CN110598833A - High-dimensional particle swarm optimization method for packet evolution - Google Patents
High-dimensional particle swarm optimization method for packet evolution Download PDFInfo
- Publication number
- CN110598833A CN110598833A CN201910869500.1A CN201910869500A CN110598833A CN 110598833 A CN110598833 A CN 110598833A CN 201910869500 A CN201910869500 A CN 201910869500A CN 110598833 A CN110598833 A CN 110598833A
- Authority
- CN
- China
- Prior art keywords
- dimension
- particle
- group
- evolution
- dimensions
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000002245 particle Substances 0.000 title claims abstract description 62
- 238000000034 method Methods 0.000 title claims abstract description 21
- 238000005457 optimization Methods 0.000 title claims abstract description 21
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims 1
- 230000007547 defect Effects 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Abstract
The invention discloses a high-dimensional particle swarm optimization method for packet evolution, which comprises the following steps: and (3) dividing the dimensions of the high-dimensional particles into groups, sequentially evolving by taking the groups as units, and calculating the fitness value after each group is evolved so as to determine whether the group completes the evolution. In an iteration process, all the dimension grouping evolution is completed, and the particle position is updated. In the process of high-dimensional particle swarm evolution, the invention not only solves the defect of poor evolution and optimization performance of all dimensions by grouping and taking the sub-dimension groups as evolution units, but also solves the defect of slow evolution and optimization speed of each dimension in turn.
Description
Technical Field
The invention relates to a particle swarm optimization method, in particular to a high-dimensional particle swarm optimization method based on grouping evolution.
Background
The particle swarm optimization is a heuristic global optimization algorithm, and the characteristics of the next searching direction and speed are adjusted according to the individuals in the whole swarm who are in the optimal position at present and the historical optimal position of the individuals in the swarm when the whole swarm searches for a certain target. In the optimization process of the standard particle swarm algorithm, each iteration updates the numerical values of all dimensions of the particles simultaneously with the particles as a whole, and although the updating by the method can enable each dimension of the particles to be close to the current optimal solution, the method cannot ensure that each dimension of the particles moves towards the optimal direction, so that the optimization effect is influenced.
Instead of updating all dimensions of the solution vector at the same time, each dimension of the solution vector is updated in turn, and the fitness value after each dimension update is evaluated, if the fitness value is not better than the value before the update, the solution vector will not update the value of the dimension, and the value of the dimension remains unchanged. And the optimization performance is greatly improved through sub-dimension optimization. When the dimensionality is not high, the optimization speed is acceptable. When the dimension of the solution vector is very high, the optimizing speed of the method is greatly reduced, so that a grouping method is adopted, high dimensions are grouped, and the sequential evolution is carried out by taking the groups as units, so that the dimension grouping is ensured to move towards the optimal direction to a certain extent, and the optimizing speed of high-dimensional particles is improved.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provide a high-dimensional particle swarm optimization method, which adopts a grouping method to group high dimensions and sequentially evolves by taking groups as units, thereby ensuring that dimension groups move to the optimal direction to a certain extent and improving the optimization speed of high-dimensional particles.
In order to achieve the above object, the present invention provides a high-dimensional particle swarm optimization method, which comprises the following steps:
step 1, randomly initializing a population, wherein the population comprises n particles, and a search space is D-dimensional. The initial position of each particle is zi= (zi1,zi2,...,ziD)。
Step 2, calculating the fitness value of each particle, and selecting the particle z with the optimal fitness valuegAs a global optimum particle pg=zg。
Step 3, grouping the D-dimensional space, and setting the number of the sub-dimensions of each group as the number in consideration of the balance of the number of the groups and the number of the dimensions in the groupThe number of groups is,
(1)
The number of sub-dimensions in the d-th group after grouping determination is
When in useWhen the temperature of the water is higher than the set temperature, (2)
when in useWhen the temperature of the water is higher than the set temperature,
step 4. according to the grouping result of the previous step, the position of the ith particle (i =1, 2.. n) can be represented as a position vector zi= (zi1,zi2,...,zim)。
Step 5, in the evolution process of the particle swarm, each dimension group is evolved in sequence, the speed update of each dimension group refers to the speed update of the standard particle swarm, the position only updates the position of the current dimension group,is the velocity vector for the d-dimensional grouping of the ith particle,is the position of the ith particle in the d-dimensional group.
(3)
(4)
Calculating the fitness value after updating each dimension, and if the fitness value is better than the fitness value before updating, performing position updating on the grouped sub-dimensions, so that the current position of the particle is always the current optimal value of the particle, only moving to the current global optimal value is needed, and only considering the global optimal value p in the formula (3)g。
If it is (5)
If it isThen the sequential evolution of the sub-dimensions within the group is performed,is the velocity in dimension j in the dimension d packet of the ith particle,is the position of the jth dimension in the jth dimension component of the ith particle
(6)
(7)
。 (8)
Step 6, when all dimension groups of the solution vector are updated, executing
(9)
And 7, when the algorithm runs to the maximum iteration times or the fitness value of the optimal solution searched by the algorithm is lower than the set minimum fitness threshold value, outputting a result.
In the process of high-dimensional particle swarm evolution, the invention not only solves the defect of poor evolution and optimization performance of all dimensions by grouping and taking the sub-dimension groups as evolution units, but also solves the defect of slow evolution and optimization speed of each dimension in turn.
Drawings
FIG. 1 is a schematic flow diagram of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings and examples.
Example (b):
as shown in fig. 1: the invention discloses an improvement method based on a high-dimensional particle swarm optimization algorithm, which comprises the following steps of:
step 1: and (3) randomly initializing a population, wherein the population comprises 1000 particles, the search space is 10000 dimensions, and the maximum iteration number is 1000. The initial position of each particle is zi= (zi1,zi2,...,zi10000)。
Step 2: calculating the fitness value of each particle, and selecting the particle z with the optimal fitness valuegAs a global optimum particle pg=zg。
And step 3: the 10000-dimensional space is divided into groups, and the number of the sub-dimensions of each group is preliminarily set to be the number considering the balance of the number of the groups and the number of the dimensions in the groupThe number of the packets is m,
(1)
the number of sub-dimensions in the d-th group after grouping determination is
When in useWhen the temperature of the water is higher than the set temperature, (2)
when in useWhen the temperature of the water is higher than the set temperature,
and calculating to obtain: g =100, m =100, i.e. 10000 dimensions, into 100 groups, each group containing 100 dimensions
And 4, step 4: in the evolution process of the particle swarm, simultaneously evolving the 100 dimensions in each group of the particles in sequence, updating the speed of each dimension group according to the speed of the standard particle swarm, updating the position of the current dimension group only by the position,is the velocity vector for the d-dimensional grouping of the ith particle,is the position of the ith particle in the d-dimensional group.
(3)
(4)
Calculating the fitness value of each group after updating, and if the fitness value is superior to the fitness value before updating, updating the positions of the sub-dimensions of the group, so that the current position of the particle is always the current optimal value of the particle, only moving to the current global optimal value is needed, and only considering the full fitness value in the formula (3)Local optimum value pg。
(5)
Performing a sequential evolution of 100 sub-dimensions within the group,is the velocity in dimension j in the dimension d packet of the ith particle,is the position of the jth dimension in the jth dimension component of the ith particle
(6)
(7)
(8)
And 5: when the 100 dimensional groups of the solution vector are updated, the execution is carried out
(9)
Step 6: and when the fitness value of the optimal solution searched by the algorithm is lower than the set minimum fitness threshold value or the maximum iteration number is 1000, outputting the result.
Claims (1)
1. A high-dimensional particle swarm optimization method for packet evolution comprises the following steps:
step 1, randomly initializing a population, wherein the population comprises n particles, a search space is D-dimension, and the initial position of each particle is zi= (zi1,zi2,...,ziD);
Step 2, calculating the fitness value of each particle, and selecting the particle z with the optimal fitness valuegAs a global optimum particle pg=zg;
Step 3, grouping the D-dimensional space, and setting the number of the sub-dimensions of each group asThe number of groups is,
;
The number of sub-dimensions in the d-th group after grouping determination is:
When in useWhen the temperature of the water is higher than the set temperature,
when in useWhen the temperature of the water is higher than the set temperature,;
step 4. according to the grouping result of step 3, the position of the ith particle can be expressed as a position vector zi= (zi1,zi2,...,zim),i=1,2,...n;
Step 5, granulatingIn the evolution process of subgroups, each dimension group is evolved in turn, the speed update of each dimension group refers to the speed update of a standard particle swarm, the position only updates the position of the current dimension group,is the velocity vector for the d-dimensional grouping of the ith particle,is the position of the ith particle in the d-dimension group:
calculating the fitness value after each dimension is updated, if the fitness value is better than the fitness value before updating, updating the positions of the grouped sub-dimensions, so that the current position of the particle is always the current optimal value of the particle, only moving to the current global optimal value,only the global optimum p is considered in the calculation formulag;
If it is
If it isThen the sequential evolution of the sub-dimensions within the group is performed,is the velocity in dimension j in the dimension d packet of the ith particle,is the position of the jth dimension in the ith particle's jth dimension grouping:
;
step 6, when all dimension groups of the solution vector are updated, executing
;
And 7, when the algorithm runs to the maximum iteration times or the fitness value of the optimal solution searched by the algorithm is lower than the set minimum fitness threshold value, outputting a result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910869500.1A CN110598833A (en) | 2019-09-16 | 2019-09-16 | High-dimensional particle swarm optimization method for packet evolution |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910869500.1A CN110598833A (en) | 2019-09-16 | 2019-09-16 | High-dimensional particle swarm optimization method for packet evolution |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110598833A true CN110598833A (en) | 2019-12-20 |
Family
ID=68859569
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910869500.1A Pending CN110598833A (en) | 2019-09-16 | 2019-09-16 | High-dimensional particle swarm optimization method for packet evolution |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110598833A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112949154A (en) * | 2021-03-19 | 2021-06-11 | 上海交通大学 | Parallel asynchronous particle swarm optimization method and system and electronic equipment |
CN113450029A (en) * | 2021-08-30 | 2021-09-28 | 广东电网有限责任公司湛江供电局 | Super-dimensional triangular optimization method and power resource scheduling optimization system |
-
2019
- 2019-09-16 CN CN201910869500.1A patent/CN110598833A/en active Pending
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112949154A (en) * | 2021-03-19 | 2021-06-11 | 上海交通大学 | Parallel asynchronous particle swarm optimization method and system and electronic equipment |
CN112949154B (en) * | 2021-03-19 | 2023-02-17 | 上海交通大学 | Parallel asynchronous particle swarm optimization method and system and electronic equipment |
CN113450029A (en) * | 2021-08-30 | 2021-09-28 | 广东电网有限责任公司湛江供电局 | Super-dimensional triangular optimization method and power resource scheduling optimization system |
CN113450029B (en) * | 2021-08-30 | 2022-01-25 | 广东电网有限责任公司湛江供电局 | Super-dimensional triangular optimization method and system for power resource scheduling optimization system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111353582B (en) | Particle swarm algorithm-based distributed deep learning parameter updating method | |
CN113296496B (en) | Gravity self-adaptive step length bidirectional RRT path planning method based on multiple sampling points | |
CN110598833A (en) | High-dimensional particle swarm optimization method for packet evolution | |
CN104035438A (en) | Self-adaptive multi-target robot obstacle avoidance algorithm based on population diversity | |
CN107506865A (en) | A kind of load forecasting method and system based on LSSVM optimizations | |
CN115509239B (en) | Unmanned vehicle route planning method based on air-ground information sharing | |
CN114047770B (en) | Mobile robot path planning method for multi-inner-center search and improvement of wolf algorithm | |
CN111173573A (en) | Identification method for power object model of steam turbine regulating system | |
CN110530373B (en) | Robot path planning method, controller and system | |
CN114611801A (en) | Traveler problem solving method based on improved whale optimization algorithm | |
CN115525038A (en) | Equipment fault diagnosis method based on federal hierarchical optimization learning | |
CN110852435A (en) | Neural evolution calculation model | |
CN114708479A (en) | Self-adaptive defense method based on graph structure and characteristics | |
CN115222006A (en) | Numerical function optimization method based on improved particle swarm optimization algorithm | |
CN104566797A (en) | Central air conditioner cooling tower fan frequency control method | |
CN109378036A (en) | A kind of Heuristic Method based on two-line hybrid rice breeding mechanism | |
CN110929851A (en) | AI model automatic generation method based on computational graph subgraph | |
CN107507210B (en) | Genetic algorithm-based image edge detection method and device | |
CN113188243B (en) | Comprehensive prediction method and system for air conditioner energy consumption | |
CN114970728A (en) | DHSSA (distributed Hash analysis for optimization) optimized K-means complementary iterative vehicle type information data clustering method | |
CN116859903A (en) | Robot smooth path planning method based on improved Harris eagle optimization algorithm | |
CN112148030B (en) | Underwater glider path planning method based on heuristic algorithm | |
CN108111535A (en) | A kind of optimal attack path planing method based on improved Monte carlo algorithm | |
CN114519457A (en) | Provincial intelligent energy service platform task scheduling method and system based on particle swarm | |
CN107679326A (en) | A kind of two-value FPRM circuit areas and delay comprehensive optimization method |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191220 |