CN103116805A - Staged replacing method for renewing genetic populations - Google Patents
Staged replacing method for renewing genetic populations Download PDFInfo
- Publication number
- CN103116805A CN103116805A CN201310054227XA CN201310054227A CN103116805A CN 103116805 A CN103116805 A CN 103116805A CN 201310054227X A CN201310054227X A CN 201310054227XA CN 201310054227 A CN201310054227 A CN 201310054227A CN 103116805 A CN103116805 A CN 103116805A
- Authority
- CN
- China
- Prior art keywords
- population
- hereditary
- adaptive value
- populations
- individuality
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 27
- 230000002068 genetic effect Effects 0.000 title claims abstract description 25
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 26
- 230000000717 retained effect Effects 0.000 claims abstract 2
- 230000003044 adaptive effect Effects 0.000 claims description 38
- 230000011218 segmentation Effects 0.000 claims description 16
- 230000035772 mutation Effects 0.000 claims description 8
- 230000008774 maternal effect Effects 0.000 claims description 5
- 238000010367 cloning Methods 0.000 claims description 3
- 238000012804 iterative process Methods 0.000 claims description 3
- 230000002028 premature Effects 0.000 abstract description 5
- 238000013377 clone selection method Methods 0.000 abstract 1
- 210000000349 chromosome Anatomy 0.000 description 16
- 230000008569 process Effects 0.000 description 8
- 238000010353 genetic engineering Methods 0.000 description 4
- 230000002759 chromosomal effect Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000036039 immunity Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000004083 survival effect Effects 0.000 description 2
- YBJHBAHKTGYVGT-ZKWXMUAHSA-N (+)-Biotin Chemical compound N1C(=O)N[C@@H]2[C@H](CCCCC(=O)O)SC[C@@H]21 YBJHBAHKTGYVGT-ZKWXMUAHSA-N 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000010429 evolutionary process Effects 0.000 description 1
- 238000012248 genetic selection Methods 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- FEPMHVLSLDOMQC-UHFFFAOYSA-N virginiamycin-S1 Natural products CC1OC(=O)C(C=2C=CC=CC=2)NC(=O)C2CC(=O)CCN2C(=O)C(CC=2C=CC=CC=2)N(C)C(=O)C2CCCN2C(=O)C(CC)NC(=O)C1NC(=O)C1=NC=CC=C1O FEPMHVLSLDOMQC-UHFFFAOYSA-N 0.000 description 1
Images
Abstract
The invention discloses a staged replacing method for renewing genetic populations. In the method, the populations are treated at three stages according to a practical situation after selection, cross and variation, the proper populations are directly retained to a next generation, the populations in the middle are replaced by the proper populations through a clone selection method, and residual improper populations are replaced by nascent populations. By the aid of the method, the whole population keeps diversity, premature convergence is avoided, the usage rate of the superior populations is increased, and accordingly, genetic algorithm performance can be improved.
Description
Technical field
The present invention relates to a kind of follow-on genetic algorithm (Genetic Algorithm, GA), relate in particular to a kind of segmentation replacement method that upgrades hereditary population.
Background technology
Genetic algorithm is the optimisation technique of a kind of practicality, efficient, strong robustness, the way of search of its abandoning tradition, and the process of simulating nature circle biological evolution, carry out random optimization search to object space.In genetic algorithm, the corresponding feasible solution of body one by one, each feasible solution usually can be encoded into and be referred to as chromosomal symbol string.According to the biological evolution process of Darwinian natural selection and the survival of the fittest, the population be comprised of a plurality of individualities is carried out repeatedly based on genetic operation.Genetic algorithm is widely used in the different field such as industry, economic management, communications and transportation, industrial design.
In genetic algorithm, utilize adaptive value to measure the individual adaptedness for living environment, number individual in population is called population scale.Mainly comprise three steps based on genetic operation, i.e. selection, crossover and mutation.Selecting operation is that the next generation is arrived in the individual inheritance that adaptive value in population is high, realizes the survival of the fittest; Crossover and mutation is to produce new individual main method, can maintain the diversity of population, prevents to a certain extent precocity.
In the initial stage of GA operation, relatively whole population, may have a few individual adaptive value very high in colony.Select operation if use, as the roulette system of selection, determine that when whether certain individuality is selected, several chromosomes with higher adaptive value can occupy very high ratio in colony of future generation.Under extreme case or when population size hour, new colony is comprised of a few such chromosome even fully.No matter carry out wherein interlace operation owing to thering are identical chromosomal two individualities, can not produce the chromosome made new advances, will make like this diversity of colony reduce, cause genetic algorithm generation Premature convergence.The later stage simultaneously moved at GA, in colony, the average adaptive value of all individualities is close to the adaptive value of optimum individual in colony, all individualities all are genetic to the next generation with the probability approached, thereby make the evolutionary process regression, are a kind of random selection course, lack competitive.
Summary of the invention
The object of the present invention is to provide a kind of segmentation replacement method that upgrades hereditary population.
For achieving the above object, the present invention has adopted following technical scheme:
In the iterative process of hereditary population, preserve all the time the individual b of a global optimum in the memory population
opt, and in the iterative process each time of genetic algorithm (GA), all can select the individual b that the adaptive value of an epicycle iteration is the highest and be kept in the memory population, simultaneously relatively b and b
optadaptive value, if the adaptive value of b is higher than b
opt, the relevant information of b is copied to b
opt, otherwise b
optremain unchanged.After based on genetic selection, three steps of crossover and mutation, increase the 4th the segmentation replacement step that the present invention proposes when a hereditary population, algorithm flow chart as shown in Figure 1.Operate as follows:
(1) population segmentation
As shown in Figure 2, the individuality in hereditary population is sorted from high to low by adaptive value, then is divided into three parts:
Category-A: 50% population that adaptive value is the highest.In the heredity population by adaptive value sequence in front 50% part will remain into the next iteration process;
Category-B: 37.5% population that adaptive value is lower.In the heredity population, by the part of adaptive value sequence between front 50% and rear 12.5%, with the individuality of replacing in the candidate population, replaced;
C class: 12.5% population that adaptive value is minimum.Be regarded as abandoning population by adaptive value sequence in rear 12.5% part in the heredity population, simulating nature death is abandoned, and with the random new population generated, directly replaces.
When utilizing this algorithm to carry out solving of particular problem, for three parts, can be according to actual conditions, ratio according to category-A is not less than 40%, the ratio of C class is no more than 30% principle, and the ratio that remains into next iterative part (category-A) in population and the ratio that abandons part (C class) are adjusted.
(2) the category-B population is replaced
B in GA
optto judge whether certain individuality can become a good reference standard of cloned object, if because this individual chromosome and b
optdiversity factor lower, this individual adaptive value is also higher.Therefore, after completing (1), calculate the globally optimal solution b of all individualities and genetic algorithm in hereditary population
optthe diversity factor value, then all individualities in hereditary population are sorted from low to high by the diversity factor value, the present invention only gets in hereditary population maternal as the clone at front 25% individuality by the sequence of diversity factor value, these clones are maternal according to document [Liu Xingbao, Cai Zixing etc. for the mixed immunity evolution algorithm [J] of Global Optimal Problem. Xian Electronics Science and Technology University's journal (natural science edition), 2010, 37 (5): 971-980.] the dynamic Strategies For The Cloning in forms the clone population, the clone population is according to after certain variation probability variation, form and replace the candidate population together with the memory population.
Finally will replace the order sequence from high to low according to adaptive value of individuality in the candidate population, and choose successively the individuality that adaptive value is higher and replace in hereditary population by all individualities in the part (category-B) of adaptive value sequence between front 50% and rear 12.5%.
(3) C class population is replaced
C class population belongs to and is not suitable with type, can be considered as abandoning population, uses newborn individual population directly to replace.
Set M=[m for described memory population
1, m
2..., m
t-1] mean, the t time iteration, after the genetic manipulation of experience selection, three steps of crossover and mutation, selected the individual m that adaptive value is the highest from hereditary population
t, by individual m
tadd in set M
The present invention is directed to the Premature convergence and the low defect of high-quality population utilization factor that occur in the GA population, increase again a segmentation replacement step after the hereditary variation operation, can make GA obtain and improve the utilization factor with high-quality chromosome individuality and the compromise of avoiding Premature Convergence, therefore can improve the performance of genetic algorithm.
The accompanying drawing explanation
Fig. 1 is based on the genetic algorithm process flow diagram of segmentation replacement method;
Fig. 2 is the segmentation replacement policy schematic diagram that the present invention designs.
Embodiment
Below in conjunction with accompanying drawing, the invention will be further described.
Genetic algorithm is simulation biological evolution process, optimizes the raw data vector, makes it have higher adaptability.The raw data vector x=[x (1), x (2) ..., x (V)]
t, wherein V is vectorial length, supposes that initial individual collections is Q (0)=[b
1, b
2..., b
i..., b
l], wherein: L is population scale, i individual chromosome is expressed as b
i=[b
i(1), b
i(2) ..., b
i(v) ..., b
i(V)]
t, b
i(v) ∈ 1 ,-1}, v=1 ..., V.
The step of genetic optimization is as follows:
The first step: initialization population.
Determine the adaptive value function of algorithm according to concrete application background, as when reducing the high peak-to-average force ratio of OFDM, adaptive value function [L.Yang, R.S.Chen, Y.M.Siu and K.K.Soo.An Efficient Sphere Decoding Approach for PTS Assisted PAPR Reduction of OFDM Signals[J] .AEU-International Journal of Electronics and Communications, 2007,61 (10): 684-688.] may be defined as:
In formula, max{.} means to choose the element of the value maximum in vector, and E{.} means that vectorial all elements gets average.
Second step: the genetic manipulations such as selection, crossover and mutation.
Selecting operation is to choose some individual inheritance to the next generation from parent, and the roulette method is the most frequently used selection operation, in the t time iteration, population Q (t-1) is carried out to following steps, with acquisition
wherein, t ∈ [1 ..., T], T is maximum iteration time.
(1) by the adaptive value f (b of all individualities
i) be mapped to [0,1] interval, obtain according to formula (2) the weight q that each individual adaptive value accounts for the adaptive value summation of all individualities
i, and calculate accumulation weight A according to formula (3)
i,
I=1 wherein, 2 ..., L.
(2) produce one [0,1] interval interior equally distributed random number r, and and A
irelatively, select to meet A
ithe b corresponding to minimum i value of>=r condition
i, first accumulates weight A
ibe more than or equal to the individuality of r, join
in;
(3) repeat top two steps, until
till the number of middle individuality reaches L.
Interlace operation refers to that two individualities that mutually match exchange chromosomal portion gene separately mutually by certain mode, produces two new individual methods, and it is the most frequently used a kind of cross method that single-point intersects.At first current population
l individuality in random select two individualities, and a point of crossing is set in their chromosome at random, then this some place according to formula (5) in crossover probability p
cmutually exchange the chromosome dyad of two pairing chromosomes.
Variation is to produce new individual householder method, can determine the local search ability of genetic algorithm, in mutation process, and the variation Probability p that individual chromosome obtains by (4) formula
mchange the genic value of variation position, if the original value of variation position is 1, after variation, it is set to-1; Vice versa.
P
c(i)=q
i(1+p
m(i)) (i=1,2 ..., L) b in (5) formula
opt=[b
opt(1), b
opt(2) ..., b
opt(V)]
tit is the globally optimal solution of genetic algorithm.
After selection, crossover and mutation, obtain population
After the t time iteration experience genetic manipulation, from
in select the individual m that adaptive value is the highest
t, by m
tadd memory population set M=[m to
1, m
2..., m
t-1] in, compare m simultaneously
twith b
optadaptive value, if m
tadaptive value compare b
optheight, by m
trelevant information be copied to b
opt, otherwise b
optremain unchanged.
Yet at genetic search in earlier stage, selection may make the chromosome that several adaptive values are higher occupy very high ratio in population, produces precocious phenomenon; In the search later stage, in population between individual chromosome difference very little, can cause evolves stagnates.Therefore, by the segmentation replacement method pair of the 3rd step
adjusted, obtained Q (t), improved the effect of optimization of this iteration.
The 3rd step: segmentation is replaced:
(1) population segmentation
As shown in Figure 2, by population
middle individuality sorts from high to low by adaptive value, and is divided into three sections.
c section: remaining
individuality [b
7L/8+1, b
7L/8+2..., b
l].
(2) B section and C section are partly replaced
The globally optimal solution b of genetic algorithm
optbe one and well judge that can certain individuality become the standard of cloned object, if because this individual chromosome and b
optdiversity factor lower, this individual adaptive value is also higher.Specifically, the computation process of the diversity factor between two individualities is as follows: calculate the corresponding position of two chromosomes (i=1 ..., difference V), finally get addition after the absolute value of difference, and divided by chromosome length V, the mean value of gained is exactly diversity factor.
After the population segmentation, according to formula (6), calculate
middle individual b
iand b
optdiversity factor S (b
i), and according to S (b
i) will
resequence from low to high by the diversity factor value, obtain D.Before getting in D
individuality maternal as the clone, according to document [Liu Xingbao, Cai Zixing etc. for the mixed immunity evolution algorithm [J] of Global Optimal Problem. Xian Electronics Science and Technology University's journal (natural science edition), 2010,37 (5): 971-980.] the dynamic Strategies For The Cloning in, with the variation probability
after variation, form clone population E.To remember population set M admixed together with clone population E, after rearranging from high to low by adaptive value, before selection
individuality is replaced
the individuality of middle B section part, and this part is kept in Q (t).
Finally the number of times when iteration surpasses T, or b
optadaptive value while meeting the end condition of threshold value ξ, just can stop optimizing, directly export b
optas the result after optimizing; Otherwise proceed genetic manipulation and segmentation replacement method.The setting of T and threshold value ξ need to determine according to concrete application background, and while reducing the peak-to-average force ratio (PAPR) of OFDM (OFDM) as applied the present invention to, T and ξ get respectively 5 and 0.1385.
While in the communications field, reducing the ofdm system peak-to-average force ratio, exist and avoid Premature Convergence and reduce the problem that two indexs of algorithm complex are difficult to take into account.Simulation result shows, while applying the present invention to this problem, in algorithm result of calculation, under basically identical condition, the iterations that the present invention needs be only 9.84% left and right of Application standard genetic algorithm, so greatly reduces the time complexity of algorithm.
Claims (3)
1. a segmentation replacement method that upgrades hereditary population, it is characterized in that: in the iterative process each time of genetic algorithm, when hereditary population through selecting, after three steps of crossover and mutation, individuality in hereditary population is sorted from high to low by adaptive value, retained in front 50% part by adaptive value sequence in hereditary population; In the heredity population, by the part of adaptive value sequence between front 50% and rear 12.5%, with the individuality of replacing in the candidate population, replaced; In the heredity population, by the adaptive value sequence, in rear 12.5% part, use the random new population generated to replace.
2. a kind of segmentation replacement method that upgrades hereditary population according to claim 1, it is characterized in that: the diversity factor value of calculating all individualities and globally optimal solution in hereditary population, then all individualities in hereditary population are sorted from low to high by the diversity factor value, get in hereditary population maternal as the clone at front 25% individuality by the sequence of diversity factor value, to clone maternal according to dynamic Strategies For The Cloning composition clone population, after clone's Population Variation, together with the memory population, form and replace the candidate population.
3. a kind of segmentation replacement method that upgrades hereditary population according to claim 1, it is characterized in that: will replace the order sequence from high to low according to adaptive value of individuality in the candidate population, and choose successively the individuality that adaptive value is higher and replace in hereditary population by all individualities in the part of adaptive value sequence between front 50% and rear 12.5%.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310054227.XA CN103116805B (en) | 2013-02-20 | 2013-02-20 | A kind of segmentation replacement method upgrading genetic groups |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310054227.XA CN103116805B (en) | 2013-02-20 | 2013-02-20 | A kind of segmentation replacement method upgrading genetic groups |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103116805A true CN103116805A (en) | 2013-05-22 |
| CN103116805B CN103116805B (en) | 2016-02-03 |
Family
ID=48415174
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310054227.XA Expired - Fee Related CN103116805B (en) | 2013-02-20 | 2013-02-20 | A kind of segmentation replacement method upgrading genetic groups |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103116805B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105306400A (en) * | 2015-08-28 | 2016-02-03 | 褚振勇 | Method for reducing peak-to-average power ratio of transform domain communication system |
| CN108460463A (en) * | 2018-03-20 | 2018-08-28 | 合肥工业大学 | High-end equipment flow line production dispatching method based on improved adaptive GA-IAGA |
| CN110084354A (en) * | 2019-04-09 | 2019-08-02 | 浙江工业大学 | A method of based on genetic algorithm training ANN Control game role behavior |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100076913A1 (en) * | 2008-09-24 | 2010-03-25 | Nec Laboratories America, Inc. | Finding communities and their evolutions in dynamic social network |
| CN102054039A (en) * | 2010-12-30 | 2011-05-11 | 长安大学 | Fitness scaling method for improving overall search capability of genetic algorithm |
-
2013
- 2013-02-20 CN CN201310054227.XA patent/CN103116805B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100076913A1 (en) * | 2008-09-24 | 2010-03-25 | Nec Laboratories America, Inc. | Finding communities and their evolutions in dynamic social network |
| CN102054039A (en) * | 2010-12-30 | 2011-05-11 | 长安大学 | Fitness scaling method for improving overall search capability of genetic algorithm |
Non-Patent Citations (3)
| Title |
|---|
| CLARA PIZZUTI: "A Genetic Algorithm for Community Detection in Social Networks", 《PARALLEL PROBLEM SOLVING FROM NATURE》, 30 December 2008 (2008-12-30), pages 1081 - 1090 * |
| 刘星宝等: "用于全局优化问题的混合免疫进化算法", 《西安电子科技大学学报（自然科学版）》, vol. 37, no. 5, 20 October 2010 (2010-10-20), pages 971 - 980 * |
| 周明等: "《遗传算法原理及应用》", 30 June 1999, 国防工业出版社, article "《遗传算法原理及应用》", pages: 21-25 - 41-50 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105306400A (en) * | 2015-08-28 | 2016-02-03 | 褚振勇 | Method for reducing peak-to-average power ratio of transform domain communication system |
| CN105306400B (en) * | 2015-08-28 | 2018-09-11 | 西京学院 | A method of reducing transform domain communication system peak-to-average power ratio |
| CN108460463A (en) * | 2018-03-20 | 2018-08-28 | 合肥工业大学 | High-end equipment flow line production dispatching method based on improved adaptive GA-IAGA |
| CN108460463B (en) * | 2018-03-20 | 2020-09-01 | 合肥工业大学 | High-end equipment assembly line production scheduling method based on improved genetic algorithm |
| CN110084354A (en) * | 2019-04-09 | 2019-08-02 | 浙江工业大学 | A method of based on genetic algorithm training ANN Control game role behavior |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103116805B (en) | 2016-02-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022193642A1 (en) | Reservoir scheduling multi-objective optimization method based on graph convolutional network and nsga-ii | |
| CN109670650B (en) | Multi-objective optimization algorithm-based solving method for cascade reservoir group scheduling model | |
| CN107229972A (en) | A kind of global optimization based on Lamarch inheritance of acquired characters principle, search and machine learning method | |
| CN106651628B (en) | Regional cooling, heating and power comprehensive energy optimal allocation method and device based on graph theory | |
| CN103036234B (en) | Power distribution network anti-error optimization method | |
| CN104764980B (en) | A kind of distribution line failure Section Location based on BPSO and GA | |
| CN107122843A (en) | A kind of traveling salesman problem method for solving based on improved adaptive GA-IAGA | |
| CN105552892A (en) | Distribution network reconfiguration method | |
| CN104615869A (en) | Multi-population simulated annealing hybrid genetic algorithm based on similarity expelling | |
| CN104268077A (en) | Chaos genetic algorithm based test case intensive simple algorithm | |
| CN104820977A (en) | BP neural network image restoration algorithm based on self-adaption genetic algorithm | |
| CN103116805B (en) | A kind of segmentation replacement method upgrading genetic groups | |
| CN109217284A (en) | A kind of reconstruction method of power distribution network based on immune binary particle swarm algorithm | |
| CN106067074B (en) | A method of network system robustness is promoted by optimizing the switch state of link | |
| CN106127304A (en) | One is applicable to power distribution network Network Topology Design method | |
| CN108537370A (en) | Especially big basin water station group Optimization Scheduling based on hybrid intelligent dimension-reduction algorithm | |
| CN107706907A (en) | A kind of Distribution Network Reconfiguration and device | |
| CN103279796A (en) | Method for optimizing genetic algorithm evolution quality | |
| CN109002878A (en) | A kind of GA Optimized BP Neural Network | |
| CN106815656B (en) | Method for acquiring cascade reservoir energy storage dispatching diagram | |
| CN105550947A (en) | Power distribution network reconstruction method | |
| CN108665068A (en) | The improved adaptive GA-IAGA of water distribution hydraulic model automatic Check problem | |
| CN105376185A (en) | Constant modulus blind equalization processing method based on optimization of DNA shuffled frog leaping algorithm in communication system | |
| CN104867164A (en) | Vector quantization codebook designing method based on genetic algorithm | |
| CN105977966B (en) | A kind of distribution network planning method considering distributed generation resource and distributing automation apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information |
Inventor after: Liang Zhonghua Inventor after: Zhang Shaobo Inventor after: Feng Xingle Inventor after: Lu Ping Inventor after: Yang Nan Inventor after: Xue Guowei Inventor after: Zhang Huaikai Inventor after: Bai Wenhao Inventor after: Chen Li Inventor before: Feng Xingle Inventor before: Zhang Shaobo Inventor before: Lu Ping Inventor before: Yang Nan Inventor before: Xue Guowei |
|
| COR | Change of bibliographic data | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160203 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |