CN109242290A - A kind of unmanned aerial vehicle group action scheme automatic generation method - Google Patents
A kind of unmanned aerial vehicle group action scheme automatic generation method Download PDFInfo
- Publication number
- CN109242290A CN109242290A CN201810988692.3A CN201810988692A CN109242290A CN 109242290 A CN109242290 A CN 109242290A CN 201810988692 A CN201810988692 A CN 201810988692A CN 109242290 A CN109242290 A CN 109242290A
- Authority
- CN
- China
- Prior art keywords
- harmony
- task
- platform
- subtask
- priority
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06316—Sequencing of tasks or work
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Abstract
The present invention discloses a kind of unmanned aerial vehicle group action scheme automatic generation method: one, determining task core and scene information；Two, task core is decomposed into several subtasks using genetic algorithm by subtask decomposing module；Three, priority of subtask auto-sequencing module --- harmony population generates；Four, scheduling of resource module carries out harmony evaluation；Five, priority of subtask auto-sequencing module --- harmony data base updates；Six, excellent alternative is exported.The present invention gets rid of artificial participation, realizes increasingly automated, quick response rate and intelligent coordinatedization；The scheme of generation guarantees outstanding property, diversity, can sort by outstanding degree for selecting；The action scheme of generation is more efficient when executing, purpose is more clear；More efficient when search, being not required to traverse all arrangements can reach outstanding solution；The evaluation of optimal scheduling and best priority sequence harmony individual can be carried out simultaneously, while guaranteeing the accuracy of timing and scheduling of resource, increase computational efficiency；Modular algorithm compatibility is higher.
Description
Technical field
The present invention relates to a kind of unmanned aerial vehicle group action scheme automatic generation method, especially a kind of decomposed based on subtask to lose
Propagation algorithm, the priority of subtask sequence harmonic search algorithm, scheduling of resource Dynamic and Multi dimensional list planning algorithm unmanned aerial vehicle group row
Dynamic scheme automatic generation method, can be quick, effective and without the completely automatic generation unmanned aerial vehicle group action side of human intervention
Case belongs to intelligent algorithm module and unmanned aerial vehicle group mission command scheduling field.
Background technique
During the intelligence of complicated unmanned aerial vehicle group, corporate action schemes generation, in order to increase formation efficiency and side
The outstanding and diversity of case often introduces the generation that various intelligent algorithms are automated using computer.In recent years, with nothing
The development of man-machine technology, unmanned plane are widely used in every field.And single rack unmanned plane limitation is stronger, in large size
During task execution, unmanned plane is often formed into columns in the form of unmanned aerial vehicle group and is equipped.Unmanned aerial vehicle group has quick response, efficiently association
With, task execution is accurate the features such as, and if commanded manually using people as commanding officer, can not undoubtedly play its special advantage.Therefore nothing
The research of man-machine group's action scheme automatic generation method is necessary.Outstanding automatic generation method can be such that unmanned aerial vehicle group executes
More quickly, efficiently, open strong, reaction speed is fast for task, and the speed that can be faster than mankind commander carries out the amendment of task
And in face of more diversified, more complicated task.So unmanned aerial vehicle group action scheme automatic generation method has good development
Prospect and practical application value.
Currently, it is subtask taking human as division that unmanned aerial vehicle group action scheme, which generates mostly, empirically manual alignment priority,
It is generation of the input using algorithm progress action scheme with subtask.And this generation method, due to subtask decompose with it is excellent
First grade generation phase, there is artificial participation, so can not be known as automatic generation method, there is significant limitation, can not be detached from people
For factor.Educational circles there is no a kind of completely automatic action scheme generation method at present, have scholar to use CS (cuckoo search)
Algorithm has certain automaticity in conjunction with MPDLS (multipriority list Dynamic Programming) algorithm in priority degree, but
Due to also needing artificial decomposition subtask and priority level initializing, unmanned aerial vehicle group action scheme is much also not achieved and automatically generates
Required automation precision.Therefore in order to meet the needs of unmanned aerial vehicle group action scheme generation, a kind of unmanned aerial vehicle group action scheme
The it is proposed of the method automatically generated is meaningful.
Summary of the invention
The present invention is improved for traditional unmanned aerial vehicle group action scheme generation method, proposes a kind of unmanned aerial vehicle group row
Dynamic scheme automatic generation method.The method of the present invention is based on three modules, and each module is based on a kind of algorithm, i.e. mould is decomposed in subtask
Block, priority of subtask sorting module, scheduling of resource module respectively correspond subtask Decomposition Genetic Algorithm, priority sequence cloth
Paddy bird searching algorithm and multidimensional list dynamic programming algorithm.The method of the present invention can be realized Given task core, i.e., one is always appointed
Business, such as obtain article, eliminate target, reach designated position or target is given first aid to automatically generate by modularization operation and hold completion
The time series of the execution specific action of each platform of the action scheme of task core, i.e. unmanned aerial vehicle group.The inventive method emphasis
It is to propose that subtask decomposing module and priority of subtask auto-sequencing module can be efficiently and rapidly without artificial participation
Generate the excellent alternative of multiplicity.
Basic ideas of the invention are as follows:
(1) subtask decomposing module
The input of the method for the present invention is a task core, by this task core by the income of each task object of determination
Value.First, it needs to be modeled according to task environment.Our unmanned plane is modeled as resource platform, that is, carrying has different energy
The platform of the resource of power, a unmanned aerial vehicle group are exactly a platform group.Moving distance and financial value needed for platform, which weights, to be formed
Function is comprehensive income degree, which is clustered by comprehensive income degree using genetic algorithm, by comprehensive income
Degree is clustered at a distance of closer task object together into the same subtask.
(2) priority of subtask auto-sequencing module
Second module of the method for the present invention is priority of subtask auto-sequencing module, raw using harmonic search algorithm
At several harmony populations, each harmony population represents a priority sequence, and each priority sequence of generation enters money
A harmony evaluation is returned to after carrying out optimal scheduling in the scheduler module of source, then iteration finds out optimal harmony individual.
(3) scheduling of resource module
The third module of the method for the present invention is scheduling of resource module, presses second using Dynamic and Multi dimensional list planning algorithm
The priority of priority sequence in module carries out scheduling of resource, and will evaluate at this time optimal scheduling scheme, returns to the
Two modules are evaluated as harmony.The individual corresponding optimal scheduling scheme of optimal harmony is optimal action scheme, and can be on-demand
Retain time excellent alternative.
A kind of unmanned aerial vehicle group action scheme automatic generation method of the invention, includes the following steps:
Step 1: task core and scene information are determined.
Step 2: task core is decomposed into several subtasks using genetic algorithm by subtask decomposing module.
(1) genetic algorithm chromosome indicates
The chromosome expression of genetic algorithm uses binary coded form, based on the scene information that step 1 determines, each
Chromosome is made of matrix F, and wherein n is the number of subtask, and m is the number of known target.Each element takes in F matrix
Value is determined that F matrix coding situation is as shown in table 1 by following formula.
Table 1
After target distribution schemes matrix F has been determined, each unfriendly target under battlefield surroundings can be according to F
Matrix result is respectively allocated to the subtask of different stage.Then each chromosome of genetic algorithm is a F matrix.
(2) fitness function of subtask Decomposition Genetic Algorithm
First when the existing perception target of enemy is carried out mission planning, to be added according to the position and income degree of target
Power obtains its " comprehensive income index " weight:
Weight=location × α+gain × (1- α)
Wherein factor alpha weight is provided by method user, and meaning is method user for positioning index and proceeds indicatior
Pay the utmost attention to degree.The comprehensive income index that each target is matched with subtask is organized into matrix W.By matrix W
With matrix F dot product, comprehensive income index matrix A after being distributed.
The variance of every a line in comprehensive income index matrix A is calculated, then by each row variance phase of comprehensive income index matrix A
Add, if sum of variance is smaller, then it is assumed that individual fitness is higher.Therefore, the concrete form of fitness function are as follows:
In formula, weight (fik) be matrix F in element (i.e. target) comprehensive income index, E (weightn) it is synthesis
The expectation of data line in proceeds indicatior matrix A.
(3) selection operation of subtask decomposition algorithm
The method of the present invention uses wheel disc method, i.e., the selection of individual is carried out using fitness ratio selection mode, per each and every one
The select probability of body is directly proportional to its fitness.If population scale is N, i-th (i=1,2 ..., N) individual is suitable in population
Response is fi, then i-th individual select probability pi, shown in following formula:
(4) crossover operation of subtask decomposition algorithm
Due to the operation of subtask decomposition algorithm object and output the result is that target distribution schemes one by one, i.e., per each and every one
The number of " 1 " is equal to distributed number of targets in body, so if changing the structure of original individual and making in crossover process
The number of " 1 " becomes more or tails off in individual, will lead to the task boundary scheme that algorithm provides occur target it is idle or occur it is extra
The problems such as target, this apparently not it is expected the result seen.
To ensure that the number of " 1 " in every row individual after crossover operation immobilizes, propose that a kind of regulating measures calculate heredity
The basic crossover operation of method is improved.Specific step is as follows: vacation lets a and b be two male parents to be intersected, and have one it is laggard
The storehouse first gone out, As, Bs represent the numerical value on the position s of two male parent's chromosome, if As, Bs two different, first not temporarily
Exchange this two, but they be put into storehouse and is stored, then proceed to search A and B subsequent bit, if discovery there is also
Ai, Bi this to phase dystopy, and Ai and Bi be also it is different, then exchange two groups of As, Bs and Ai, Bi gene positions simultaneously, thus may be used
To ensure that two individuals of A and B still ensure that in individual that the number of " 1 " immobilizes after crossover operation.
(5) mutation operation of subtask decomposition algorithm
Equally, it is not intended to the mutation operation of genetic algorithm to destroy the stability of original scheme individual, also here to ensure to become
The number of " 1 " immobilizes in each individual after ETTHER-OR operation, must just use special variation method.Specific step is as follows: root
Determine whether a certain position s of a line genes of individuals sequence morphs according to given mutation probability, if it is, selecting in same a line
Take another random order i, it is assumed that this two be also it is different, just s and i are directly exchanged, otherwise do not exchanged, to reach change
Different effect.
(6) hereditary termination condition
Specified genetic algebra just stops the progress of genetic algorithm.
It, can be according to existing scene information (the mainly geographical location seat of known target by Revised genetic algorithum
Be marked with and income degree) task core is decomposed into the subtask of different stage.
Step 3: priority of subtask auto-sequencing module --- harmony population generates.
Harmony population is generated using harmonic search algorithm.This step is priority of subtask auto-sequencing module and step 4
Scheduling of resource module the nested first step.Nested WFTA module realizes function are as follows: uses priority of subtask auto-sequencing harmony
Searching algorithm generates several harmony individual, and each harmony individual represents a kind of priority of subtask sortord, by each
Priority of subtask sortord enters step in the Dynamic and Multi dimensional list planning algorithm of four scheduling of resource module and carries out resource
Scheduling, obtained resource utilization and task completion time are the harmony evaluation of harmony individual.Nested WFTA module shares step
Three arrive three step of step 5.
(1) initial parameter
Following parameter need to be initialized:
The size HMS of harmony data base: for the size of harmony population.
Harmony data base probability HMCR: the probability of a harmony is taken out from existing population (HM harmony library).
Tone finely tunes probability P AR: the probability being finely adjusted to the harmony taken out.
Tone finely tunes bandwidth BW: the amplitude of fine tuning.
The number T of creationmax: number, that is, iteration number of adjustment.
(2) harmony data base HMS is initialized
Generate HMS harmony (harmony is understood as individual, and HMS harmony is understood as population) at random in solution space, with to
Measure X1,X2,…,XHMSIt indicates, each harmony represents the sortord of a priority of subtask.It is put into harmony data base, and
Record corresponding f (X), the form in harmony library are as follows:
Wherein, n represents the number of subtask, i-th of harmony XiInRepresent j-th subtask in harmony individual
Priority size.F (X) represents the harmony evaluation of the harmony individual (i.e. priority of subtask sequence).
Step 4: scheduling of resource module --- harmony evaluation
The step is the scheduling of resource module in nested algorithm, and function is the f (X) calculated in HM harmony library, i.e. harmony
Evaluation.
It is individual for each harmony, i.e. row X in HMi, (MDLS) algorithm is planned using Dynamic and Multi dimensional list to realize
Scheduling of resource of the given priority to task.
(1) priority of task
When all predecessor tasks (having to completing for task before task processing) of a certain task have all been handled
When complete, which just enters in eligible task collection READY locating for allowable resource task.Concentrate selection excellent in READY
First gradeMaximum subtask j carries out platform resource scheduling first, as waiting task.
(2) idle platform collection
Either with or without processing task can calling platform be put into idle platform collection FREE, the selection of platform just directly from
It is selected in FREE.
(3) platform group selection
Platform group selection is the platform group that selection executes waiting task, and platform group selection is Dynamic and Multi dimensional list planning
The key component of MDLS algorithm.Here the selection of platform group is to calculate priority P to platform one by one, i.e. platform is applicable in task
Degree, formula are as follows:
P=T+ ∑ R*T
In formula, T represents the time (normalized) that platform is moved to task objective ground, and ∑ R*T represents platform items and appoints
The sum of the processing capacity for required ability of being engaged in.
Then, according to the ability of the priority P sequence resource of arrangement superposition from big to small, check superposition ability whether be more than
The resource requirement of task stops the selection to platform once the demand for reaching task, then starts to trim platform group, picks
Except the platform of redundancy.
After so completing platform group selection, i.e., the task distribution platform is terminated.By required platform from idle platform
It is removed in collection FREE, and the task is removed from eligible task collection READY, which is put into a new set ALREADY, i.e.,
To have distributed task-set.
(4) time updates
It, will when the resource capability needed for waiting task is greater than the sum of all platform capabilities in idle platform collection FREE
Carry out the update of time.Time passage has task completion until having to have distributed in task-set ALREADY, then by the completion task
In next task merging eligible task collection READY, and the completion task institute is discharged using platform, is placed in idle platform collection FREE
In, and record each platform time for being used, j-th of platform be denoted as tj.Step (1) priority of task relaying is returned to later
Continuous algorithm, matches until without task dividable.At this point, scheduling of resource complete, record at this time the time passage at the time of T.
(5) harmony is evaluated
Scheduling of resource module final step is after having carried out Dynamic and Multi dimensional list planning algorithm, to dispatch and carry out to Current resource
Harmony evaluation.
Firstly, when algorithm proceeds to here for the first time, for X at this timei, harmony evaluation is following to be calculated:
By step (4) the time t that is used of task total time T (i) at this time and each platformj(i), it calculates separately
Complete task time reference value TP and platform utilization rate reference value PP.
TP=T
In formula, T represents the first sub-distribution task total time, tjRepresent the time that each platform of the first sub-distribution is used, N
Represent platform sum.
Therefore using platform utilization rate reference value, completion task time and platform utilization rate are normalized, obtained
Time priority weight coefficient T when each algorithm carries outP(i) and PP(i)
In formula, T (i) represents i-th distribution task total time, tjIt represents i-th and distributes the time that each platform is used,
N represents platform sum.
For this method, it is desirable to obtain completing that task time is most short and the highest action scheme of platform utilization efficiency.
Open parameter beta is set, and the size of β is determined for completing task time tendentious height.β is bigger, represents method use
Person is higher for shortening completion task time requirement.
Comprehensive priority coefficient are as follows:
Note harmony is evaluated as the inverse of comprehensive priority coefficient, is f (Xi)=Pr (i)
Then by each f (Xi) return in HM, complete harmony library HM.
Step 5: priority of subtask auto-sequencing module --- harmony data base updates
(1) a new harmony is generated
A random number r is generated between [0,1], is compared with harmony data base probability HMCR, if r <
HMCR takes out a harmony variable at random from harmony data base, otherwise, generates a harmony variable at random from solution space.By
Above content obtains a harmony variable, if this harmony variable be from harmony library memory obtained in, it is necessary to this and
Sound variable is finely adjusted, and a random number s is generated between [0,1].If s < PAR (tone fine tuning probability), according to fine tuning bandwidth
BW obtains a new harmony variable to be adjusted to obtained harmony variable；Otherwise, it does not make any adjustments.It finally obtains
New harmony Xnew。
(2) harmony is evaluated
By XnewIt substitutes into and carries out harmony evaluation in step 4.
(3) harmony data base is updated
To XnewIt is assessed, i.e. f (Xnew), if one worst better than the functional value in the HM of harmony library, i.e. f (Xnew)<f
(Xworst), then by XnewInstead of the worst harmony X of functional value in the HM of harmony libraryworst；Otherwise, it does not make an amendment.
(4) step (1) to step (3) is repeated, until number of repetition (adjustment number) reaches Tmax。
Step 6: output excellent alternative.
Reach T in step 5 number of repetitionmaxAfterwards in gained harmony library HM, f (X) maximum harmony individual X is detectedi, then
The i priority of subtask is ordered as optimal sequencing.X in step 4 scheduling of resource module where itiCorresponding scheduling of resource is
Optimal scheduling of resource, that is, optimal action scheme.
A kind of unmanned aerial vehicle group action scheme automatic generation method of the invention, advantage and effect are:
1. propose a kind of full automatic unmanned aerial vehicle group action scheme automatic generation method, get rid of that artificial participate in can be by appointing
Business core automatically generates action scheme, realizes increasingly automated, quick response rate and intelligent coordinatedization.
2. scheme generated both ensure that outstanding property, diversity is in turn ensured, and can be ordered for selecting by outstanding degree
With.
3. the subtask that subtask decomposing module is decomposed is sorted out by comprehensive income degree, when action scheme generated executes
It is more efficient, purpose is more clear.
4. use harmonic search algorithm, since the performance individual (i.e. the priority of individual task) in single harmony can be with
Voluntarily by suggesting turning down/being turned up part, therefore more efficient in search, it is not required to traverse all arrangements and can reach outstanding solution.
5. priority of subtask module and scheduling of resource module combine, while carrying out optimal scheduling and best priority sequence
The evaluation of harmony individual, while guaranteeing the accuracy of timing and scheduling of resource, and increase computational efficiency.
6. three modules that the method for the present invention uses, modular algorithm compatibility are higher.
Detailed description of the invention
Fig. 1 is the method for the present invention basic flow chart.
Fig. 2 is subtask Decomposition Genetic Algorithm basic flow chart.
Fig. 3 is subtask decomposing module and scheduling of resource modules nests flow chart.
Fig. 4 is output result gunter diagram form.
Specific embodiment
The method of the present invention is described in further detail presently in connection with attached drawing.
A kind of basic procedure of unmanned aerial vehicle group action scheme automatic generation method is as shown in Figure 1, its main part is that son is appointed
Business decomposing module, priority of subtask sorting module, three modules of scheduling of resource module.This process is pressed below, in conjunction with specific
Embodiment elaborates.
Embodiment is as follows: fire occurs for certain forest, shares 10, ignition point to be put out.Before now sending 20 frame unmanned plane and fire-extinguishings
Toward Fight Fire in Forest.Task core is to put out the flame of all ignition points, and prevent from further spreading.First according to task core
Model of place is established, the first step is clustered in the decomposing module of subtask, by unmanned plane away from ignition point distance and intensity of a fire size
Several subtasks are decomposed into, second step generates several priority sequence waiting moneys using harmonic search algorithm to these subtasks
Source scheduling, third step carry out the scheduling of unmanned plane resource to the priority sequence of each harmony individual of harmonic search algorithm, press
According to parameters such as control of fire ability, intensity of a fire extinguishing capability, unmanned plane speed as resource capability, planned using Dynamic and Multi dimensional list
Algorithm carries out scheduling of resource.Finally using the corresponding optimal scheduling of the optimal harmony individual of harmonic search algorithm as optimal action scheme
Output.Detailed process is as follows:
S1 determines task core and scene information:
According to task core, it may be determined that final task target., should be in conjunction with concrete scene information when using this method, judgement
The information and quantity of the information and quantity and target of platform out, is modeled.It should be determined in task according to scene characteristics first
Required elimination or the resource requirement for reaching target, such as detection demand, the defence demand, bombing demand in battlefield, and determine flat
The resource capability that there is platform corresponding these types of demand to have, such as anti-detectivity, penetration ability can resist bombing capacity.
Demand and ability is corresponding, scene is modeled.
It in the embodiment, puts out the flame of all ignition points and prevents from further spreading, be task core.According to this task
Core, will be shown in scene modeling following 2 and table 3.
| Ignition point number | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| Intensity of a fire size | 10 | 8 | 6 | 8 | 10 | 8 | 6 | 8 | 12 | 12 |
| Fire spreading ability | 12 | 6 | 16 | 10 | 12 | 6 | 16 | 10 | 8 | 8 |
| Away from base distance | 10 | 24 | 8 | 12 | 10 | 24 | 8 | 12 | 20 | 10 |
Table 2
| Unmanned plane number | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| Intensity of a fire extinguishing capability | 8 | 8 | 8 | 16 | 16 | 8 | 8 | 8 | 16 | 16 |
| Control of fire ability | 5 | 5 | 10 | 10 | 10 | 10 | 10 | 5 | 5 | 5 |
| Speed | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 6 |
| Unmanned plane number | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
| Intensity of a fire extinguishing capability | 8 | 8 | 8 | 16 | 16 | 8 | 8 | 8 | 16 | 16 |
| Control of fire ability | 5 | 5 | 10 | 10 | 10 | 10 | 10 | 5 | 5 | 5 |
| Speed | 12 | 12 | 12 | 12 | 12 | 12 | 12 | 12 | 12 | 12 |
Table 3
S2, subtask decomposing module:
Subtask Decomposition Genetic Algorithm basic flow chart is as shown in Figure 2.The module uses genetic algorithm, by task core point
Solution is several subtasks, each subtask is the set of several kindling point targets.In this step, method is needed to use
Person provides the basic parameter of genetic algorithm according to their needs.Design parameter size, routinely genetic algorithm convention selects.At this
In embodiment, population scale 30, crossover probability 0.97, mutation probability 0.01, maximum number of iterations 500 are taken.
1. chromosome indicates
The chromosome expression of genetic algorithm uses binary coded form, based on the scene information that step 1 determines, each
Chromosome is made of matrix F, wherein the number of known target is 10, resolves into 5 subtasks altogether.Each member in F matrix
The value of element is determined that the citing of F matrix coding situation is as shown in table 4 by following formula.
Table 4
After target distribution schemes matrix F has been determined, each unfriendly target under battlefield surroundings can be according to F
Matrix result is respectively allocated to the subtask of different stage.Then each chromosome of genetic algorithm is a F matrix.
2. fitness function
When carrying out mission planning, to be weighted to obtain its " comprehensive income index " according to the position and income degree of target
Weight:
Weight=location × α+gain × (1- α)
Wherein factor alpha weight is provided by method user, calculates the value side weight of each subtask in chromosome
Difference, then by each subtask Variance Addition, if sum of variance is smaller, then it is assumed that the fitness of the individual is higher.Therefore, fitness
The concrete form of function are as follows:
In formula, weight is the comprehensive income index of element (i.e. unfriendly target) in chromosome, and E is a son in chromosome
The expectation of task, fitness fitness are that the inverse of variance reaches the maximum of fitness when then genetic algorithm restrains, this
The outstanding decomposing scheme of sample ensures that each subtask by the target cluster of different comprehensive income indexs to together.
3. the selection operation of subtask decomposition algorithm
The method of the present invention uses wheel disc method, i.e., the selection of individual is carried out using fitness ratio selection mode, per each and every one
The select probability of body is directly proportional to its fitness.If population scale is N, i-th (i=1,2 ..., N) individual is suitable in population
Response is fi, then i-th individual select probability pi, shown in following formula:
4. the crossover operation of subtask decomposition algorithm
Due to the operation of subtask decomposition algorithm object and output the result is that target distribution schemes one by one, i.e., per each and every one
The number of " 1 " is equal to distributed number of targets in body, so if changing the structure of original individual and making in crossover process
The number of " 1 " becomes more or tails off in individual, will lead to the task boundary scheme that algorithm provides occur target it is idle or occur it is extra
The problems such as target, this apparently not it is expected the result seen.
To ensure that the number of " 1 " in every row individual after crossover operation immobilizes, propose that a kind of regulating measures calculate heredity
The basic crossover operation of method is improved.Specific step is as follows: vacation lets a and b be two male parents to be intersected, and have one it is laggard
The storehouse first gone out, As, Bs represent the numerical value on the position s of two male parent's chromosome, if s need to intersect, and As, Bs two
It is different, then this two are not first temporarily exchanged, but they are put into storehouse and is stored, then proceed to the subsequent of search A and B
Position, if discovery there is also Ai, Bi this to phase dystopy, and Ai and Bi be also it is different, then exchange two groups of As, Bs and Ai, Bi simultaneously
Gene position can ensure that two individuals of A and B still ensure that in individual that the number of " 1 " is fixed not after crossover operation in this way
Become.
Judging whether the concrete operations for needing to intersect are as follows: concrete operation method is to generate a random number x (0≤x≤1),
If x < crossover probability Pj, then crossover operation is carried out, if x > crossover probability Pj, then it is not processed.
5. the mutation operation of subtask decomposition algorithm
Equally, it is not intended to the mutation operation of genetic algorithm to destroy the stability of original scheme individual, also here to ensure to become
The number of " 1 " immobilizes in each individual after ETTHER-OR operation, must just use special variation method.Specific step is as follows: root
Determine whether a certain position s of a line genes of individuals sequence morphs according to given mutation probability, if it is, selecting in same a line
Take another random order i, it is assumed that this two be also it is different, just s and i are directly exchanged, otherwise do not exchanged, to reach change
Different effect.
Judging whether the concrete operations for needing to make a variation are as follows: concrete operation method is to generate a random number x (0≤x≤1),
If x < mutation probability Pm, then mutation operation is carried out, if x > mutation probability Pm, then it is not processed.
6. hereditary termination condition
Specified genetic algebra just stops the progress of genetic algorithm.
It, can be according to existing scene information (the mainly geographical location seat of known target by Revised genetic algorithum
Be marked with and income degree) task core is decomposed into the subtask of different stage.
In the present embodiment, subtask distribution condition is as follows:
Subtask 1: target 9, target 10.
Subtask 2: target 1, target 5.
Subtask 3: target 2, target 6.
Subtask 4: target 3, target 7.
Subtask 5: target 4, target 8.
S3, priority of subtask auto-sequencing module --- harmony population generate:
10 target ignition points are divided in step S2 and generate 5 subtasks, it will be to this 5 subtasks in this step
Carry out priority ranking.
It is modeled as follows first.(HS) algorithm is searched for using harmony, includes five singers in each harmony individual,
Five subtasks are represented, the melody sung represents priority of subtask size.Generate 10 harmony at random in solution space
Body represents 10 kinds of priority of subtask sortords.
Following parameter: the size HMS:10 of harmony data base need to be provided by method user.Harmony data base probability
HMCR:0.5.Tone finely tunes probability P AR:0.1.Tone fine tuning bandwidth BW: 0.1.The number T of creationmax: 100.
(2) harmony data base HMS is initialized
10 harmony are generated at random in solution space, with vector X1,X2,…,X10It indicates, each harmony represents a son
The sortord of task priority.It is put into harmony data base, and records corresponding f (X), harmony library are as follows:
S4, scheduling of resource module --- harmony evaluation
To each of current harmony population harmony individual, i.e. one group of priority of subtask sortord carries out resource
Scheduling.(MDLS) algorithm is planned using Dynamic and Multi dimensional list, and scheduling of resource is carried out with the established model of step S1.Scheduling of resource is complete
Bi Hou, as each individual corresponding optimal scheduling of harmony.Harmony evaluation is carried out to this optimal scheduling.
(1) priority of task
When all predecessor tasks (having to completing for task before task processing) of a certain task have all been handled
When complete, which just enters in eligible task collection READY locating for allowable resource task.Concentrate selection excellent in READY
First gradeMaximum subtask j carries out platform resource scheduling first, as waiting task.
(2) idle platform collection
Either with or without processing task can calling platform be put into idle platform collection FREE, the selection of platform just directly from
It is selected in FREE.
(3) platform group selection
Platform group selection is the platform group that selection executes waiting task, and platform group selection is Dynamic and Multi dimensional list planning
The key component of MDLS algorithm.Here the selection of platform group is to calculate priority P to platform one by one, i.e. platform is applicable in task
Degree, formula are as follows:
P=T+ ∑ R*T
In formula, T represents the time (normalized) that platform is moved to task objective ground, and ∑ R*T represents platform items and appoints
The sum of the processing capacity for required ability of being engaged in.
Then, according to the ability of the priority P sequence resource of arrangement superposition from big to small, check superposition ability whether be more than
The resource requirement of task stops the selection to platform once the demand for reaching task, then starts to trim platform group, picks
Except the platform of redundancy.
After so completing platform group selection, i.e., the task distribution platform is terminated.By required platform from idle platform
It is removed in collection FREE, and the task is removed from eligible task collection READY, which is put into a new set ALREADY, i.e.,
To have distributed task-set.
(4) time updates
It, will when the resource capability needed for waiting task is greater than the sum of all platform capabilities in idle platform collection FREE
Carry out the update of time.Time passage has task completion until having to have distributed in task-set ALREADY, then by the completion task
In next task merging eligible task collection READY, and the completion task institute is discharged using platform, merging FREE is concentrated, and is remembered
Record each platform time for being used, j-th of platform be denoted as tj.It is returned in step (1) priority of task later and continues algorithm,
Match until without task dividable.At this point, scheduling of resource complete, record at this time the time passage at the time of T.
(5) harmony is evaluated
Scheduling of resource module final step is after having carried out Dynamic and Multi dimensional list planning algorithm, to dispatch and carry out to Current resource
Harmony evaluation.
Firstly, when algorithm proceeds to here for the first time, for X at this timei, harmony evaluation is following to be calculated:
By step (4) the time t that is used of task total time T (i) at this time and each platformj(i), it calculates separately
Complete task time reference value TP and platform utilization rate reference value PP.
TP=T
In formula, T represents the first sub-distribution task total time, tjRepresent the time that each platform of the first sub-distribution is used, N
Represent platform sum.
Therefore using platform utilization rate reference value, completion task time and platform utilization rate are normalized, obtained
Time priority weight coefficient T when each algorithm carries outP(i) and PP(i)
In formula, T (i) represents i-th distribution task total time, tjIt represents i-th and distributes the time that each platform is used,
N represents platform sum.
For this method, it is desirable to obtain completing that task time is most short and the highest action scheme of platform utilization efficiency.
Open parameter beta is set, and the size of β is determined for completing task time tendentious height.β is bigger, represents method use
Person is higher for shortening completion task time requirement.
Comprehensive priority coefficient are as follows:
Note harmony is evaluated as the inverse of comprehensive priority coefficient, is f (Xi)=Pr (i)
Then by each f (Xi) return in HM, complete harmony library HM.
Calculate to obtain harmony evaluation are as follows:
S5, priority of subtask auto-sequencing module --- the memory of harmony library updates:
(1) a new harmony is generated
A random number r is generated between [0,1], is compared with harmony data base probability HMCR, if r <
HMCR takes out a harmony variable at random from harmony data base, otherwise, generates a harmony variable at random from solution space.By
Above content obtains a harmony variable, if this harmony variable is obtained in the harmony data base, it is necessary to this and
Sound variable is finely adjusted, and a random number s is generated between [0,1].If s < PAR (tone fine tuning probability), according to fine tuning bandwidth
BW obtains a new harmony variable to be adjusted to obtained harmony variable.Otherwise, it does not make any adjustments.It finally obtains
New harmony Xnew。
(2) harmony is evaluated
By XnewIt substitutes into and carries out harmony evaluation in step 4.
(3) harmony data base is updated
To XnewIt is assessed, i.e. f (Xnew), if one worst better than the functional value in the HM of harmony library, i.e. f (Xnew)<f
(Xworst), then by XnewInstead of the worst harmony X of functional value in the HM of harmony libraryworst；Otherwise, it does not make an amendment.
(4) step (1) to step (3) is repeated, until number of repetition (adjustment number) reaches Tmax。
Finally obtained harmony population is as follows:
Step S3-S5 is that priority of subtask auto-sequencing module is nested with scheduling of resource module, flow chart such as Fig. 3
It is shown.
S6, action scheme generate
It is 10 outstanding harmony individuals that step S5, which reaches obtained after convergence times at this time, i.e. the priority of subtask is arranged
Sequence and the corresponding optimum allocation of each individual.Allocation plan can be indicated with Gantt chart, such as Fig. 4.Scheduling scheme is with step S2's
Subtask decomposing scheme combines the output result of as this unmanned aerial vehicle group action scheme automatic generation method.
Claims (5)
1. a kind of unmanned aerial vehicle group action scheme automatic generation method, it is characterised in that: this method comprises the following steps:
Step 1: task core and scene information are determined；
Step 2: task core is decomposed into several subtasks using genetic algorithm by subtask decomposing module；
Step 3: priority of subtask auto-sequencing module --- harmony population generates；
Harmony population is generated using harmonic search algorithm: being generated using priority of subtask auto-sequencing harmonic search algorithm several
A harmony individual, each harmony individual represent a kind of priority of subtask sortord, each priority of subtask are sorted
Mode enters step in the Dynamic and Multi dimensional list planning algorithm of four scheduling of resource module and carries out scheduling of resource, obtained resource benefit
It is the harmony evaluation of harmony individual with rate and task completion time；
Step 4: scheduling of resource module --- harmony evaluation
It is individual for each harmony, i.e. row X in HMi, it is given preferential to realize to plan MDLS algorithm using Dynamic and Multi dimensional list
Scheduling of resource of the grade to task；
Step 5: priority of subtask auto-sequencing module --- harmony data base updates
Step 6: output excellent alternative；
It updates number of repetition in step 5 harmony data base to reach after Tmax in gained harmony library HM, detection harmony evaluation f (X) is most
Big harmony individual Xi, then i-th of priority of subtask is ordered as optimal sequencing；Xi in step 4 scheduling of resource module where it
Corresponding scheduling of resource is optimal scheduling of resource, that is, optimal action scheme.
2. a kind of unmanned aerial vehicle group action scheme automatic generation method according to claim 1, it is characterised in that: the step
Two detailed process is as follows:
(1) genetic algorithm chromosome indicates
The chromosome of genetic algorithm indicates to use binary coded form, based on the scene information that step 1 determines, each dyeing
Body is made of matrix F, and wherein n is the number of subtask, and m is the number of known target；In F matrix the value of each element by
Following formula determines:
After target distribution schemes matrix F has been determined, each unfriendly target under battlefield surroundings can be according to F matrix
As a result it is respectively allocated to the subtask of different stage；Then each chromosome of genetic algorithm is a F matrix；
(2) fitness function of subtask Decomposition Genetic Algorithm
First when the existing perception target of enemy is carried out mission planning, to be weighted according to the position and income degree of target
To its " comprehensive income index " weight:
Weight=location × α+gain × (1- α)
Wherein factor alpha weight is provided by method user, and meaning is method user for the excellent of positioning index and proceeds indicatior
First consider degree；The comprehensive income index that each target is matched with subtask is organized into matrix W；By matrix W and square
Battle array F dot product, the comprehensive income index matrix A after being distributed；
The variance of every a line in comprehensive income index matrix A is calculated, then by each row Variance Addition of comprehensive income index matrix A, if
Sum of variance is smaller, then it is assumed that individual fitness is higher；Therefore, the concrete form of fitness function are as follows:
In formula, weight (fik) be matrix F in element (i.e. target) comprehensive income index, E (weightn) refer to for comprehensive income
Mark the expectation of data line in matrix A；
(3) selection operation of subtask decomposition algorithm
Using wheel disc method, i.e., carried out using fitness ratio selection mode individual selection, the select probability of each individual and
Its fitness is directly proportional；If population scale is N, the fitness of i-th (i=1,2 ..., N) individual is f in populationi, then i-th
The select probability p of individuali, shown in following formula:
(4) crossover operation of subtask decomposition algorithm
To ensure that the number of " 1 " in every row individual after crossover operation immobilizes, propose a kind of regulating measures to genetic algorithm
Basic crossover operation is improved, the specific steps are as follows: vacation lets a and b be two male parents to be intersected, and has one last in, first out
Storehouse, As, Bs represent the numerical value on the position s of two male parent's chromosome, if As, Bs two different, temporarily first do not exchange
This two, but they are put into storehouse and is stored, then proceed to the subsequent bit of search A and B, if discovery there is also Ai,
Bi this to phase dystopy, and Ai and Bi be also it is different, then exchange two groups of As, Bs and Ai, Bi gene positions simultaneously, thus can be true
It protects two individuals of A and B and still ensures that in individual that the number of " 1 " immobilizes after crossover operation；
(5) mutation operation of subtask decomposition algorithm
Equally, it is not intended to the mutation operation of genetic algorithm to destroy the stability of original scheme individual, also here to ensure the behaviour that makes a variation
The number of " 1 " immobilizes in each individual after work, must just use special variation method；Specific step is as follows: according to giving
Fixed mutation probability determines whether a certain position s of a line genes of individuals sequence morphs, if it is, choosing in same a line another
An outer random order i, it is assumed that this two be also it is different, just s and i are directly exchanged, otherwise do not exchanged, to reach variation
Effect；
(6) hereditary termination condition
Specified genetic algebra just stops the progress of genetic algorithm.
3. a kind of unmanned aerial vehicle group action scheme automatic generation method according to claim 1, it is characterised in that: the step
Three detailed process is as follows:
(1) initial parameter need to initialize following parameter:
The size HMS of harmony data base: for the size of harmony population；
Harmony data base probability HMCR: the probability of a harmony is taken out from existing population, that is, HM harmony library；
Tone finely tunes probability P AR: the probability being finely adjusted to the harmony taken out；
Tone finely tunes bandwidth BW: the amplitude of fine tuning；
The number T of creationmax: number, that is, iteration number of adjustment；
(2) harmony data base HMS is initialized
Generate HMS harmony at random in solution space, wherein harmony is understood as individual, and HMS harmony is understood as population, uses vector
X1,X2,…,XHMSIt indicates, each harmony represents the sortord of a priority of subtask；It is put into harmony data base, and is remembered
Record corresponding f (X), the form in harmony library are as follows:
Wherein, n represents the number of subtask, i-th of harmony XiInRepresent the preferential of j-th subtask in harmony individual
Grade size；F (X) represents the harmony evaluation of the i.e. priority of subtask sequence of the harmony individual.
4. a kind of unmanned aerial vehicle group action scheme automatic generation method according to claim 1, it is characterised in that: the step
Four detailed process is as follows:
(1) priority of task
When a certain task all predecessor tasks be the task processing before have to complete task, all handled when
It waits, which just enters in eligible task collection READY locating for allowable resource task；Selection priority is concentrated in READY
Maximum subtask j carries out platform resource scheduling first, as waiting task；
(2) idle platform collection
Either with or without processing task can calling platform be put into idle platform collection FREE, the selection of platform is just directly from FREE
Middle selection；
(3) platform group selection
Platform group selection is the platform group that selection executes waiting task, specifically calculates priority P, i.e. platform to platform one by one
To the appropriate of task, formula is as follows:
P=T+ ∑ R*T
In formula, T represents the time that platform is moved to task objective ground, and ∑ R*T represents the place of the required ability of platform each task
The sum of reason ability；
Then, according to the ability of the priority P sequence resource of arrangement superposition from big to small, check whether the ability of superposition is more than task
Resource requirement, stop the selection to platform once the demand for reaching task, then start to trim platform group, reject it is superfluous
Remaining platform；
After so completing platform group selection, i.e., the task distribution platform is terminated；By required platform from idle platform collection
It is removed in FREE, and the task is removed from eligible task collection READY, which is put into a new set ALREADY, as
Task-set is distributed；
(4) time updates
When the resource capability needed for waiting task is greater than the sum of all platform capabilities in idle platform collection FREE, it will carry out
The update of time；Time passage has task completion until having to have distributed in task-set ALREADY, then by the next of the completion task
Task is placed in eligible task collection READY, and the completion task institute is discharged using platform, is placed in idle platform collection FREE,
And record each platform time for being used, j-th of platform be denoted as tj；It returns in step (1) priority of task and continues later
Algorithm is matched until without task dividable；At this point, scheduling of resource complete, record at this time the time passage at the time of T；
(5) harmony is evaluated
Firstly, for Xi, harmony evaluation is following to be calculated:
By step (4) the time t that is used of task total time T (i) at this time and each platformj(i), completion is calculated separately to appoint
Be engaged in temporal reference value TP and platform utilization rate reference value PP；
TP=T
In formula, T represents the first sub-distribution task total time, tjThe time that each platform of the first sub-distribution is used is represented, N is represented
Platform sum；
Using platform utilization rate reference value, completion task time and platform utilization rate are normalized, calculated every time
Time priority weight coefficient T when method carries outP(i) and PP(i)
In formula, T (i) represents i-th distribution task total time, tjIt represents i-th to distribute the time that each platform is used, N is represented
Platform sum；
Open parameter beta is set, and the size of β is determined for completing task time tendentious height；β is bigger, represents method
User is higher for shortening completion task time requirement；
Comprehensive priority coefficient are as follows:
Note harmony is evaluated as the inverse of comprehensive priority coefficient, is f (Xi)=Pr (i)
Then by each f (Xi) return in HM, complete harmony library HM.
5. a kind of unmanned aerial vehicle group action scheme automatic generation method according to claim 1, it is characterised in that: the step
Five detailed process is as follows:
(1) a new harmony is generated
Between [0,1] generate a random number r, be compared with harmony data base probability HMCR, if r < HMCR, from
A harmony variable is taken out in sound memory library at random, otherwise, generates a harmony variable at random from solution space；It is obtained by above content
To a harmony variable, if this harmony variable be from harmony library memory obtained in, it is necessary to this harmony variable carry out
Fine tuning generates a random number s between [0,1]；If s < PAR, according to fine tuning bandwidth BW, to be carried out to obtained harmony variable
Adjustment, obtains a new harmony variable；Otherwise, it does not make any adjustments；Finally obtain new harmony Xnew；
(2) harmony is evaluated
By XnewIt substitutes into and carries out harmony evaluation in step 4；
(3) harmony data base is updated
To XnewIt is assessed, i.e. f (Xnew), if one worst better than the functional value in the HM of harmony library, i.e. f (Xnew)<f
(Xworst), then by XnewInstead of the worst harmony X of functional value in the HM of harmony libraryworst；Otherwise, it does not make an amendment；
(4) step (1) to step (3) is repeated, until number of repetition reaches Tmax。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810988692.3A CN109242290B (en) | 2018-08-28 | 2018-08-28 | Automatic generation method for unmanned aerial vehicle group action scheme |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810988692.3A CN109242290B (en) | 2018-08-28 | 2018-08-28 | Automatic generation method for unmanned aerial vehicle group action scheme |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109242290A true CN109242290A (en) | 2019-01-18 |
| CN109242290B CN109242290B (en) | 2020-11-27 |
Family
ID=65068045
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810988692.3A Active CN109242290B (en) | 2018-08-28 | 2018-08-28 | Automatic generation method for unmanned aerial vehicle group action scheme |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109242290B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109947130A (en) * | 2019-04-03 | 2019-06-28 | 广东电网有限责任公司 | A kind of transmission of electricity multiple aircraft formation tour management method and system based on grid-based management |
| CN113050688A (en) * | 2021-03-22 | 2021-06-29 | 中国人民解放军国防科技大学 | Planning method for multi-unmanned aerial vehicle collaborative search path in key target sealing control |
| CN113553777A (en) * | 2021-09-18 | 2021-10-26 | 中国人民解放军国防科技大学 | Anti-unmanned aerial vehicle swarm air defense deployment method, device, equipment and medium |
| CN115239204A (en) * | 2022-09-19 | 2022-10-25 | 中国电子科技集团公司第十四研究所 | Collaborative task planning method for multi-platform unmanned aerial vehicle-mounted radio frequency system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001350799A (en) * | 2000-06-09 | 2001-12-21 | Sanyo Electric Co Ltd | Concurrent design supporting system |
| CN103279812A (en) * | 2013-06-19 | 2013-09-04 | 河海大学 | Task and resource allocation optimizing method capable of eliminating priority conflicts |
| CN103631758A (en) * | 2013-11-21 | 2014-03-12 | 陕西理工学院 | Method for solving non-linear programming and absolute value equation through improved harmony search algorithm |
| CN104698838A (en) * | 2014-12-23 | 2015-06-10 | 清华大学 | Discourse domain based dynamic division and learning fuzzy scheduling rule mining method |
| US20150170093A1 (en) * | 2013-12-17 | 2015-06-18 | Asd Inc. | Break Management in Business Center or the Like |
| CN106600147A (en) * | 2016-12-15 | 2017-04-26 | 合肥工业大学 | Resolvable task oriented task assigning method and apparatus for multiple unmanned aerial vehicles |
-
2018
- 2018-08-28 CN CN201810988692.3A patent/CN109242290B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001350799A (en) * | 2000-06-09 | 2001-12-21 | Sanyo Electric Co Ltd | Concurrent design supporting system |
| CN103279812A (en) * | 2013-06-19 | 2013-09-04 | 河海大学 | Task and resource allocation optimizing method capable of eliminating priority conflicts |
| CN103631758A (en) * | 2013-11-21 | 2014-03-12 | 陕西理工学院 | Method for solving non-linear programming and absolute value equation through improved harmony search algorithm |
| US20150170093A1 (en) * | 2013-12-17 | 2015-06-18 | Asd Inc. | Break Management in Business Center or the Like |
| CN104698838A (en) * | 2014-12-23 | 2015-06-10 | 清华大学 | Discourse domain based dynamic division and learning fuzzy scheduling rule mining method |
| CN106600147A (en) * | 2016-12-15 | 2017-04-26 | 合肥工业大学 | Resolvable task oriented task assigning method and apparatus for multiple unmanned aerial vehicles |
Non-Patent Citations (4)
| Title |
|---|
| GEEM Z W ETAL.: "Application of harmony search to vehicle routing", 《AMERICAN JOURNAL OF APPLIED SCIENCES》 * |
| 姜华 等: "混合变邻域和声搜索的独立任务调度问题研究", 《计算机工程与设计》 * |
| 阳东升 等: "战役任务计划的数学描述与求解算法研究", 《系统工程理论与实践》 * |
| 陈云霞 等: "基于遗传算法的多学科设计优化分解方法", 《北京航空航天大学学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109947130A (en) * | 2019-04-03 | 2019-06-28 | 广东电网有限责任公司 | A kind of transmission of electricity multiple aircraft formation tour management method and system based on grid-based management |
| CN113050688A (en) * | 2021-03-22 | 2021-06-29 | 中国人民解放军国防科技大学 | Planning method for multi-unmanned aerial vehicle collaborative search path in key target sealing control |
| CN113050688B (en) * | 2021-03-22 | 2022-03-29 | 中国人民解放军国防科技大学 | Planning method for multi-unmanned aerial vehicle collaborative search path in key target sealing control |
| CN113553777A (en) * | 2021-09-18 | 2021-10-26 | 中国人民解放军国防科技大学 | Anti-unmanned aerial vehicle swarm air defense deployment method, device, equipment and medium |
| CN113553777B (en) * | 2021-09-18 | 2021-11-30 | 中国人民解放军国防科技大学 | Anti-unmanned aerial vehicle swarm air defense deployment method, device, equipment and medium |
| CN115239204A (en) * | 2022-09-19 | 2022-10-25 | 中国电子科技集团公司第十四研究所 | Collaborative task planning method for multi-platform unmanned aerial vehicle-mounted radio frequency system |
| CN115239204B (en) * | 2022-09-19 | 2023-02-14 | 中国电子科技集团公司第十四研究所 | Collaborative task planning method for multi-platform unmanned aerial vehicle-mounted radio frequency system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109242290B (en) | 2020-11-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109242290A (en) | A kind of unmanned aerial vehicle group action scheme automatic generation method | |
| CN114545975B (en) | Multi-unmanned aerial vehicle system task allocation method integrating multi-target evolution algorithm and contract network algorithm | |
| CN108846472A (en) | A kind of optimization method of Adaptive Genetic Particle Swarm Mixed Algorithm | |
| CN110956266A (en) | Multi-power-supply power system multi-target optimization scheduling method based on analytic hierarchy process | |
| CN109409773B (en) | Dynamic planning method for earth observation resources based on contract network mechanism | |
| CN104735166B (en) | The Skyline method for service selection annealed based on MapReduce and multi-target simulation | |
| CN107016461A (en) | One kind mixing multi-target evolution method | |
| CN104572297A (en) | Hadoop job scheduling method based on genetic algorithm | |
| CN113033970B (en) | AUV cluster layered distribution method for large-scale tasks | |
| CN111160511A (en) | Group intelligent method for consensus active learning | |
| CN107146039A (en) | The customized type mixed-model assembly production method and device of a kind of multiple target Collaborative Control | |
| CN105045095A (en) | Multi-unmanned-plane emergency rescue task distribution method | |
| CN114565239A (en) | Comprehensive low-carbon energy scheduling method and system for industrial park | |
| CN114841055B (en) | Unmanned aerial vehicle cluster task pre-allocation method based on generation countermeasure network | |
| Ahmad et al. | Task assignment using a problem‐space genetic algorithm | |
| CN109840551A (en) | A method of the optimization random forest parameter for machine learning model training | |
| Liang et al. | Improved nsga2 algorithm to solve multi-objective flexible job shop scheduling problem | |
| CN114580298A (en) | RBF-ISBO-based microgrid optimization scheduling method and system | |
| RU80604U1 (en) | AUTOMATED RESOURCE DISTRIBUTION SYSTEM FOR OPTIMUM SOLUTION OF TARGET TASKS | |
| Soto et al. | Top-Based Adaptive Enumeration in Constraint Programming | |
| CN110059806A (en) | A kind of multistage weighted network community structure detection method based on power-law function | |
| Li | Improved NSGA-II for the job-shop multi-objective scheduling problem | |
| CN113947321A (en) | Distributed information energy network topological structure construction method and system | |
| Ling et al. | An MOEA/D-ACO with PBI for many-objective optimization | |
| Taowei et al. | A Membrane-Fireworks Algorithm for Multi-Objective Optimization Problems |
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 |