CN115202400A - Unmanned aerial vehicle cluster task planning method based on self-adaptive penalty TAEA - Google Patents
Unmanned aerial vehicle cluster task planning method based on self-adaptive penalty TAEA Download PDFInfo
- Publication number
- CN115202400A CN115202400A CN202210930554.6A CN202210930554A CN115202400A CN 115202400 A CN115202400 A CN 115202400A CN 202210930554 A CN202210930554 A CN 202210930554A CN 115202400 A CN115202400 A CN 115202400A
- Authority
- CN
- China
- Prior art keywords
- unmanned aerial
- population
- aerial vehicle
- search
- task
- 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
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
- G05D1/10—Simultaneous control of position or course in three dimensions
- G05D1/101—Simultaneous control of position or course in three dimensions specially adapted for aircraft
- G05D1/104—Simultaneous control of position or course in three dimensions specially adapted for aircraft involving a plurality of aircrafts, e.g. formation flying
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Abstract
The invention provides an unmanned aerial vehicle cluster task planning method based on self-adaptive penalty TAEA, which comprises the following steps: constructing an unmanned aerial vehicle cluster collaborative search tracking task planning scene; establishing a multi-target multi-constraint optimization problem model for unmanned aerial vehicle searching formation searching; acquiring an optimal search task planning scheme; establishing a multi-target multi-constraint optimization problem model for searching and tracking by unmanned aerial vehicle searching formation; initializing self-adaptive punished double-file evolution algorithm TAEA parameters; updating the convergence population and the diversity population; and acquiring a planning result of the unmanned aerial vehicle cluster collaborative search tracking task. According to the method, the problem of unmanned aerial vehicle cluster searching and tracking task planning is solved through a self-adaptive punished double-file evolution algorithm, the task planning efficiency is improved, a new task planning problem model is established for optimization after the unmanned aerial vehicle searching formation finds a target, a global optimal task planning scheme is obtained, and the unmanned aerial vehicle cluster effectiveness is favorably exerted.
Description
Technical Field
The invention belongs to the technical field of unmanned aerial vehicle clusters, relates to an unmanned aerial vehicle cluster task planning method, and particularly relates to an unmanned aerial vehicle cluster task planning method based on a self-adaptive punished double-file evolution algorithm TAEA (probabilistic algorithm for area optimization) in a dynamic scene environment, which can be used in the fields of city security, environment monitoring, field search and rescue, terrorist prevention monitoring and the like.
Background
The unmanned aerial vehicle cluster task planning refers to allocating appropriate unmanned aerial vehicle formation to dynamic tasks and determining the sequence of the unmanned aerial vehicle formation to execute the tasks. According to the organization form of the unmanned aerial vehicle cluster, the unmanned aerial vehicle task allocation mode mainly comprises a centralized mode and a distributed mode. The centralized type is a system structure that all unmanned aerial vehicles in the formation communicate through a single control center to realize signal transmission and control. All unmanned aerial vehicles in the formation transmit the acquired external information and the state information of the unmanned aerial vehicles to the central processing center, the central processing center processes and makes decisions on the information, a control instruction is formed, and then the control instruction is sent to the unmanned aerial vehicles in the cluster, so that the task allocation planning of the unmanned aerial vehicle cluster is realized. Distributed refers to a control system structure in which all drones in a formation can communicate with each other. Compared with a centralized unmanned aerial vehicle cluster system, the distributed unmanned aerial vehicle cluster system has the following advantages: each unmanned aerial vehicle is a computing node, each unmanned aerial vehicle can make a decision autonomously, risks caused by damage of a central node can be reduced, and the unmanned aerial vehicle has the capability of quickly coping with scene changes; however, when the number of communication nodes of the distributed unmanned aerial vehicle system is too large, the communication information amount is extremely large, great challenges are brought to the communication capacity of the system, only the local optimal solution can be obtained, but the global optimal solution cannot be obtained, and the efficiency of the unmanned aerial vehicle for executing tasks is reduced.
The unmanned aerial vehicle cluster task planning method can be divided into an unmanned aerial vehicle cluster task planning method under a dynamic scene environment and an unmanned aerial vehicle cluster task planning method under a static scene environment according to the working scene of the unmanned aerial vehicle cluster, wherein the basic principle of the unmanned aerial vehicle cluster task planning method under the dynamic scene environment is that unmanned aerial vehicle cluster task allocation planning is carried out again after tasks are changed dynamically, so that the unmanned aerial vehicle cluster can finish the tasks efficiently, and the key point is to improve the efficiency of solving a global optimal task allocation planning scheme after new tasks are found. For example, application publication No. CN111199360A, entitled "unmanned aerial vehicle task allocation planning method", which, when a new task P needs to be executed during the task execution process of an unmanned aerial vehicle, divides the task allocation into two stages, namely, pre-allocation and negotiation, i.e., in the pre-allocation stage, a satisfaction set of executable tasks is selected by a central node according to an individual satisfaction function and an individual rejection function, and determines whether the set is empty: if the set is not empty, selecting the most suitable unmanned aerial vehicle to execute a new task by the set of unmanned aerial vehicles meeting the conditions through mutual negotiation, and if the set is empty, executing the same task by combining a plurality of unmanned aerial vehicles. The method mainly solves the problem that the computing time of the central node is too long when task allocation is carried out in the prior art, ensures real-time dynamic allocation of tasks and load balance of unmanned aerial vehicle tasks, and improves the overall performance of unmanned aerial vehicle task execution, but the method has the defects that: 1. after a new task appears, task allocation planning is carried out, a new task planning scheme is obtained through a mutual negotiation mechanism of the unmanned aerial vehicle, the time consumption of the negotiation process is large, and the efficiency of solving the task planning scheme is low; 2. according to the method, only one locally optimal task planning scheme can be obtained through a mutual negotiation mechanism of the unmanned aerial vehicles, the unmanned aerial vehicle cluster can not obtain the maximum benefit at the minimum cost when executing tasks according to the task planning scheme, and therefore the unmanned aerial vehicle cluster cannot exert the maximum effectiveness.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides an unmanned aerial vehicle cluster task planning method based on self-adaptive penalty TAEA (target area algorithm), which is used for solving the technical problem of low planning efficiency in the prior art.
In order to achieve the purpose, the technical scheme adopted by the invention comprises the following steps:
(1) Constructing an unmanned aerial vehicle cluster collaborative search tracking task planning scene:
constructing an unmanned aerial vehicle cluster collaborative search tracking task planning scene, wherein the scene comprises M square grid task areas T = { T = distributed on the ground 1 ,T 2 ,...,T m ,...,T M K area targets T' = { T } distributed within grid task area T 1 ′,T′ 2 ,...,T′ k ,...,T′ K Searching N unmanned aerial vehicles distributed in the same horizontal plane in spaceFormation F = { F = { (F) 1 ,F 2 ,...,F n ,...,F N And central processing unit Θ, each grid task area T m The central point coordinate, the threat factor and the threat level parameter of (x) are respectively m ,y m )、w m 、λ m Per target T' k The threat factor and the threat level parameter of (1) are w' k 、λ′ k Each drone search formation F n Including the flight speeds ofThe starting time of all unmanned aerial vehicles for searching and forming to execute the searching task is t 0 Maximum flight time of T max In each task area T m The time of internal search is phi, the time interval of mission planning is tau, and each unmanned aerial vehicle searches for formation F n The number of the task areas covered by executing one search task is L, wherein M is more than or equal to 100 m Denotes the m-th task area, K is more than or equal to 1,T' k Represents the kth regional target, N is more than or equal to 2 n Representing the nth unmanned aerial vehicle searching formation, wherein L is more than or equal to 2;
(2) Establishing a multi-target multi-constraint optimization problem model for unmanned aerial vehicle search formation search:
(2a) Formation of unmanned aerial vehicle search F n And a grid task area T m As the allocation decision variable assign [ m, n ]]Searching and grouping unmanned aerial vehicles into an execution grid task area set Gamma n The sequence of (A) is used as a sequence decision variable order [ n ]]And passes through assign [ m, n ]]Value of (d) determines each drone search formation F n Assigned set of grid task regions tau n Then through order [ n ]]Gamma n All the distributed grid task areas are sequenced to obtain a sequenced grid task area set T nL ;
(2b) Searching formation F by each drone n Flying speed ofAnd a set T of ordered grid task areas nL Calculating unmanned aerial vehicle searchCable formation F n Completion task region T nl Time of day of
(2c) Sum S of threat values established to minimize τ over time for all grid task areas t Integral f of 1 And minimizing the maximum time f for the n unmanned aerial vehicle search formations to complete the L grid task areas respectively distributed 2 For the objective function, the drones search for the mutual distance d between the formation at the sampling time points ti (F a ,F b ) Each drone search formation F n Maximum time of flight constraint c 2 Each grid task area T m Threat value constraint c of 3 Multi-objective multi-constraint optimization problem model Ψ for task planning for searching grid task region T for constraint condition F 1 Wherein:
min f 1
min f 2
c 3 :s(m,t)≤S m,max
where s (m, T) denotes each grid task area T at time T m E represents the natural logarithm base number, t is t 0 Is a time variable at the start time, t' is a time offset, d min To indicate nobodyMinimum safe distance between machine search formations, S m,max Representing a grid task area T m The threat value upper limit value of (2);
(3) Obtaining an optimal search task planning scheme:
by assigning a decision variable assign [ m, n ]]And order decision variable order [ n ]]The decimal system randomly generating includesA convergence population CA and a diversity population DA of an initial individual x, and a constraint-oriented double-file evolution algorithm is adopted to solve a multi-target multi-constraint optimization problem model psi by updating the CA and the DA 1 To obtain a convergence population CA 0 And diversity population DA 0 Then from CA 0 Optionally selecting one individual gamma in the front face of Paretor 1 As an optimal search task planning scheme;
(4) Establishing a multi-target multi-constraint optimization problem model for searching and tracking of unmanned aerial vehicle searching formation:
(4a) Central processing unit theta selects unmanned aerial vehicle search formation F μ Flight direction grid task area T ν Search formation F with unmanned aerial vehicle η Form unmanned aerial vehicle tracking formation F μ ″ η To F, for n Using a search mission planning scheme gamma 1 For grid task region T ν Searching for found target T' k Set up trace, remainder divide by F μ And F η N-2 unmanned aerial vehicle search formations F 'based on a decision variable assign [ m, N']And order [ n']Determining a mesh task area search order allocated by unmanned aerial vehicle search formation, F ' = { F ', F ' 2 ,...,F n ″,...,F′ N-2 Where μ = sent [ n ″ ]],sent[n″]To assign a decision variable, N ∈ { 1., N } \ η;
(4b) Establishing to minimize the sum S 'of threat values for all grid task areas over time τ' t Integral f 'of' 1 Minimizing the maximum time f for N-2 unmanned aerial vehicles to search and form to complete L grid task areas distributed to the unmanned aerial vehicles 2 ' and minimize drone tracking formation F μ Reach grid taskRegion T ν Time f of 3 ' As an objective function, the drones search for the mutual distance between the formation at the sampling time pointsSearching mutual distance of formation at sampling time point by unmanned aerial vehicleSafety constraint of c 1 ', maximum time of flight constraint for each drone search formation c 2 ', each grid task area T m Threat value constraint c 3 ', target T ' has been found ' k Threat value constraint c 4 'unmanned aerial vehicle search formation F' as constraint condition and unmanned aerial vehicle tracking formation carry out searching and tracking in grid task area T by multi-target multi-constraint optimization problem model psi 2 Wherein:
min f 1 ′
min f 2 ′
min f 3 ′
c′ 3 :s(m,t)≤S m,max
c′ 4 :s(k,t)≤S k,max
wherein s (k, T) represents that target T 'is found at time T' k T' is a time offset, S k,max Denotes that target T has been found' k The upper limit value of the threat degree;
(5) Initializing parameters of a self-adaptive punished double-file evolution algorithm TAEA:
initializing the convergence population CA obtained in the step (3) 0 For the initial generation of the convergent population CA, randomly generated comprisesTaking the initial individual x as an initial generation population of the diversity population DA, wherein the iteration frequency is R, the maximum iteration frequency is R, and R =0;
(6) Updating the convergence population and the diversity population:
(6a) The r generation convergence population CA r And the r-th generation diversity population DA r Combining to obtain population Hm r Separately calculating the population Hm r In CA r And DA r Non-dominant individual population Hm of r Pc and Pd, and Pc is judged>If Pd is true, if so, from CA r In the random selectionIndividual as a population P 1r Otherwise, from DA r In the random selectionIndividual as population P 1r (ii) a Computing population CA r The ratio of PC to non-dominant individual and producing a composition comprisingA vector pf of random numbers, the f-th value pf (f) of the vector pf is judged>Whether PC is true, if so, from CA r In the method, one individual is randomly selected as P 2r Of the f individual, otherwise from the DA r In the method, one individual is randomly selected as P 2r The f th individual of (1), P 1r And P 2r Form aGroup parent individual P r To P r Each group of parent individuals in the group are subjected to uniform cross operation to obtainGroup P ru To P ru Each group of individuals in the group is subjected to partial matching and cross operation to obtainGroup P rp And from P rp Randomly selecting one individual from each group of individuals to form P rhalf Then to P rhalf Is uniformly mutated to obtain P rum To P rum Performing insertion variation operation on each individual to obtain a filial generation population Q r ;
(6b) The r generation convergence population CA r And the offspring population Q r Composition of a population HC r The r-th generation convergence population DA r And progeny population Q r Make up population HD r Species HC r Feasible individuals and infeasible individuals in the group respectively form a group Fs r And a population Is r ;
(6c) Respectively through HC r 、HD r For convergence population CA r Diverse group DA r Updating is carried out;
(7) Acquiring a planning result of a collaborative search tracking task of the unmanned aerial vehicle cluster:
judging whether R = R is true, if yes, selecting the R-th generation convergence population CA according to a weight method r In the pareto frontier of (1) optionally selecting one individual gamma 2 The unmanned aerial vehicle search formation determined by the value of the decision variable is used for executing the grid task area allocated by the grid task area, the sequence of the grid task area allocated by the search and the designated search unmanned aerial vehicle formation are used as a task planning scheme for unmanned aerial vehicle cluster search tracking, otherwise, r = r +1 is carried out, and the step (6) is carried out;
compared with the prior art, the invention has the following advantages:
1. when the problem of unmanned aerial vehicle cluster search and tracking task planning is solved, the global optimal task planning scheme is obtained by using valuable and infeasible individuals to accelerate population convergence through a self-adaptive punishment-based double-file evolution algorithm, so that the defect that the negotiation process in the prior art consumes a lot of time is overcome, and the task planning efficiency is improved.
2. According to the invention, a new task planning problem model is established after the unmanned aerial vehicles search for the formation to find the target, and a global optimal task planning scheme is obtained by optimizing the task planning problem model, so that the unmanned aerial vehicle cluster effectiveness is favorably exerted.
Drawings
FIG. 1 is a flow chart of an implementation of the present invention.
Fig. 2 is a schematic diagram of an effective search range of the unmanned aerial vehicle search formation of the present invention.
FIG. 3 is a schematic diagram of the square grid task areas and the threat values of each grid task area at time t according to the present invention.
Detailed Description
The invention is described in further detail below with reference to the figures and specific examples.
Referring to fig. 1, the present invention includes the steps of:
step 1) constructing an unmanned aerial vehicle cluster collaborative search tracking task planning scene:
constructing an unmanned aerial vehicle cluster collaborative search tracking task planning scene, wherein the scene comprises M square grid task areas T = { T = distributed on the ground 1 ,T 2 ,...,T m ,...,T M K area targets T' = { T } distributed within the grid task area T 1 ′,T′ 2 ,...,T′ k ,...,T′ K N unmanned aerial vehicle search formations F = { F) distributed in the same horizontal plane of space 1 ,F 2 ,...,F n ,...,F N And central processing unit Θ, each grid task area T m The central point coordinate, the threat factor and the threat level parameter are respectively (x) m ,y m )、w m 、λ m Per target T' k The threat factor and the threat level parameter of (2) are w' k 、λ′ k Each drone search formation F n Involving flyingAll speeds areThe starting time of all unmanned aerial vehicles for searching and forming to execute the searching task is t 0 Maximum flight time of T max In each task area T m The time of the internal search is phi, the time interval of the task planning is tau, and each unmanned aerial vehicle searches for a formation F n The number of the task areas covered by executing one search task is L, wherein M is more than or equal to 100 m Denotes the m-th task region, K.gtoreq.1,T' k Denotes the kth regional target, N ≧ 2 n Represents the nth drone search formation, L ≧ 2, in this example, M =400, K =1, N =5,w m is [ 1X 10 ] -5 ,1×10 -4 ]Random number in the range, λ m Is [ 1X 10 ] -11 ,1×10 -5 ]Random number in the range, w' k =1×10 -1 ,λ′ k =1×10 -5 ,t 0 =0,T max =1×10 4 Second, φ =10 seconds, τ =200 seconds, L =7, each grid task area T m The inscribed regular quadrangle is represented as an effective search circular range for unmanned aerial vehicle search formation, the side length of each grid area is 100 meters, as shown in fig. 2, a dotted circle represents an effective search range for unmanned aerial vehicle search formation, and for simplifying a scene, the effective search range for unmanned aerial vehicle search formation is represented as a circular inscribed gray square.
Step 2), establishing a multi-target multi-constraint optimization problem model for unmanned aerial vehicle search formation search:
step 2 a) search formation F of unmanned aerial vehicles n And a grid task area T m As the allocation decision variable assign [ m, n ]]Searching and grouping unmanned aerial vehicles into an execution grid task area set Gamma n The sequence of (A) is used as a sequence decision variable order [ n ]]And passes through assign [ m, n ]]Value of (d) determines each drone search formation F n Assigned set of grid task areas T n Then through order [ n ]]Gamma n All the distributed grid task areas are sequenced to obtain a sequenced grid task area set T nL ;
Step 2 b) search formation F by each drone n Flying speed ofAnd a set T of ordered grid task areas nL And calculating unmanned aerial vehicle search formation F n Completion task region T nl Time of day ofThe calculation formula is as follows:
wherein (x) nl ,y nl ) And (x) n(l+1) ,y n(l+1) ) Are each T nL Two grid task areas adjacent to each other in the execution sequence;
step 2 c) set up to minimize the sum S of threat values of τ over time for all grid task areas t Integral f of 1 And minimizing the maximum time f for the n unmanned aerial vehicle search formations to complete the L grid task areas respectively allocated 2 For the objective function, the drones search for the mutual distance d between the formation at the sampling time points ti (F a ,F b ) Each drone search formation F n Maximum time of flight constraint c 2 Each grid task area T m Threat value constraint c 3 Multi-objective multi-constraint optimization problem model Ψ for mission planning for searching constraint conditions F in grid mission region T 1 Wherein:
min f 1
min f 2
c 3 :s(m,t)≤S m,max
where s (m, T) denotes each grid task area T at time T m E represents the natural logarithm baseA number, t being t 0 Is a time variable at the start time, t' is a time offset, d min Representing the minimum safe distance between unmanned aerial vehicle search formations, S m,max Representing a grid task area T m Upper limit value of the threat value of t i For sampling time series, f s In order to be able to do so at a time sampling rate,is t i Unmanned aerial vehicle search formation F at moment n The location of the location; in this example d min =20 m, S m,max =1×10 5 ,f s =1, each side of the grid task area T contains 20 square grid task areas, as shown in fig. 3, where α =20, the darker the color of the grid task area represents the greater the threat value of the grid area.
Step 3) obtaining an optimal search task planning scheme:
by assigning a decision variable assign [ m, n ]]And order decision variable order [ n ]]The decimal encoding mode randomly generates a decimal code containingA convergence population CA and a diversity population DA of an initial individual x, and a constraint-oriented double-file evolution algorithm is adopted to solve a multi-target multi-constraint optimization problem model psi by updating the CA and the DA 1 To obtain a convergence population CA 0 And diversity population DA 0 Then from CA 0 The pareto frontier of the system can select an individual gamma 1 As an optimal search task planning scheme.
Step 4), establishing a multi-target multi-constraint optimization problem model for searching and tracking of unmanned aerial vehicle searching formation:
step 4 a) selecting unmanned aerial vehicle search formation F by the central processing unit theta μ Flight direction grid task area T ν Search formation F with unmanned aerial vehicle η Form unmanned aerial vehicle tracking formation F μη To F, for n Using a search mission planning scheme gamma 1 For grid task region T ν Searching for found target T' k Set up trace, remainder divide by F μ And F η N-2 unmanned aerial vehicle search formations F 'based on a decision variable assign [ m, N']And order [ n']Determining a mesh task area search order allocated by unmanned aerial vehicle search formation, F '= { F' 1 ,F′ 2 ,...,F′ n′ ,...,F′ N-2 Wherein μ = send [ n ″ ")],sent[n″]To assign a decision variable, N ∈ { 1., N } \ η;
step 4 b) set up to minimize the sum S 'of threat values for all grid task areas over time τ' t Integral f' 1 Minimizing the maximum time f for N-2 unmanned aerial vehicles to search and form to complete L grid task areas distributed to the unmanned aerial vehicles 2 ' and minimize drone tracking formation F μ Reach grid task area T ν Time f of 3 ' As an objective function, the drones search for the mutual distance between the formation at the sampling time pointsSearching mutual distance of formation at sampling time point by unmanned aerial vehicleSafety constraint of c 1 ', maximum time of flight constraint for each drone search formation c 2 ', each grid task area T m Threat value constraint c 3 ', found target T' k Threat value constraint c of 4 'unmanned aerial vehicle search formation F' as constraint condition and unmanned aerial vehicle tracking formation carry out searching and tracking in grid task area T multi-target multi-constraint optimization problem model psi of task planning 2 Wherein:
min f 1 ′
min f 2 ′
min f 3 ′
c′ 3 :s(m,t)≤S m,max
c′ 4 :s(k,t)≤S k,max
wherein s (k, T) represents that the target T 'has been found at time T' k T' is a time offset, S k,max Denotes found target T' k Threat degree upper limit value, S' t Representing all grid task areas and found target T 'at moment T' k The sum of the threat values of (a) is,representing unmanned aerial vehicle search formation F μ Reach grid task area T ν The time of (d); in this example S k,max =1×10 7 。
Step 5) initializing parameters of a self-adaptive punished double-file evolution algorithm TAEA:
initializing the convergence population CA obtained in the step (3) 0 For the initial generation of the convergent population CA, randomly generated comprisesTaking the initial individual x as the initial generation population of the diversity population DA, the iteration number is r, and the maximum iteration isThe times are R, and R =0; in the present example, the first and second substrates were,R=300。
step 6) updating the convergence population and the diversity population:
step 6 a) the r-th generation of the convergent population CA r And the r-th generation diversity population DA r Combining to obtain population Hm r Separately calculating the population Hm r In the genus of CA r And DA r Non-dominant individual population Hm of r Pc and Pd, pc is judged>If Pd is true, if yes, from CA r In the random selectionIndividual as P 1r Else from DA r In the random selectionIndividual as P 1r (ii) a Computing population CA r The ratio of PC to non-dominant individual and producing a composition comprisingA vector pf of random numbers, determining the f-th value pf (f) of the vector pf>If PC is true, if so, from CA r In the method, one individual is randomly selected as P 2r Of the f individual, otherwise from the DA r In the method, one individual is randomly selected as P 2r Of the f individual, P 1r And P 2r Form aGroup parent individual P r To P is to P r Each group of parent individuals in the group are subjected to uniform cross operation to obtainGroup P ru To P is to P ru Each group of individuals in the group is subjected to partial matching and crossing operation to obtainGroup P rp And from P rp Randomly selecting one individual from each group of individuals to form P rhalf Then to P rhalf Is uniformly mutated to obtain P rum To P rum Performing insertion mutation operation on each individual to obtain a filial generation population Q r ;
Step 6 b) the r-th generation of the convergent population CA r And the offspring population Q r Composition of a population HC r The r-th generation convergence population DA r And the offspring population Q r Make up a population HD r Population HC r Feasible individuals and infeasible individuals in the group respectively form a group Fs r And a population Is r ;
Step 6 c) by separately passing the population HC r And population Is r For convergent population CA r Update is performed through HD r For diversity population DA r Updating, wherein the updating method comprises the following steps: when in useWhen a time of CA r =X r When is coming into contact withWhen a time of CA r =S r ;Wherein, X r Is a population Is r N- | Fs selected according to a two-dimensional optimization problem formed by a constraint violation value of each individual and a modified Chebyshev decomposition value based on the constraint r Set of | optimal individuals, by X hr Hexix- hr+1 Composition is carried out; s lr Showing population HC solved by self-adaptive punishment based double-file evolution algorithm TAEA r The average number of the most optimal individuals,population HD obtained by dual-file evolution algorithm TAEA (probabilistic algorithm) for expressing self-adaptive punishment r Well-optimized individuals x b Are respectively represented as:
X r ={X hr ,χ hr+1 }
x b =argmin{g tch (x|w i ,z * )}
v 3 =max(s(m,t)-S m,max ,0)
v 4 =max(s(k,t)-S k,max ,0)
X hr Representing vs. population Is r The individuals in the population are subjected to rapid non-dominated sorting according to a dual-objective optimization problem gamma (x), and the population Is r The individuals in (1) are divided into different grades, a set formed by the individuals of the first hr grades is taken, and hr is satisfiedMaximum integer value of χ h+1 Indicating normalized constraint violation in individuals taking the hr-th levelIs smallerSet of individual entities, S lr Representing a population Fs r According to the correction objective function valuePerforming rapid non-dominated sorting to select the species Fs r The individuals in (1) are divided into different grades, the first lr grades of the individuals are taken to form a set, and lr is satisfiedThe minimum value of (a) is determined,penalty based modified chebyshev decomposition method g in individuals representing the lr th level mtch-penalty (x|w i ,z * ) Selected ofWorst individual x w Set of compositions, X CAr Denotes CA r A collection of partial individuals as a reference set,method g for representing penalty-based modified Chebyshev decomposition mtch-penalty (x|w i ,z * ) From HD r Selected optimal individual x b R (x) represents a constructed dual-objective optimization problem including constraint violation values and modified chebyshev decomposition values based on adaptive penalties,express the integrated individual x normalized objective function value And an adaptive penalty value p 1 、p 2 、p 3 The obtained correction objective function value;respectively representing the objective function f 1 、f 2 、f 3 Normalized value of (p) 1 、p 2 、p 3 Respectively representing an objective function f 1 、f 2 、f 3 An adaptive penalty value of (a) is set,to representThe weighted average of (a) of (b),representing violation value v 1 ,v 2 ,v 3 ,v 4 Normalized value of r f Representing the ratio of feasible individuals in the population;
step 7) acquiring a planning result of the unmanned aerial vehicle cluster collaborative search tracking task:
judging whether R = R is true, if so, selecting the R-th generation of convergence population CA according to the weight method r Optionally selecting one individual gamma from front surface of middle Paretor 2 And (3) determining the execution grid task area allocated by the unmanned aerial vehicle search formation, the sequence of the execution grid task area allocated by the execution search formation and the designated search unmanned aerial vehicle formation as a task planning scheme for unmanned aerial vehicle cluster search tracking, and if not, making r = r +1 and executing the step (6).
Claims (6)
1. An unmanned aerial vehicle cluster task planning method based on self-adaptive penalty TAEA (task adaptive algorithm) in a dynamic scene environment is characterized by comprising the following steps of:
(1) Constructing an unmanned aerial vehicle cluster collaborative search tracking task planning scene:
constructing an unmanned aerial vehicle cluster collaborative search tracking task planning scene, wherein the scene comprises M square grid task areas T = { T = distributed on the ground 1 ,T 2 ,...,T m ,...,T M K area targets T' = { T } distributed within the grid task area T 1 ′,T 2 ′,...,T k ′,...,T K ' }, N unmanned aerial vehicle search formations F = { F = distributed in the same horizontal plane of space 1 ,F 2 ,...,F n ,...,F N } and a central processor Θ, each grid task area T m The central point coordinate, the threat factor and the threat level parameter are respectively (x) m ,y m )、w m 、λ m Per target T' k The threat factor and the threat level parameter of (2) are w' k 、λ′ k Each drone search formation F n Including the flight speeds ofThe starting time of all unmanned aerial vehicles for searching and forming to execute the searching task is t 0 Maximum flight time of T max In each task area T m The time of internal search is phi, the time interval of mission planning is tau, and each unmanned aerial vehicle searches for formation F n The number of the task areas covered by the execution of one search task is L, wherein M is more than or equal to 100 m Denotes the m-th task region, K ≧ 1 k ' denotes the kth regional target, N.gtoreq.2 n Representing the nth unmanned aerial vehicle searching formation, wherein L is more than or equal to 2;
(2) Establishing a multi-target multi-constraint optimization problem model for unmanned aerial vehicle search formation search task allocation planning:
(2a) Formation of unmanned aerial vehicle search F n And a grid task area T m As the allocation decision variable assign [ m, n ]]And passes through assign [ m, n ]]Value of (d) determines each drone search formation F n Assigned set of grid task areas T n Search and formation of unmanned aerial vehicle into task area set Gamma n The sequence of (A) is used as a sequence decision variable order [ n ]]Then through order [ n ]]Gamma n All the distributed grid task areas are sequenced to obtain a sequenced grid task area set T nL ;
(2b) Searching formation F by each drone n Flying speed ofAnd a set T of ordered grid task areas nL And calculating unmanned aerial vehicle search formation F n Completion task region T nl Time of day of
(2c) Set up to minimize the sum S of threat values of all grid task areas over time t Integral f of 1 And minimizing the maximum time f for the N unmanned aerial vehicles to search and form a formation to complete the L grid task areas distributed to the N unmanned aerial vehicles 2 For the objective function, the drones search the mutual distance between the formation at the sampling time pointsEach drone search formation F n Maximum time of flight constraint c 2 Each grid task area T m Threat value constraint c 3 Task planning multi-objective multi-constraint optimization problem model psi for searching grid task region T for F with constraint conditions 1 Wherein:
min f 1
min f 2
c 3 :s(m,t)≤S m,max
where s (m, T) denotes each grid task area T at time T m E represents a natural logarithm base number, and t is t 0 Is a time variable at the start time, t' is a time offset, d min Representing the minimum safe distance between unmanned aerial vehicle search formations, S m,max Representing a grid task area T m The upper limit value of the threat value of (1);
(3) Obtaining an optimal search task allocation planning scheme:
by assigning a decision variable assign m, n]And order decision variable order [ n ]]The decimal encoding mode randomly generates a decimal code containingA convergence population CA and a diversity population DA of an initial individual x, and a constraint-oriented double-file evolution algorithm is adopted to solve a multi-target multi-constraint optimization problem model psi by updating the CA and the DA 1 To obtain a convergent population CA 0 And diversity population DA 0 Then from CA 0 The pareto frontier of the system can select an individual gamma 1 As an optimal search task planning scheme;
(4) Establishing a multi-target multi-constraint optimization problem model for searching and tracking task allocation planning of unmanned aerial vehicle search formation:
(4a) Central processing unit theta selects unmanned aerial vehicle search formation F μ Flight direction grid task area T ν Search formation F with unmanned aerial vehicle η Form unmanned aerial vehicle tracking formation F μη To F n Using search mission planning scheme gamma 1 For grid task region T ν Target T found by searching k ' establish track, remainder divide F μ And F η N-2 unmanned aerial vehicle search formation F 'according to decision variable assign [ m, N']And order [ n']Determining a grid task area search sequence distributed by unmanned aerial vehicle search formation, wherein F' = { F = } 1 ′,F′ 2 ,...,F′ n′ ,...,F′ N-2 Where μ = sent [ n ″ ]],sent[n″]To assign a decision variable, N ∈ { 1., N } \ η;
(4b) Established to minimize all grid task areas and discovered target T' k Sum of threat values S over time τ t ' integral f 1 ' minimizing the maximum time f for N-2 unmanned aerial vehicle search teams to complete respectively distributed L grid task areas 2 ' and minimize drone search formation F μ Reach grid task area T ν Time f of 3 ' As an objective function, the drones search for the mutual distance d ' to be queued between the sampling time points ' ti (F′ a ,F′ b ) Searching the mutual distance of the formation at the sampling time point by the unmanned planeSafety constraint c 1 ', maximum time of flight constraint for each drone search formation c 2 ', each grid task area T m Threat value constraint c of 3 ', found target T' k Threat value constraint c 4 'unmanned aerial vehicle search formation F' under constraint condition and unmanned aerial vehicle tracking formation carry out search and tracking in grid task area T by task allocation planning multi-target multi-constraint optimization problem model psi 2 Wherein:
min f 1 ′
min f 2 ′
min f 3 ′
c′ 3 :s(m,t)≤S m,max
c′ 4 :s(k,t)≤S k,max
wherein s (k, T) represents that target T 'is found at time T' k T' is a time offset, S k,max Denotes that target T has been found' k The upper limit value of the threat degree;
(5) Initializing parameters of a self-adaptive punished double-file evolution algorithm TAEA:
initializing the convergence population CA obtained in the step (3) 0 For the initial generation of the convergent population CA, randomly generated comprisesTaking the initial individual x as an initial generation population of the diversity population DA, wherein the iteration frequency is R, the maximum iteration frequency is R, and R =0;
(6) Updating the convergence population and the diversity population:
(6a) The r generation convergence population CA r And the r-th generation diversity population DA r Combining to obtain population Hm r Separately calculating the population Hm r In the genus of CA r And DA r Non-dominant individual population Hm of r Pc and Pd, and Pc is judged>If Pd is true, if yes, from CA r In the random selectionIndividual as a population P 1r Else from DA r In the random selectionIndividual as a population P 1r (ii) a Computing population CA r The ratio of PC to non-dominant individual and producing a composition comprisingOne is followedThe value of the f-th value pf (f) of the vector pf is judged according to the vector pf of the machine number>Whether PC is true, if so, from CA r In the method, one individual is randomly selected as P 2r Of the f individual, otherwise from the DA r In the method, one individual is randomly selected as P 2r The f th individual of (1), P 1r And P 2r Form a compositionGroup parent individual P r To P is to P r Each group of parent individuals in the group is subjected to uniform cross operation to obtainGroup P ru To P is to P ru Each group of individuals in the group is subjected to partial matching and crossing operation to obtainGroup P rp And from P rp Randomly selecting one individual from each group of individuals to form P rhalf Then to P rhalf Is uniformly mutated to obtain P rum To P is to P rum Performing insertion variation operation on each individual to obtain a filial generation population Q r ;
(6b) The r generation convergence population CA r And the offspring population Q r Composition of a population HC r The r-th generation convergence population DA r And the offspring population Q r Make up population HD r Population HC r Feasible individuals and infeasible individuals in the group respectively form a group Fs r And a population Is r ;
(6c) Respectively through the population HC r Group Fs r And population Is r For convergence population CA r Update, go through HD r For diversity population DA r Updating is carried out;
(7) Acquiring an unmanned aerial vehicle cluster collaborative search tracking task allocation planning result:
judging whether R = R is established, if yes, substituting the convergent population CA from the R r In the front surface of the pareto, one individual is arbitrarily selectedΥ 2 The sequence of the grid task areas distributed by the unmanned aerial vehicle search formation and the assigned search unmanned aerial vehicle formation determined by the value of the decision variable are used as a task allocation planning scheme for unmanned aerial vehicle cluster search tracking, otherwise, r = r +1 is set, and the step (6) is executed.
2. The method according to claim 1, wherein the set of mesh task regions t at step (2 a) n And a set of grid task areas T nL The acquisition method comprises the following steps:
(2a1) Assign a decision variable assign [ m, n ]]Randomly assigning binary values to the binary values and assigning an assign [ m, n ] to the assign]=1 for drone search formation F n And grid task area T m There is an allocation relation, assign [ m, n ] to]=0 expressed as F n And T m There is no assignment, and then will be F n All grid task area composition F with distribution relation n Set of grid task areas t n ;
(2a2) For sequential decision variables order n]Random assignment of [1, L]L non-repeating integers in between, and according to order [ n ]]Sequence of (2) to T n Sequence number of task area of all gridsSorting and then sorting the set by the sorted sequence numbersDetermination of F n Set T of grid task areas to which it is allocated to be executed n In the order of execution, get the set T arranged in the execution order nL 。
4. The method of claim 3, wherein the sum of threat values S in step (2 c) t Integral f of 1 Maximum time f 2 And the mutual distance d between the unmanned aerial vehicle search formation at the sampling time points ti (F a ,F b ) The calculation formulas are respectively as follows:
5. The method of claim 4, wherein the sum of threat values S in step (4 b) t ', and drone search formation F μ Reach grid task area T ν Time f' 3 The calculation formulas are respectively as follows:
6. A method according to claim 5, characterized in that step (6 c) is performed by HC respectively r 、Fs r And Is r For convergence population CA r Update by CA r And HD r For diversity population DA r Updating is carried out; the updating method comprises the following steps: when the temperature is higher than the set temperatureWhen a time of CA r =X r When it comes toWhen a time of CA r =S r ;Wherein X Is a population Is r N-I Fs selected according to a two-dimensional optimization problem formed by a constraint violation value of each individual and a corrected Chebyshev decomposition value based on adaptive punishment r A set of | optimal individuals; s r Representing a population Fs r The N optimal individuals selected by the correction Chebyshev decomposition method based on the self-adaptive penalty,modified chebyshev factorization method to represent population HD through adaptive penalties r And CA r The optimal individual set obtained in (1).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210930554.6A CN115202400A (en) | 2022-08-04 | 2022-08-04 | Unmanned aerial vehicle cluster task planning method based on self-adaptive penalty TAEA |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210930554.6A CN115202400A (en) | 2022-08-04 | 2022-08-04 | Unmanned aerial vehicle cluster task planning method based on self-adaptive penalty TAEA |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115202400A true CN115202400A (en) | 2022-10-18 |
Family
ID=83585699
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210930554.6A Pending CN115202400A (en) | 2022-08-04 | 2022-08-04 | Unmanned aerial vehicle cluster task planning method based on self-adaptive penalty TAEA |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115202400A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115454147A (en) * | 2022-10-29 | 2022-12-09 | 中国电子科技集团公司第十五研究所 | Unmanned aerial vehicle cluster cooperative task allocation method based on dynamic optimization |
CN116187594A (en) * | 2023-04-27 | 2023-05-30 | 北京玻色量子科技有限公司 | Multi-target multi-task path scheduling cost optimization method, device, medium and equipment |
CN116542468A (en) * | 2023-05-06 | 2023-08-04 | 中国人民解放军32370部队 | Unmanned aerial vehicle cluster task planning method |
-
2022
- 2022-08-04 CN CN202210930554.6A patent/CN115202400A/en active Pending
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115454147A (en) * | 2022-10-29 | 2022-12-09 | 中国电子科技集团公司第十五研究所 | Unmanned aerial vehicle cluster cooperative task allocation method based on dynamic optimization |
CN115454147B (en) * | 2022-10-29 | 2023-03-14 | 中国电子科技集团公司第十五研究所 | Unmanned aerial vehicle cluster cooperative task allocation method based on dynamic optimization |
CN116187594A (en) * | 2023-04-27 | 2023-05-30 | 北京玻色量子科技有限公司 | Multi-target multi-task path scheduling cost optimization method, device, medium and equipment |
CN116187594B (en) * | 2023-04-27 | 2023-09-08 | 北京玻色量子科技有限公司 | Multi-target multi-task path scheduling cost optimization method, device, medium and equipment |
CN116542468A (en) * | 2023-05-06 | 2023-08-04 | 中国人民解放军32370部队 | Unmanned aerial vehicle cluster task planning method |
CN116542468B (en) * | 2023-05-06 | 2023-10-20 | 中国人民解放军32370部队 | Unmanned aerial vehicle cluster task planning method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115202400A (en) | Unmanned aerial vehicle cluster task planning method based on self-adaptive penalty TAEA | |
Zhu et al. | An efficient evolutionary grey wolf optimizer for multi-objective flexible job shop scheduling problem with hierarchical job precedence constraints | |
Ding et al. | Improved particle swarm optimization algorithm based novel encoding and decoding schemes for flexible job shop scheduling problem | |
CN103279793B (en) | A kind of unmanned vehicle formation method for allocating tasks determined under environment | |
CN108632831B (en) | Unmanned aerial vehicle cluster frequency spectrum resource allocation method based on dynamic flight path | |
Wang et al. | Cooperative task allocation for heterogeneous multi-UAV using multi-objective optimization algorithm | |
CN108880663A (en) | Incorporate network resource allocation method based on improved adaptive GA-IAGA | |
Shang et al. | A multi-population cooperative coevolutionary algorithm for multi-objective capacitated arc routing problem | |
Pongchairerks | A two-level metaheuristic algorithm for the job-shop scheduling problem | |
CN110109358B (en) | Feedback-based hybrid multi-agent cooperative control method | |
CN112580943B (en) | Task decomposition and capability matching method applied to unmanned aerial vehicle cluster | |
CN112561395A (en) | Unmanned aerial vehicle cooperation method, system, device, electronic equipment and storage medium | |
CN112801540B (en) | Intelligent collaborative architecture design method based on unmanned clusters | |
CN113033970B (en) | AUV cluster layered distribution method for large-scale tasks | |
CN109242290B (en) | Automatic generation method for unmanned aerial vehicle group action scheme | |
CN113031650B (en) | Unmanned aerial vehicle cluster cooperative target distribution design method under uncertain environment | |
García-Galán et al. | Swarm fuzzy systems: knowledge acquisition in fuzzy systems and its applications in grid computing | |
CN108399105B (en) | Software and hardware partitioning method based on improved brainstorming algorithm | |
CN114167898A (en) | Global path planning method and system for data collection of unmanned aerial vehicle | |
CN110442143A (en) | A kind of unmanned plane situation data clustering method based on combination multiple target dove group's optimization | |
Wu et al. | A modified MOEA/D algorithm for solving bi-objective multi-stage weapon-target assignment problem | |
CN102708407A (en) | Self-adaptive hybrid multi-objective evolutionary method on basis of population decomposition | |
Jing et al. | Cooperative task assignment for heterogeneous multi-UAVs based on differential evolution algorithm | |
CN112668871A (en) | Dynamic assignment method for expert weight in multi-round group decision | |
CN116777170A (en) | Multi-robot task allocation method based on chaotic self-adaptive dung beetle optimization algorithm |
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 |