CN105426965B - A kind of sort method applied to multiple target section genetic algorithm - Google Patents
A kind of sort method applied to multiple target section genetic algorithm Download PDFInfo
- Publication number
- CN105426965B CN105426965B CN201510902573.8A CN201510902573A CN105426965B CN 105426965 B CN105426965 B CN 105426965B CN 201510902573 A CN201510902573 A CN 201510902573A CN 105426965 B CN105426965 B CN 105426965B
- Authority
- CN
- China
- Prior art keywords
- design
- individual
- population
- sub
- hypervolume
- 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.)
- Expired - Fee Related
Links
Abstract
The invention discloses a kind of sort method applied to multiple target section genetic algorithm, design variable and design object are determined according to design object；Search meets the individual of design requirement, and the fitness of new individual, i.e. design object value are calculated using Fluid Mechanics Computation method；All individuals are ranked up according to design object value, while using hypervolume size and non-dominated ranking, generate sub- population one, sub- population two；In sub- population one, all individual hypervolumes are calculated, according to hypervolume size, are ranked up from big to small, it is contemplated that the section characteristic of solution；In sub- population two, using non-dominated ranking method, all individuals are ranked up, solve multi-objective problem；By sub- population one and sub- two ordering by merging of population；New population is produced using selection operation：Until new population meets convergence, design process terminates.The present invention solves the problems, such as the multi-objective problem of genetic algorithm and interval solutions at the same time.
Description
Technical field
The present invention relates to block design technical field, more particularly to a kind of row applied to multiple target section genetic algorithm
Sequence method.
Background technology
Genetic algorithm is a kind of global optimization approach, it is expressed as the possible solution of each variable the piece of one chromosome
Section, all possible solutions of variable are arranged in a complete chromosome in order, and every chromosome represents one in evolutionary process
Individual, all individuals in the same generation are known as a population.Before genetic algorithm is performed, initial population is provided first,
Namely some solutions assumed.Then, these hypothesis solutions are placed in the environment of problem, and according to the principle of the survival of the fittest, led to
Cross the processes such as intersection, variation, selection and produce the population of new generation for more adapting to environment.So evolve, finally will from generation to generation
Convergence obtains the optimal solution of problem.
During using genetic algorithm, often it is related to multi-objective problem and interval solutions problem.Multi-objective problem refers to optimize
During there are multiple optimization aims, need to consider influence of each target to optimum results at the same time.Interval solutions problem refers to hereditary calculation
In method calculating process, each individual represents a range of variables, and optimum results need to use the form in section to represent.
For multi-objective problem, the method for non-dominated ranking can be used to be ranked up all individuals, and then select most
Excellent individual.Here by taking maximization problems as an example, non-dominant relation is stated, for two any individuals A, B：
(1) for arbitrary target function f, it is satisfied by f (A)>F (B), claims A to be dominantOr A dominates B；
(2) if A is not dominant in B, and B is not dominant when A, claims A and B indifferences.
Non-dominant individual refers to not by any one other individual for being dominated of individual, they the characteristics of be if to wherein
One target is improved, and will damage other at least one targets.All non-dominant individuals form or Pareto forward position, in pa
Individual on tired support forward position dominates from any other individual.
For interval solutions problem, the section performance of hypervolume evaluation individual can be used, and then according to hypervolume size to institute
There is individual to be ranked up.Hypervolume calculation formula is as follows：
λ is Lebesgue measure, and y is Pareto forward position, yrefFor reference point.
When being designed using genetic algorithm to indoor environment, often there are multiple design objects, and desired set
It is often an interval range to count result, therefore need to solve the problems, such as multi-objective problem and interval solutions at the same time.Using only non-dominant row
Sequence method is ranked up individual, it is impossible to embodies the section characteristic of individual；Individual is ranked up according only to hypervolume size, is needed
All individual hypervolumes are calculated, it is computationally intensive.Therefore, existing method cannot solve the problems, such as this very well.
The content of the invention
For the above-mentioned prior art and there are the problem of, the present invention, which proposes, a kind of to be applied to the heredity of multiple target section and calculates
The sort method of method, when being designed using genetic algorithm to indoor environment, is realized and is solved at the same time in an interval range
The problem of multi-objective problem and interval solutions.
The present invention proposes a kind of sort method applied to multiple target section genetic algorithm, comprises the following steps：.
Step 1, according to design object determine design variable and design object；The design variable of Cabin model is entrance velocity
Vin, inlet temperature Tin, inlet angle Ain, entry position LinAnd outlet port Lout；The design object of Cabin model is (1) |
PMVc|<0.5；(2)0.1m/s<Vhead<0.2m/s；(3)Vfeat<0.2m/s；(4)ΔT<2.8℃.Vhead、VfeatRepresented with Δ T
Head speed, foot's speed and the vertical temperature difference, PMVcIt is as follows for modified PMV, its calculation formula：
PMVc=-0.0758PMV2+0.6757PMV-0.1262；
Step 2, using the cross and variation process of genetic algorithm obtain multigroup design variable value for meeting design requirement, calculates
The design object value of new individual, i.e. the boundary condition of design object is determined according to design variable value, is calculated in design object
The distribution of the speed, temperature in portion, genetic algorithm crossover probability are 0.8, mutation probability 0.1, are often 24 for number of individuals, obtain every
The corresponding design object value of individual, the condition of convergence is to obtain 8 interval solutions for meeting design requirement；In design process, at the same time
Individual is ranked up using non-dominated ranking and hypervolume.It is true according to the design object value of individual during using non-dominated ranking
Fixed all individual sequences；When calculating hypervolume, it is first determined the subinterval of different designs variable, i.e. variable change scope, so
The method for using latin hypercube sampling afterwards, a certain amount of individual is extracted in subinterval, is calculated, is designed accordingly
Desired value, finally selects the individual for meeting design requirement as local solution, calculates these hypervolumes being locally deconstructed into；And then
To design object value；
Step 3, according to design object value, while using hypervolume size and non-dominated ranking arrange all individuals
Sequence：If the design object value of individual meets design standard, sub- population one is classified to；Set if the design object value of individual is not met
Meter standard, is classified to sub- population two；In sub- population one, all individual hypervolumes are calculated, according to hypervolume size, from big
It is ranked up to small, it is contemplated that the section characteristic of solution, using the section characteristic of individual hypervolume size evaluation solution, hypervolume is got over
Greatly, the section where illustrating the individual is better；In sub- population two, using non-dominated ranking method, all individuals are arranged
Sequence, solves multi-objective problem；
Step 4, come sub- population one and sub- two ordering by merging of population, sub- population one before sub- population two；
Step 5, by non-dominated ranking method be ranked up the individual in population, and is selected using tournament algorithm
Select, new population is produced using selection operation：If new population meets convergence, design process terminates；Otherwise continue to intersect
Mutation process, produces new population.
The calculating of the hypervolume comprises the following steps：
The subinterval of different designs variable, i.e. variable change scope are determined, then using the side of latin hypercube sampling
Method, a certain amount of individual is extracted in subinterval, is calculated with Fluid Mechanics Computation, is obtained corresponding design object value, is selected
Meet that the individual of design requirement as local solution, calculates these hypervolumes being locally deconstructed into.
Compared with prior art, the present invention solves the problems, such as the multi-objective problem of genetic algorithm and interval solutions at the same time.
Brief description of the drawings
One kind that Fig. 1 is the present invention is applied to multiple target section genetic algorithm procedural model schematic diagram；
Fig. 2 is Cabin model schematic diagram；
Fig. 3 is solution quantity and the relation of genetic algebra；
Fig. 4 designs calculation amount and the relation of genetic algebra for Cabin model.
Embodiment
Below in conjunction with the drawings and the specific embodiments, technical scheme is described in further detail.
The design variable of Cabin model is entrance velocity Vin, inlet temperature Tin, inlet angle Ain, entry position LinAnd
Outlet port Lout.The size of entrance and exit is identical, and 14 optional entrances are provided with the upside of window, and downside is set
16 optional entrances.
Table 1, Cabin model design variable parameter
Table 1 is the excursion and change interval of each design variable.Due to entrance velocity, inlet temperature and inlet angle
It is continuous variable, entry position and outlet port are discrete variables, and interval solutions is just for continuous variable, therefore calculate hypervolume
When only consider three entrance velocity, inlet temperature and inlet angle variables.
The design object of Cabin model is as follows：
(1)|PMVc|<0.5；(2)0.1m/s<Vhead<0.2m/s；(3)Vfeat<0.2m/s；(4)ΔT<2.8℃.Vhead、
VfeatHead speed, foot's speed and the vertical temperature difference are represented with Δ T.PMVcFor modified PMV (predicted mean vote)
Formula, its calculation formula are as follows：
PMVc=-0.0758PMV2+0.6757PMV-0.1262 (2)
Genetic algorithm crossover probability is 0.8, mutation probability 0.1, is often 24 for number of individuals, uses Fluid Mechanics Computation skill
Art (CFD) obtains the corresponding design object value of each individual, and the condition of convergence is to obtain 8 interval solutions for meeting design requirement.If
During meter, while individual is ranked up using non-dominated ranking and hypervolume.During using non-dominated ranking, according to individual
Design object value determines all individual sequences.When calculating hypervolume, it is first determined the subinterval of different designs variable, i.e. variable
Excursion, then using the method for latin hypercube sampling, extracts a certain amount of individual in subinterval, carries out CFD calculating,
Corresponding design object value is obtained, the individual for meeting design requirement is finally selected as local solution, calculates these parts and be deconstructed into
Hypervolume.
Fig. 3 is change curve of the solution quantity with genetic algebra, and when genetic algorithm was calculated to 13 generation, design process terminates, 8
Solution results from the 1st respectively, 3,4,7,8,11,13 generations, wherein the 7th generation generate two solutions at the same time.
Design process has obtained 8 solutions, is numbered by the sequencing of appearance using alphabetical a~h, table 2, table 3 divide
Each corresponding design variable value and design object value Wei not be solved.Due in design standard and undeclared head speed it is more big more
It is good or the smaller the better, therefore during analysis optimal solution, on the premise of on head, speed meets design requirement, only to other three
Design object is analyzed.The results show that there is no any one solution, its all design object value is set to be superior to setting for other solutions
Desired value is counted, therefore can not determine optimal solution by simply comparing design object value.
Table 2, respectively solve corresponding design variable value
Table 3, respectively solve corresponding design object value
When calculating hypervolume, only three entrance velocity, inlet temperature and inlet angle parameters are sampled, sample size
For 20, the subinterval scope of entrance velocity is ± 0.05m/s, and the subinterval scope of inlet temperature is ± 0.5 DEG C, inlet angle
Subinterval scope be ± 5 °, if interval range exceed design variable excursion, be used as area using the boundary of design variable
Between border.Since head speed has two boundary values, can not decision content be the bigger the better or the smaller the better, therefore only considered |
PMVc |, foot's speed and the vertical temperature difference and three design objects.Reference point is (0.5,0.2,2.8) when calculating hypervolume, three
Coordinate represents respectively | PMVc |, foot's speed (m/s) and the vertical temperature difference (DEG C).
Table 4, local solution quantity and hypervolume
Table 4 represents local solution quantity for the corresponding local solution quantity of each solution and hypervolume, N, and H represents hypervolume size.Solve e
Possess maximum hypervolume, for this optimal solution of design process, the interval solutions represented using control interval form is：Vin=1.25~
1.3m/s, Tin=21.1~22.1 DEG C, Ain=-41~-31 °, Lin=No.4, Lout=No.15.Wherein：Entrance velocity Vin, enter
Mouth temperature Tin, inlet angle Ain, entry position LinAnd outlet port Lout.No represent numbering, entrance and exit totally 30 can
Bit selecting is put, and No.1 represents first position.
Fig. 4 is the curve that calculation amount changes with genetic algebra.When genetic algorithm restrains, 355 case have been calculated altogether and (have been calculated
Amount), solved wherein 195 case are used to search for, 160 case are used for the hypervolume for calculating solution.
Claims (2)
1. a kind of sort method applied to multiple target section genetic algorithm, it is characterised in that this method comprises the following steps：
Step (1), according to design object determine design variable and design object；The design variable of Cabin model is entrance velocity
Vin, inlet temperature Tin, inlet angle Ain, entry position LinAnd outlet port Lout；The design object of Cabin model is (1) |
PMVc|<0.5；(2)0.1m/s<Vhead<0.2m/s；(3)Vfeat<0.2m/s；(4)ΔT<2.8℃；Vhead、VfeatRepresented with Δ T
Head speed, foot's speed and the vertical temperature difference, PMVcIt is as follows for modified PMV, its calculation formula：
PMVc=-0.0758PMV2+0.6757PMV-0.1262
Step (2), obtain multigroup design variable value for meeting design requirement using the cross and variation process of genetic algorithm, calculates new
The design object value of individual, i.e. the boundary condition of design object is determined according to design variable value, is calculated inside design object
Speed, temperature distribution, genetic algorithm crossover probability is 0.8, mutation probability 0.1, is often 24 for number of individuals, obtains each
The corresponding design object value of individual, the condition of convergence is to obtain 8 interval solutions for meeting design requirement；In design process, make at the same time
Individual is ranked up with non-dominated ranking and hypervolume；During using non-dominated ranking, determined according to the design object value of individual
All individual sequences；When calculating hypervolume, it is first determined the subinterval of different designs variable, i.e. variable change scope, then
Using the method for latin hypercube sampling, a certain amount of individual is extracted in subinterval, is calculated, obtain designing mesh accordingly
Scale value, finally selects the individual for meeting design requirement as local solution, calculates these hypervolumes being locally deconstructed into；And then obtain
Design object value；
Step (3), according to design object value, while using hypervolume size and non-dominated ranking be ranked up all individuals：
If the design object value of individual meets design standard, sub- population one is classified to；If the design object value of individual does not meet design
Standard, is classified to sub- population two；In sub- population one, calculate all individual hypervolumes, according to hypervolume size, from greatly to
It is small to be ranked up, it is contemplated that the section characteristic of solution, using the section characteristic of individual hypervolume size evaluation solution, hypervolume is bigger,
Section where illustrating the individual is better；In sub- population two, using non-dominated ranking method, all individuals are ranked up,
Solve multi-objective problem；
Step (4), come sub- population one and sub- two ordering by merging of population, sub- population one before sub- population two；
Step (5), by non-dominated ranking method be ranked up the individual in population, and is selected using tournament algorithm
Select, new population is produced using selection operation：If new population meets convergence, design process terminates；Otherwise continue to intersect
Mutation process, produces new population.
2. it is applied to the sort method of multiple target section genetic algorithm as claimed in claim 1, it is characterised in that wherein described
The calculating of hypervolume comprises the following steps：
Determine the subinterval of different designs variable, i.e. variable change scope, then using the method for latin hypercube sampling,
Quantitative individual is extracted in subinterval, is calculated with Fluid Mechanics Computation, obtains corresponding design object value, selects satisfaction design
It is required that individual as local solution, calculate the hypervolume being locally deconstructed into.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510902573.8A CN105426965B (en) | 2015-12-09 | 2015-12-09 | A kind of sort method applied to multiple target section genetic algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510902573.8A CN105426965B (en) | 2015-12-09 | 2015-12-09 | A kind of sort method applied to multiple target section genetic algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105426965A CN105426965A (en) | 2016-03-23 |
| CN105426965B true CN105426965B (en) | 2018-05-04 |
Family
ID=55505160
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510902573.8A Expired - Fee Related CN105426965B (en) | 2015-12-09 | 2015-12-09 | A kind of sort method applied to multiple target section genetic algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105426965B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107038306A (en) * | 2017-04-14 | 2017-08-11 | 哈尔滨工业大学 | A kind of optimal Latin hypercube experimental design method based on self-adapted genetic algorithm |
| CN112825159B (en) * | 2019-11-20 | 2022-11-29 | 天津大学 | Gasification process parameter interval determining device and method |
| EP4063469A4 (en) * | 2019-11-20 | 2023-08-30 | Tianjin University | Reverse design device for gasification process, and method |
| CN111582552B (en) * | 2020-04-16 | 2023-04-25 | 浙江大学城市学院 | Shared bicycle parking point distribution method based on multi-target genetic algorithm |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7996344B1 (en) * | 2010-03-08 | 2011-08-09 | Livermore Software Technology Corporation | Multi-objective evolutionary algorithm based engineering design optimization |
| US8015127B2 (en) * | 2006-09-12 | 2011-09-06 | New York University | System, method, and computer-accessible medium for providing a multi-objective evolutionary optimization of agent-based models |
| CN103440377A (en) * | 2013-08-27 | 2013-12-11 | 北京航空航天大学 | Aircraft aerodynamic configuration optimum design method based on improved parallel DE algorithm |
| CN103473597A (en) * | 2013-09-05 | 2013-12-25 | 中国寰球工程公司 | Method for optimizing natural gas liquefaction process technological parameters based on genetic algorithm |
| CN103473424A (en) * | 2013-09-23 | 2013-12-25 | 北京理工大学 | Optimum design method for aircraft system based on sequence radial basis function surrogate model |
-
2015
- 2015-12-09 CN CN201510902573.8A patent/CN105426965B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8015127B2 (en) * | 2006-09-12 | 2011-09-06 | New York University | System, method, and computer-accessible medium for providing a multi-objective evolutionary optimization of agent-based models |
| US7996344B1 (en) * | 2010-03-08 | 2011-08-09 | Livermore Software Technology Corporation | Multi-objective evolutionary algorithm based engineering design optimization |
| CN103440377A (en) * | 2013-08-27 | 2013-12-11 | 北京航空航天大学 | Aircraft aerodynamic configuration optimum design method based on improved parallel DE algorithm |
| CN103473597A (en) * | 2013-09-05 | 2013-12-25 | 中国寰球工程公司 | Method for optimizing natural gas liquefaction process technological parameters based on genetic algorithm |
| CN103473424A (en) * | 2013-09-23 | 2013-12-25 | 北京理工大学 | Optimum design method for aircraft system based on sequence radial basis function surrogate model |
Non-Patent Citations (2)
| Title |
|---|
| 应用遗传算法求解流体力学反问题;吴宏平;《人民黄河》;20080331;第30卷(第3期);全文 * |
| 用于区间参数多目标优化问题的遗传算法;孙靖;《中国博士学位论文全文数据库 信息科技辑》;20121015;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105426965A (en) | 2016-03-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105426965B (en) | A kind of sort method applied to multiple target section genetic algorithm | |
| CN108764517B (en) | Method, equipment and storage medium for predicting change trend of silicon content in molten iron of blast furnace | |
| CN109634121B (en) | Multi-parent genetic algorithm air source heat pump multi-objective optimization control method based on radial basis function neural network | |
| CN106527381B (en) | A kind of fast evaluation method towards parallel batch processing machine dynamic dispatching | |
| WO2016169286A1 (en) | Workshop layout method for discrete manufacturing system | |
| CN110363344A (en) | Probability integral parameter prediction method based on MIV-GP algorithm optimization BP neural network | |
| CN109657847A (en) | Failure prediction method in industrial production based on particle group optimizing support vector regression | |
| CN107203590B (en) | Personalized movie recommendation method based on improved NSGA-II | |
| CN104375478B (en) | A kind of method and device of Rolling production process product quality on-line prediction and optimization | |
| CN106650314A (en) | Method and system for predicting amino acid mutation | |
| CN103488561B (en) | A kind of kNN fault detection method of online upgrading master sample model | |
| CN106934722A (en) | Multi-objective community detection method based on k node updates Yu similarity matrix | |
| KR20170078256A (en) | Method and apparatus for time series data prediction | |
| CN108460213A (en) | Based on the T-S models of multi-cluster prototype to the prediction technique and program of gas flowrate in bosh | |
| CN108334898A (en) | A kind of multi-modal industrial process modal identification and Fault Classification | |
| CN114167898A (en) | Global path planning method and system for data collection of unmanned aerial vehicle | |
| CN102129242A (en) | Product quality control method during batch processing production process based on two-layer hybrid intelligent optimization | |
| CN109493916A (en) | A kind of Gene-gene interactions recognition methods based on sparsity factorial analysis | |
| CN111833970A (en) | Construction method and application of cement clinker quality characterization parameter prediction model | |
| CN105389442A (en) | Reverse design method for coupling genetic algorithm, neural network and numerical simulation | |
| Zhang et al. | Embedding multi-attribute decision making into evolutionary optimization to solve the many-objective combinatorial optimization problems | |
| CN105550437A (en) | Reverse design method for indoor environment based on genetic algorithm | |
| CN103902706A (en) | Method for classifying and predicting big data on basis of SVM (support vector machine) | |
| CN108537249A (en) | A kind of industrial process data clustering method of density peaks cluster | |
| CN113468794A (en) | Temperature and humidity prediction and reverse optimization method for small-sized closed space |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180504 Termination date: 20211209 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |