmfIf the number of times of correction is less than the set maximum number of times, go to step 1042;
step 1042, the correction formula is as follows:
in the formula, VnewIs the corrected target value;
step 1043, update correction times Nmf
Nmf＝Nmf+1
Step 1044, change V to VbestStep 102 is performed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710823081.9A CN107609273B (en) | 2017-09-13 | 2017-09-13 | Engineering product design method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710823081.9A CN107609273B (en) | 2017-09-13 | 2017-09-13 | Engineering product design method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107609273A CN107609273A (en) | 2018-01-19 |
| CN107609273B true CN107609273B (en) | 2021-01-15 |
Family
ID=61063570
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710823081.9A Active CN107609273B (en) | 2017-09-13 | 2017-09-13 | Engineering product design method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107609273B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105631093A (en) * | 2015-12-18 | 2016-06-01 | 吉林大学 | M-BSWA multi-target optimization based mechanical structure design method |
| CN106934175A (en) * | 2017-03-29 | 2017-07-07 | 南京航空航天大学 | A kind of negative poisson's ratio structure energy-absorption box and its Multipurpose Optimal Method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7756688B2 (en) * | 2004-05-10 | 2010-07-13 | Board Of Trustees Of Michigan State University | Design optimization system and method |
-
2017
- 2017-09-13 CN CN201710823081.9A patent/CN107609273B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105631093A (en) * | 2015-12-18 | 2016-06-01 | 吉林大学 | M-BSWA multi-target optimization based mechanical structure design method |
| CN106934175A (en) * | 2017-03-29 | 2017-07-07 | 南京航空航天大学 | A kind of negative poisson's ratio structure energy-absorption box and its Multipurpose Optimal Method |
Non-Patent Citations (1)
| Title |
|---|
| 参数化诱导槽设计的吸能盒结构抗撞性多目标优化;郝亮 等;《吉林大学学报(工学版)》;20130131;第43卷(第1期);第1-6页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107609273A (en) | 2018-01-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Improved non-dominated sorting genetic algorithm (NSGA)-II in multi-objective optimization studies of wind turbine blades | |
| Oh et al. | Design automation by integrating generative adversarial networks and topology optimization | |
| Wen et al. | Structural deep incomplete multi-view clustering network | |
| CN108491576B (en) | Optimization design method for reinforcing composite material wing opening | |
| CN105760619A (en) | Composite material layer optimization design method taking fatigue reliability into account | |
| Liu et al. | Proximal alternating direction network: A globally converged deep unrolling framework | |
| CN107609273B (en) | Engineering product design method | |
| Fu et al. | Exploring structural sparsity of deep networks via inverse scale spaces | |
| CN114692488A (en) | Method and device for generating supercritical airfoil, electronic equipment and storage medium | |
| CN106776579B (en) | The sampling accelerated method of Biterm topic model | |
| CN110688860B (en) | Weight distribution method based on multiple attention mechanisms of transformer | |
| CN105427364A (en) | Multi-point touch two-dimensional animation production method | |
| CN116739923B (en) | Face area guided image blind restoration method based on meta learning framework | |
| Berry et al. | A generalized plasma dispersion function for electron damping in tokamak plasmas | |
| Shirvani et al. | A deep learning‒genetic algorithm approach for aerodynamic inverse design via optimization of pressure distribution | |
| CN102289430B (en) | Method for analyzing latent semantics of fusion probability of multi-modality data | |
| Gan et al. | Generative Adversarial Networks with Learnable Auxiliary Module for Image Synthesis | |
| WO2022248867A1 (en) | Methods and apparatus for optical fibre design and production | |
| Wang et al. | Automatic Network Architecture Search for RGB-D Semantic Segmentation | |
| Li et al. | Interactive Editing of Discrete Chebyshev Nets | |
| Heo et al. | Neural-network model for linear mhd stability analysis of tokamak edge pedestals | |
| Xie et al. | Design Space Exploration of CNN Accelerators based on GSA Algorithm | |
| CN112232482A (en) | Method and equipment for constructing neural network architecture search and intelligent equipment | |
| CN117709164B (en) | Self-supporting structure topology optimization design method based on phase field method | |
| Long et al. | An Overview of Sequential Approximation in Topology Optimization of Continuum Structure. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |