CN113671987B - Multi-machine distributed time sequence task allocation method based on non-deadlock contract net algorithm - Google Patents
Multi-machine distributed time sequence task allocation method based on non-deadlock contract net algorithm Download PDFInfo
- Publication number
- CN113671987B CN113671987B CN202110861873.1A CN202110861873A CN113671987B CN 113671987 B CN113671987 B CN 113671987B CN 202110861873 A CN202110861873 A CN 202110861873A CN 113671987 B CN113671987 B CN 113671987B
- Authority
- CN
- China
- Prior art keywords
- task
- unmanned aerial
- aerial vehicle
- deadlock
- algorithm
- 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.)
- Active
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 102
- 238000000034 method Methods 0.000 title claims abstract description 57
- 230000007246 mechanism Effects 0.000 claims abstract description 22
- 230000008569 process Effects 0.000 claims abstract description 20
- 238000005457 optimization Methods 0.000 claims abstract description 10
- 230000008878 coupling Effects 0.000 claims abstract description 7
- 238000010168 coupling process Methods 0.000 claims abstract description 7
- 238000005859 coupling reaction Methods 0.000 claims abstract description 7
- 238000004364 calculation method Methods 0.000 claims description 14
- 238000011156 evaluation Methods 0.000 claims description 14
- 238000012163 sequencing technique Methods 0.000 claims description 10
- 230000008859 change Effects 0.000 claims description 9
- 238000001514 detection method Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000013459 approach Methods 0.000 claims description 5
- 125000003275 alpha amino acid group Chemical group 0.000 claims description 2
- 150000001875 compounds Chemical class 0.000 claims description 2
- 230000006870 function Effects 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
-
- 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 discloses a multi-machine distributed time sequence task allocation method based on a non-deadlock contract net algorithm, and belongs to the technical field of task allocation. The implementation method of the invention comprises the following steps: aiming at the time sequence constraint problem of the cooperative task allocation of multiple unmanned planes, a time sequence task allocation model is established, and a time sequence task deadlock criterion is provided; in the process of ordering the tasks of the contract net algorithm, the influence of the coupling time sequence tasks is fully considered, the nearest neighbor tasks are preferentially selected, the deadlock criterion recursion backtracking for eliminating the infeasible ordering scheme is combined, the task ordering scheme meeting the deadlock constraint is generated, the selecting mechanism and the algorithm convergence condition of the bidding unmanned aerial vehicle are improved, a non-deadlock contract net (DF-CNP) algorithm based on a competition mechanism is further provided, the non-deadlock contract net (DF-CNP) algorithm is parallel under the distributed architecture, and the non-deadlock time sequence task allocation result of the multi-unmanned aerial vehicle is output efficiently. The invention is based on a competition mechanism, so that the invention has the characteristic of high optimization efficiency.
Description
Technical Field
The invention relates to a multi-machine distributed time sequence task allocation method based on a non-deadlock contract net algorithm, and belongs to the technical field of task allocation.
Background
The coupled task timing constraint brings great technical challenges to multi-machine task allocation, so that the phenomenon that a plurality of unmanned planes are in deadlock due to timing conflict is easily caused, and a feasible solution cannot be obtained. Therefore, it is important to effectively avoid the deadlock problem in time-series task allocation.
The time sequence task allocation method mainly comprises a centralized method and a distributed method. Under a centralized architecture, according to a deadlock generation mechanism, non-deadlock task scheduling of the whole system is realized through a central node. However, the centralized task allocation architecture has a central computing node, relies on global communication, and has poor system survivability. In addition, as the scale of the task allocation problem increases, the time consumption of the task allocation method under the centralized architecture increases in an ultra-linear manner, and the requirement of online real-time solution is difficult to meet.
Under the distributed architecture, each unmanned aerial vehicle does not depend on a central computing node, and efficient allocation of tasks is realized through inter-machine communication negotiation, so that the method has better algorithm robustness and stability. The market competition mechanism method generally adopts a layered solution mode to process the distributed time sequence task allocation problem, so that the task deadlock phenomenon can be effectively avoided, but the method solidifies the allocation sequence of the time sequence tasks, reduces the feasible solution space, and the optimality of allocation results is required to be further improved. In addition, under the potential game framework, the problem of complex time sequence task allocation is processed by introducing a scheduling algorithm, so that the solving efficiency of a large-scale problem can be remarkably improved, but the optimality of the method is still difficult to ensure due to the adoption of random sampling during balanced selection.
Disclosure of Invention
The invention discloses a multi-machine distributed time sequence task allocation method based on a non-deadlock contract net algorithm, which aims to solve the technical problems that: according to the actual task demands, the influence of the coupling time sequence tasks is fully considered in the task ordering process of the contract net algorithm, the infeasible ordering scheme is eliminated by combining with the deadlock criterion, the task time sequence deadlock is avoided, the bidding unmanned aerial vehicle selection mechanism and the contract net algorithm convergence condition are improved to further expand the feasible solution space, and the optimality of the multi-unmanned aerial vehicle time sequence task allocation result is improved. The invention is based on a competition mechanism, so that the invention has the characteristic of high optimization efficiency.
The invention aims at realizing the following technical scheme:
the invention discloses a multi-machine distributed time sequence task distribution method based on a non-deadlock contract network algorithm, which aims at the time sequence constraint problem of multi-unmanned plane cooperative task distribution, establishes a time sequence task distribution model and provides a time sequence task deadlock criterion. In the process of ordering the tasks of the contract net algorithm, the influence of the coupling time sequence tasks is fully considered, the nearest neighbor tasks are preferentially selected, the deadlock criterion recursion backtracking for eliminating the infeasible ordering scheme is combined, the task ordering scheme meeting the deadlock constraint is generated, the selecting mechanism and the algorithm convergence condition of the bidding unmanned aerial vehicle are improved, a non-deadlock contract net (DF-CNP) algorithm based on a competition mechanism is further provided, the non-deadlock contract net (DF-CNP) algorithm is parallel under the distributed architecture, and the non-deadlock time sequence task allocation result of the multi-unmanned aerial vehicle is output efficiently.
The invention discloses a multi-machine distributed time sequence task allocation method based on a non-deadlock contract net algorithm, which comprises the following steps:
step one: and taking the sum of the minimum unmanned aerial vehicle task execution time and the overall task completion time as optimization targets, and establishing a multi-unmanned aerial vehicle time sequence task allocation model by taking task time sequence constraint into consideration.
Considering W targets, the targets are denoted o= { O 1 ,O 2 ,…,O W }. The unmanned plane is required to sequentially execute reconnaissance, striking and assessment tasks on each target, and a task set is expressed as T= { T 1 ,T 2 ,…,T N And recording the task number as n=t, then n=3w. Consider that there are K-frame heterogeneous unmanned aerial vehicles, denoted as u= { U 1 ,U 2 ,…,U K And is specifically classified into combat type, reconnaissance type and striking type unmanned aerial vehicles. The fight type unmanned aerial vehicle can execute all tasks in T, the reconnaissance type unmanned aerial vehicle can execute reconnaissance and evaluation tasks, and the hit type unmanned aerial vehicle only executes hit tasks. Each task is only executed by a single unmanned aerial vehicle, and the task time sequence of each target is scout, strike and evaluate.
The overall task cost of the system is reduced and the cost of each machine is balanced in the task execution process, the sum of the minimum unmanned aerial vehicle task execution time and the overall task completion time are taken as optimization targets, and a multi-unmanned aerial vehicle time sequence task allocation model is constructed as follows
x i,n ∈{0,1} (3)
In the formula (1), alpha and beta are weight coefficients. X is x i,n Is a binary variable, when task T n Assigned to unmanned plane U i When x is i,n =1, and vice versa is 0.C (C) i,n Is unmanned plane U i Completion of task T n Is expressed as the duration of
Wherein,unmanned plane U i From task point T m Approach to the task point T n Duration of->Representing U i Arrive at task point T n Later waiting period,/->Representing task T n Is executed for a long time.
In the formula (4), the amino acid sequence of the compound,representing target O w Task X above, X e { R, A, V }; t is t str (. About.) represents the time at which the task starts to execute, t end (. Cndot.) is the end time of task execution.
Step two: in the process of ordering the tasks of the contract net algorithm, the influence of the coupling time sequence tasks is fully considered, the nearest neighbor tasks are preferentially selected, the deadlock criterion recursion backtracking for eliminating the infeasible ordering scheme is combined, and the selection mechanism of the bidding unmanned aerial vehicle and the convergence condition of the algorithm are improved, so that a non-deadlock contract net (DF-CNP) algorithm based on the competition mechanism is formed. Under the distributed architecture, a parallel non-deadlock contract network (DF-CNP) algorithm outputs a multi-unmanned aerial vehicle non-deadlock time sequence task allocation result with high efficiency, task time sequence deadlock is avoided by combining deadlock criteria in the process of ordering the contract network tasks, a feasible solution space is further expanded by improving a bid-inviting unmanned aerial vehicle selection mechanism and algorithm convergence conditions, and the optimality of the multi-unmanned aerial vehicle time sequence task allocation result is improved.
The unmanned aerial vehicle selection mechanism of the signer is improved as follows: all unmanned aerial vehicles are used as signer unmanned aerial vehicles according to the numbering sequence of the unmanned aerial vehicles. The algorithm convergence condition is improved as follows: after all unmanned aerial vehicles serve as signer unmanned aerial vehicles, algorithm cost is not reduced any more, and algorithm convergence conditions are met.
Step 2.1: unmanned plane U i And initializing parameters. Unmanned plane U i The parameters include unmanned plane U i Basic parameter information, a signer number, a target position, a machine number and a type. Unmanned plane U i The basic parameter information includes an initial position, a minimum turning radius.
Step 2.2: unmanned plane U i And (5) identity judgment. Before each round of negotiation starts, unmanned plane U i Judging the identity of the unmanned plane U if the unmanned plane U i The current identity is the bidder unmanned aerial vehicle, and the step 2.6 is carried out to wait for the bidder unmanned aerial vehicle to issue a task sequence; otherwise, the current identity is the signer unmanned plane, and step 2.3 is executed.
Step 2.3: unmanned aerial vehicle U for signer i And issuing the holding task sequence and the remaining task sequences to all bidders unmanned aerial vehicles. The holding task sequence is concretely an unmanned plane U i The current task sequence to be executed, and the rest task sequences refer to unmanned plane U i And selling the remaining task sequences after each task in turn.
Step 2.4: unmanned aerial vehicle U for signer i And publishing winning bid information. Unmanned plane U i Receiving bidding books issued by all bidders unmanned aerial vehicles and corresponding bidding tasks, selecting the unmanned aerial vehicle with the largest bidding value from the bidding books as a bidder, and throwing the bidding books to all biddersThe target unmanned aerial vehicle publishes the bid-winning information. And taking the rest task sequence after selling the bidding task as an updated task ordering scheme.
Step 2.5: and judging whether the condition of updating the unmanned aerial vehicle of the signer and converging the algorithm is met or not through the updating strategy of the unmanned aerial vehicle of the signer. If the updating condition of the signer unmanned aerial vehicle is met, the number of the signer unmanned aerial vehicle is backward and forward, and meanwhile, the updating information of the signer unmanned aerial vehicle is published to all bidder unmanned aerial vehicles. Returning to the step 2.2; if the convergence condition is satisfied, unmanned plane U i And issuing algorithm convergence information to all bidding unmanned aerial vehicles, and terminating a non-deadlock contract network (DF-CNP) algorithm, wherein a task allocation result generated by each unmanned aerial vehicle is a multi-unmanned aerial vehicle feasible allocation scheme meeting time sequence constraint.
The signer unmanned aerial vehicle updating strategy is as follows: when the cost of the system after the multi-round negotiation cannot be reduced, and the unmanned aerial vehicle still exists and is not used as the signer unmanned aerial vehicle, as shown in the formula (6), the signer unmanned aerial vehicle identity is moved to the subsequent unmanned aerial vehicle according to the number.
Wherein r represents the transaction of the r round, tau is convergence specified iteration number, Z auc Numbering the signers, wherein K is the number of unmanned aerial vehicles.
The algorithm convergence condition is as follows: if the cost of the system cannot be reduced after multiple negotiations and all unmanned aerial vehicles are used as signer unmanned aerial vehicles at the moment, the algorithm is converged and exited.
Step 2.6: if unmanned plane U i The identity is judged to be the bidder unmanned aerial vehicle, and after the bidder unmanned aerial vehicle releases information, unmanned aerial vehicle U is received i And carrying out calculation of bidding books in parallel with other bidder unmanned aerial vehicles, selecting a task held by the bidder unmanned aerial vehicle to incorporate into a task sequence, and completing non-deadlock sequencing through a non-deadlock sequencing algorithm. And calculating the cost according to the sorting result and the rest task sequence of the unmanned aerial vehicle of the tenderer, selecting the task minimizing the whole cost for bidding, and issuing a bidding document and a corresponding bidding task to the unmanned aerial vehicle of the tenderer.
The non-deadlock ordering algorithm specifically comprises the following steps: unmanned plane U i The holding task is regarded as a plurality of nodes, and the current node is constructedAdjacent task set->According to the estimated cost matrix L i Selecting the nearest node j to join the unmanned plane U i Task sequence S i Marking the node j as the accessed node until the current search depth d of the non-deadlock sequencing algorithm and the unmanned plane U i Number of held tasks N i The same applies. According to S i Constructing a task time sequence directed graph H, and performing S according to deadlock criteria i And (5) deadlock detection.
If the task sequence S i Is not deadlocked, and the search times lambda is less than or equal to a preset threshold valueThe non-deadlock ordering algorithm obtains a set of feasible ordering schemes, and the non-deadlock ordering algorithm ends; if->The non-deadlock ordering algorithm exits recursively. If the task sequence S i Deadlock, then slave task sequence S i And removing nearest neighbor node j, resetting j to be an unvisited node, and selecting the rest nearest neighbor nodes to continue the ordering process. If->And if no adjacent node exists, recursively backtracking to the previous node in the task sequence to continue the ordering process until the non-deadlock ordering algorithm is finished or exits.
The estimated cost matrix L i The calculation method is as follows:
wherein unmanned aerial vehicle U i The current task set is Q i ，Unmanned plane U i Slave task T m To T n The estimated approach cost of the task is estimated by using the Euclidean distance between tasks; />0) To execute task T n Pre-estimated wait cost->Unmanned plane U i Begin to execute task T n Is (are) estimated time of day>Unmanned plane U i Ending execution task T m Is a predicted time of the time frame; />Representing task T n Is executed for a long time.
The deadlock criterion is as follows: if id (H) is 0 or od (H) is 0, the global task non-deadlock is satisfied as long as the task is adjusted and H is loop-free. If id (H) and od (H) are not 0, only the task is adjusted to have no loop, and no new boundary reachable pair is generated by H, so that the global task can be guaranteed to be not deadlocked. Wherein task adjustment refers to unmanned plane U i Selecting a task held by a signer to incorporate the task sequence of the signer; the directional edge pointing to H is the entering arc of H, the number of which is the entering degree of H, and is marked as id (H); the directional edge indicated by H is the outgoing arc of H, and the number of the outgoing arcs is the outgoing degree of H and is denoted as od (H).
The cost calculation method comprises the following steps: the overall cost change before and after negotiation is calculated by (8)
Wherein, unmanned plane U of signer i Unmanned plane U of bidder j The holding task sequences are S respectively i 、S j The task set is Q i 、Q j The number of holding tasks is N i 、N j . Unmanned plane U j From S i Medium buying task T m Adding self sequence S j Unmanned plane U through non-deadlock sequencing j The task sequence becomesCorresponding unmanned aerial vehicle U i The remaining task sequence is->Unmanned plane U i Execution of task sequence S i Alpha, beta are weight coefficients in formula (1), S i,n Representing a task sequence S i N-th task of->Representing a task sequence S i Cost accumulation of each task. />The cost of three parts is: the approach duration cost, the wait duration cost, and the execution duration cost are calculated by equation (5).
The method for generating the bidding document comprises the following steps: unmanned plane U j Sequentially to S i All tasks in the system calculate the overall cost change before and after negotiation, select the task minimizing the overall cost as the bidding task, and issue the bid value bid j,i Maximum value of the overall cost change.
Step 2.7: unmanned plane U of bidder i Judging whether to bid in the middle. Unmanned plane U i And if the winning bid information is received, updating the task sequence of the winning bid information. If not, the original task sequence is maintained.
Step 2.8: if bidder unmanned plane U i After receiving the algorithm convergence information, terminating the non-deadlock contract network (DF-CNP) algorithm and the unmanned plane U i And executing the task according to the current allocation result. If the signer update information is received, the number of the signer unmanned aerial vehicle is backward and forward. Returning to step 2.2.
The method also comprises the following step three: and (3) according to the non-deadlock time sequence task distribution result of the multi-unmanned aerial vehicle output in the step (II), the multi-unmanned aerial vehicle executes corresponding reconnaissance, striking and evaluation time sequence tasks, and the total execution time length sum and the overall task completion time of the multi-unmanned aerial vehicle tasks are effectively shortened.
The task category of the detection, striking and evaluation task in the corresponding detection, striking and evaluation time sequence task can be deleted or expanded according to the actual task requirement.
Advantageous effects
1. The invention discloses a multi-machine distributed time sequence task distribution method based on a non-deadlock contract network algorithm, which aims at the time sequence constraint problem of multi-unmanned plane cooperative task distribution, establishes a time sequence task distribution model and provides a time sequence task deadlock criterion. In the process of ordering the tasks of the contract net algorithm, the influence of the coupling time sequence tasks is fully considered, the nearest neighbor tasks are preferentially selected, the deadlock criterion recursion backtracking for eliminating the infeasible ordering scheme is combined, the task ordering scheme meeting the deadlock constraint is generated, the selecting mechanism and the algorithm convergence condition of the bidding unmanned aerial vehicle are improved, a non-deadlock contract net (DF-CNP) algorithm based on a competition mechanism is further provided, the non-deadlock contract net (DF-CNP) algorithm is parallel under the distributed architecture, and the non-deadlock time sequence task allocation result of the multi-unmanned aerial vehicle is output efficiently.
2. According to the multi-machine distributed time sequence task distribution method based on the non-deadlock contract net algorithm, the deadlock criterion is combined in the contract net task sequencing process to avoid task time sequence deadlock, and all unmanned aerial vehicles can act as the unmanned aerial vehicles of the signer according to the unmanned aerial vehicle number sequence by improving the selection mechanism of the unmanned aerial vehicles and the algorithm convergence condition, so that the feasible solution space is further expanded, and the optimality of the time sequence task distribution result of the multi-unmanned aerial vehicles is improved.
Drawings
FIG. 1 is a flow chart of a multi-machine distributed time sequence task allocation method based on a non-deadlock contract net algorithm;
FIG. 2 is a timing task allocation result;
FIG. 3 is a Gantt chart of task execution times;
FIG. 4 is a comparison of algorithm optimality versus solution time.
Detailed Description
For better illustrating the objects and advantages of the present invention, the present invention is further described below with reference to the accompanying drawings and tables by way of example of multiple unmanned aerial vehicle time-series task allocation.
Example 1:
the simulation hardware environment is a PC with 5 InterXeon E5-4620.2 GHz and 16G memory, and the simulation software environment is Visual Studio 2015. The battlefield environment comprises 5 heterogeneous unmanned aerial vehicles and 7 targets, and for each target, the unmanned aerial vehicle needs to sequentially execute the reconnaissance, the striking and the evaluation tasks, so the battlefield environment comprises 21 tasks (T 1 ～T 21 Wherein T is 1 To T 3 Is a reconnaissance, hit, evaluation task on target 1, T 4 To T 6 Is a scout, hit, evaluation task on target 2, and so on). The time sequence task allocation requirement takes the minimum task completion time as an optimization target, and takes constraint conditions such as task time sequence, unmanned aerial vehicle performance and the like into consideration, so that the multi-task is reasonably allocated to each unmanned aerial vehicle.
As shown in fig. 1, the method for distributing multi-machine distributed time sequence tasks based on the non-deadlock contract network algorithm disclosed in the embodiment specifically comprises the following implementation steps:
step one: and taking the sum of the minimum unmanned aerial vehicle task execution time and the overall task completion time as optimization targets, and establishing a multi-unmanned aerial vehicle time sequence task allocation model by taking task time sequence constraint into consideration.
And (3) establishing a time sequence task allocation mathematical model according to the unmanned aerial vehicle, the target types, the number and the optimization targets in the environment, wherein the time sequence task allocation mathematical model is shown in (11) - (15).
x i,n ∈{0,1} (13)
Step two: under the distributed architecture, a parallel non-deadlock contract network (DF-CNP) algorithm outputs a multi-unmanned aerial vehicle non-deadlock time sequence task allocation result with high efficiency, task time sequence deadlock is avoided by combining deadlock criteria in the process of ordering the contract network tasks, a feasible solution space is further expanded by improving a bid-inviting unmanned aerial vehicle selection mechanism and algorithm convergence conditions, and the optimality of the multi-unmanned aerial vehicle time sequence task allocation result is improved.
The unmanned aerial vehicle selection mechanism of the signer is improved as follows: all unmanned aerial vehicles are used as signer unmanned aerial vehicles according to the numbering sequence of the unmanned aerial vehicles. The algorithm convergence condition is improved as follows: after all unmanned aerial vehicles serve as signer unmanned aerial vehicles, algorithm cost is not reduced any more, and algorithm convergence conditions are met.
Step 2.1: unmanned plane U i And initializing parameters. Number initialization of signer is 1, unmanned plane U i The initial position, minimum turning radius, target position, machine number and type are shown in tables 1-3.
Table 1 initial position information of unmanned aerial vehicle
| U 1 | U 2 | U 3 | U 4 | U 5 | |
| Position (m) | (800,0) | (1700,700) | (2800,100) | (3700,0) | (4500,0) |
TABLE 2 target initial position information
| O 1 | O 2 | O 3 | O 4 | |
| Position (m) | (600,1200) | (1700,2000) | (1800,3500) | (2700,4200) |
| O 5 | O 6 | O 7 | ||
| Position (m) | (4000,2000) | (5000,4000) | (5200,1000) |
Table 3 unmanned aerial vehicle performance parameters
| Parameters (parameters) | U 1 | U 2 | U 3 | U 4 | U 5 |
| Type(s) | Fight type | Fight type | Striking type | Reconnaissance type | Fight type |
| Numbering device | 1 | 2 | 3 | 4 | 5 |
| Turning radius (m) | 200 | 250 | 200 | 300 | 180 |
Step 2.2: unmanned plane U i And (5) identity judgment. Before each round of negotiation starts, unmanned plane U i Judging the identity of the unmanned plane U if the unmanned plane U i The current identity is the bidder unmanned aerial vehicle, and the step 2.6 is carried out to wait for the bidder unmanned aerial vehicle to issue a task sequence; otherwise, the current identity is the signer unmanned plane, and step 2.3 is executed.
Step 2.3: unmanned aerial vehicle U for signer i Issuing holding tasks to unmanned aerial vehicles of all biddersThe sequence and the remaining task sequence. The holding task sequence is concretely an unmanned plane U i The current task sequence to be executed, and the rest task sequences refer to unmanned plane U i And selling the remaining task sequences after each task in turn.
Step 2.4: unmanned aerial vehicle U for signer i And publishing winning bid information. Unmanned plane U i And receiving bidding books issued by all bidder unmanned aerial vehicles and corresponding bidding tasks, selecting the unmanned aerial vehicle with the largest bidding value from the bidding books as a bidder, and issuing bidding information to all bidder unmanned aerial vehicles. And taking the rest task sequence after selling the bidding task as an updated task ordering scheme.
Step 2.5: and judging whether the condition of updating the unmanned aerial vehicle of the signer and converging the algorithm is met or not through the updating strategy of the unmanned aerial vehicle of the signer. If the updating condition of the signer unmanned aerial vehicle is met, the number of the signer unmanned aerial vehicle is backward and forward, and meanwhile, the updating information of the signer unmanned aerial vehicle is published to all bidder unmanned aerial vehicles. Returning to the step 2.2; if the convergence condition is satisfied, unmanned plane U i And issuing algorithm convergence information to all bidding unmanned aerial vehicles, and terminating a non-deadlock contract network (DF-CNP) algorithm, wherein a task allocation result generated by each unmanned aerial vehicle is a multi-unmanned aerial vehicle feasible allocation scheme meeting time sequence constraint.
The signer unmanned aerial vehicle updating strategy is as follows: when the system cost after the continuous 3 rounds of negotiation cannot be reduced, and the unmanned aerial vehicle still exists and is not used as the signer unmanned aerial vehicle, as shown in the formula (16), the signer unmanned aerial vehicle identity is moved to the subsequent unmanned aerial vehicle according to the number.
The algorithm convergence condition is as follows: if the cost of the system cannot be reduced after multiple negotiations and all unmanned aerial vehicles are used as signer unmanned aerial vehicles at the moment, the algorithm is converged and exited.
Step 2.6: if unmanned plane U i The identity is judged to be the bidder unmanned aerial vehicle, and after the bidder unmanned aerial vehicle releases information, unmanned aerial vehicle U is received i Performing bidding calculation in parallel with other bidder unmanned aerial vehicles, and selecting the bidder not to existThe man-machine holding task is incorporated into the task sequence, and the non-deadlock sequencing is completed through a non-deadlock sequencing algorithm. And calculating the cost according to the sorting result and the rest task sequence of the unmanned aerial vehicle of the tenderer, selecting the task minimizing the whole cost for bidding, and issuing a bidding document and a corresponding bidding task to the unmanned aerial vehicle of the tenderer.
The non-deadlock ordering algorithm specifically comprises the following steps: unmanned plane U i The holding task is regarded as a plurality of nodes, and the current node is constructedAdjacent task set->According to the estimated cost matrix L i Selecting the nearest node j to join the unmanned plane U i Task sequence S i Marking the node j as the accessed node until the current search depth d of the non-deadlock sequencing algorithm and the unmanned plane U i Number of held tasks N i The same applies. According to S i Constructing a task time sequence directed graph H, and performing S according to deadlock criteria i And (5) deadlock detection.
If the task sequence S i Is not deadlocked, and the search times lambda is less than or equal to a preset threshold valueThe non-deadlock ordering algorithm obtains a set of feasible ordering schemes, and the non-deadlock ordering algorithm ends; if λ > 15, the non-deadlock ordering algorithm exits recursively. If the task sequence S i Deadlock, then slave task sequence S i And removing nearest neighbor node j, resetting j to be an unvisited node, and selecting the rest nearest neighbor nodes to continue the ordering process. If->And if no adjacent node exists, recursively backtracking to the previous node in the task sequence to continue the ordering process until the non-deadlock ordering algorithm is finished or exits.
The estimated cost matrix L i The calculation method is as follows:
the deadlock criterion is as follows: if id (H) is 0 or od (H) is 0, the global task non-deadlock can be satisfied as long as the task is adjusted and H is loop-free. If id (H) and od (H) are not 0, the overall task is guaranteed to be not deadlocked as long as the task is adjusted to have no loop and no new boundary reachable pair is generated by H.
The cost calculation method comprises the following steps: the overall cost change before and after negotiation is calculated by (18)
The method for generating the bidding document comprises the following steps: unmanned plane U j Sequentially to S i All tasks in the system calculate the overall cost change before and after negotiation, select the task minimizing the overall cost as the bidding task, and issue the bid value bid j,i Maximum value of the overall cost change.
Step 2.7: unmanned plane U of bidder i Judging whether to bid in the middle. Unmanned plane U i And if the winning bid information is received, updating the task sequence of the winning bid information. If not, the original task sequence is maintained.
Step 2.8: if bidder unmanned plane U i After receiving the algorithm convergence information, terminating the non-deadlock contract network (DF-CNP) algorithm and the unmanned plane U i And executing the task according to the current allocation result. If the signer receives moreNew information, the signer number is forward. Returning to step 2.2.
The method also comprises the following step three: and (3) according to the non-deadlock time sequence task distribution result of the multi-unmanned aerial vehicle output in the step (II), the multi-unmanned aerial vehicle executes corresponding reconnaissance, striking and evaluation time sequence tasks, and the total execution time length sum and the overall task completion time of the multi-unmanned aerial vehicle tasks are effectively shortened.
The task category of the detection, striking and evaluation task in the corresponding detection, striking and evaluation time sequence task can be deleted or expanded according to the actual task requirement.
When the operation of a non-deadlock contract network (DF-CNP) algorithm is finished, the algorithm convergence condition is met, and the task allocation result generated by each unmanned aerial vehicle is a multi-unmanned aerial vehicle feasible allocation scheme meeting the time sequence constraint. The task allocation result is shown in fig. 2, and the Gantt chart is shown in fig. 3. In the Gantt chart, the left end moment of the square represents that the unmanned aerial vehicle starts to execute the current task, and the right end moment of the square represents that the unmanned aerial vehicle completes the current task. The time interval between blocks represents the duration of time that the drone is approaching or waiting to perform a task. As can be seen from fig. 2 and 3, the above-mentioned time-series task allocation results satisfy heterogeneous flight platform constraint and task time-series non-deadlock constraint.
To fully embody DF-CNP solution optimality advantages, the method is compared with a non-deadlock genetic algorithm (TB-GA) and a coupling constraint consistency bundle algorithm (CBBA-TCC). 50 simulation tests were performed and the statistical results are shown in fig. 4. Compared with an optimal time-consuming mean value (0.489 s), the DF-CNP objective function mean value (126.8 s) is reduced by 8.5% and 97.9% respectively by TB-GA (objective function mean value 138.6s and calculation time-consuming mean value 23.32 s), and the DF-CNP optimization efficiency has obvious advantages on the premise of ensuring the optimal solution. As can be seen from the data distribution in fig. 4, the TB-GA calculation result has a certain randomness, while DF-CNP is a deterministic algorithm, and the solution robustness is better. Compared with CBBA-TCC (target function mean value 152.8s, calculation time-consuming mean value 0.274 s), DF-CNP has the maximum task execution duration reduced by 20.5% on the premise of meeting the sub-second planning efficiency, which indicates that the optimality of DF-CNP solving result is higher.
According to the simulation result and analysis of the multi-unmanned-plane time sequence task allocation example, the non-deadlock contract net algorithm can provide a feasible allocation scheme meeting task time sequence constraint for unmanned planes in sub-second level, and the algorithm optimality is better than that of the existing method, so that the method has strong engineering practicability, and the expected aim of the invention can be achieved.
The foregoing detailed description is provided for the purpose of illustrating the invention in further detail and is to be understood that this invention is not limited to the particular embodiments disclosed, but is intended to cover all modifications, equivalents, alternatives, and improvements as fall within the spirit and scope of the invention.
Claims (4)
1. The multi-machine distributed time sequence task allocation method based on the non-deadlock contract net algorithm is characterized by comprising the following steps of: comprises the following steps of the method,
step one: taking the sum of the minimum unmanned aerial vehicle task execution time length and the overall task completion time as optimization targets, and taking task time sequence constraint into consideration, establishing a multi-unmanned aerial vehicle time sequence task distribution model;
the first implementation method of the step is that,
considering W targets, the targets are denoted o= { O 1 ,O 2 ,…,O W -a }; the unmanned plane is required to sequentially execute reconnaissance, striking and assessment tasks on each target, and a task set is expressed as T= { T 1 ,T 2 ,…,T N Recording that the task number is n= |t||, then n=3w; consider that there are K-frame heterogeneous unmanned aerial vehicles, denoted as u= { U 1 ,U 2 ,…,U K The unmanned aerial vehicle is divided into a combat type unmanned aerial vehicle, a reconnaissance type unmanned aerial vehicle and a striking type unmanned aerial vehicle; the combat type unmanned aerial vehicle can execute all tasks in T, the reconnaissance type unmanned aerial vehicle can execute reconnaissance and evaluation tasks, and the striking type unmanned aerial vehicle only executes striking tasks; each task is only executed by a single unmanned aerial vehicle, and the task time sequence of each target is scout, strike and evaluate;
the overall task cost of the system is reduced and the cost of each machine is balanced in the task execution process, the sum of the minimum unmanned aerial vehicle task execution time and the overall task completion time are taken as optimization targets, and a multi-unmanned aerial vehicle time sequence task allocation model is constructed as follows
x i,n ∈{0,1} (3)
In the formula (1), alpha and beta are weight coefficients; x is x i,n Is a binary variable, when task T n Assigned to unmanned plane U i When x is i,n =1, whereas 0; c (C) i,n Is unmanned plane U i Completion of task T n Is expressed as the duration of
Wherein,unmanned plane U i From task point T m Approach to the task point T n Duration of->Representing U i Arrive at task point T n Later waiting period,/->Representing task T n Is executed for a time period of time;
in the formula (4), the amino acid sequence of the compound,representing target O w Task X above, X e { R, A, V }; t is t str (. About.) represents the time at which the task starts to execute, t end (-) is the task execution end time;
step two: fully considering the influence of the coupling time sequence task in the task ordering process of the contract net algorithm, preferentially selecting the nearest neighbor task, combining with the deadlock criterion recursion backtracking for eliminating the infeasible ordering scheme, and improving the selecting mechanism of the bidding unmanned aerial vehicle and the convergence condition of the algorithm to form a non-deadlock contract net (DF-CNP) algorithm based on the competition mechanism; under a distributed architecture, a parallel non-deadlock contract network (DF-CNP) algorithm outputs a multi-unmanned aerial vehicle non-deadlock time sequence task allocation result with high efficiency, task time sequence deadlock is avoided by combining deadlock criteria in the process of ordering the contract network tasks, a feasible solution space is further expanded by improving a bid-inviting unmanned aerial vehicle selection mechanism and algorithm convergence conditions, and the optimality of the multi-unmanned aerial vehicle time sequence task allocation result is improved;
the implementation method of the second step is that,
step 2.1: unmanned plane U i Initializing parameters; unmanned plane U i The parameters include unmanned plane U i Basic parameter information, a signer number, a target position, and other machine numbers and types; unmanned plane U i The basic parameter information comprises an initial position and a minimum turning radius;
step 2.2: unmanned plane U i Judging the identity; before each round of negotiation starts, unmanned plane U i Judging the identity of the unmanned plane U if the unmanned plane U i The current identity is the bidder unmanned aerial vehicle, and the step 2.6 is carried out to wait for the bidder unmanned aerial vehicle to issue a task sequence; otherwise, the current identity is the signer unmanned plane, and the step 2.3 is executed;
step 2.3: unmanned aerial vehicle U for signer i Issuing a holding task sequence and a remaining task sequence to all bidders unmanned aerial vehicles; the holding task sequence is concretely an unmanned plane U i The current task sequence to be executed, and the rest task sequences refer to unmanned plane U i Selling the residual task sequences after each task in sequence;
step 2.4: unmanned aerial vehicle U for signer i Publishing winning bid information; unmanned plane U i Receiving bidding books issued by all bidder unmanned aerial vehicles and corresponding bidding tasks, selecting the unmanned aerial vehicle with the largest bidding value from the bidding books as a bidder, and issuing bidding information to all bidder unmanned aerial vehicles; taking the rest task sequence after selling the bidding task as an updated task ordering scheme;
step 2.5: judging whether the condition of updating the unmanned aerial vehicle of the signer and converging an algorithm is met or not through the updating strategy of the unmanned aerial vehicle of the signer; if the updating condition of the signer unmanned aerial vehicle is met, the number of the signer unmanned aerial vehicle is backward and forward, and meanwhile, the updating information of the signer unmanned aerial vehicle is published to all bidder unmanned aerial vehicles; returning to the step 2.2; if the convergence condition is satisfied, unmanned plane U i Issuing algorithm convergence information to all bidding unmanned aerial vehicles, and terminating a non-deadlock contract network (DF-CNP) algorithm, wherein the task allocation result generated by each unmanned aerial vehicle is a multi-unmanned aerial vehicle feasible allocation scheme meeting time sequence constraint;
the signer unmanned aerial vehicle updating strategy is as follows: when the overall cost after multiple rounds of negotiation cannot be reduced and the unmanned aerial vehicle still exists and is not used as a signer unmanned aerial vehicle, the signer unmanned aerial vehicle identity is moved to a subsequent unmanned aerial vehicle according to the number in a sequential manner as shown in the formula (6);
wherein r represents the transaction of the r round, tau is convergence specified iteration number, Z auc Numbering the signers, wherein K is the number of unmanned aerial vehicles;
the algorithm convergence condition is as follows: if the overall cost cannot be reduced after multiple negotiations and all unmanned aerial vehicles are used as signer unmanned aerial vehicles at the moment, the algorithm converges and exits;
step 2.6: if unmanned plane U i The identity is judged to be the bidder unmanned aerial vehicle, and after the bidder unmanned aerial vehicle releases information, unmanned aerial vehicle U is received i Performing bidding calculation in parallel with other bidder unmanned aerial vehicles, selecting a task held by a bidder unmanned aerial vehicle to incorporate the task sequence, and completing non-deadlock ranking through a non-deadlock ranking algorithmSequencing; according to the sorting result and the rest task sequence of the unmanned aerial vehicle of the tenderer, performing cost calculation by a cost calculation method, selecting a task minimizing the whole cost for bidding, and issuing a bidding document and a corresponding bidding task to the unmanned aerial vehicle of the tenderer;
the non-deadlock ordering algorithm specifically comprises the following steps: unmanned plane U i The holding task is regarded as a plurality of nodes, and the current node is constructedAdjacent task set->According to the estimated cost matrix L i Selecting the nearest node j to join the unmanned plane U i Task sequence S i Marking the node j as the accessed node until the current search depth d of the non-deadlock sequencing algorithm and the unmanned plane U i Number of held tasks N i The same; according to S i Constructing a task time sequence directed graph H, and performing S according to deadlock criteria i Deadlock detection;
if the task sequence S i Is not deadlocked, and the search times lambda is less than or equal to a preset threshold valueThe non-deadlock ordering algorithm obtains a set of feasible ordering schemes, and the non-deadlock ordering algorithm ends; if->The non-deadlock ordering algorithm recursively exits; if the task sequence S i Deadlock, then slave task sequence S i Removing nearest neighbor node j, resetting j to be an unaccessed node, and selecting the rest nearest neighbor nodes to continue the ordering process; if->If no adjacent node exists, recursively backtracking to the previous node in the task sequence to continue the sorting process until the non-deadlock sorting algorithm ends or exits；
The estimated cost matrix L i The calculation method is as follows:
wherein unmanned aerial vehicle U i The current task set is Q i ，Unmanned plane U i Slave task T m To T n The estimated approach cost of the task is estimated by using the Euclidean distance between tasks; />0) To execute task T n Pre-estimated wait cost->Unmanned plane U i Begin to execute task T n Is (are) estimated time of day>Unmanned plane U i Ending execution task T m Is a predicted time of the time frame; />Representing task T n Is executed for a time period of time;
the deadlock criterion is as follows: if id (H) is 0 or od (H) is 0, the global task non-deadlock is satisfied as long as the task is adjusted to have no loop; if id (H) and od (H) are not 0, only the task is adjusted to have no loop, and no new boundary reachable pair is generated by H, so that the global task can be guaranteed to be not deadlocked; wherein task adjustment refers to unmanned plane U i Selecting a signer to hold a task to be incorporated into itselfA process of a task sequence; the directional edge pointing to H is the entering arc of H, the number of which is the entering degree of H, and is marked as id (H); the directional edge indicated by H is an outgoing arc of H, the number of which is the outgoing degree of H and is marked as od (H);
the cost calculation method comprises the following steps: the overall cost change before and after negotiation is calculated by (8)
Wherein, unmanned plane U of signer i Unmanned plane U of bidder j The holding task sequences are S respectively i 、S j The task set is Q i 、Q j The number of holding tasks is N i 、N j The method comprises the steps of carrying out a first treatment on the surface of the Unmanned plane U j From S i Medium buying task T m Adding self sequence S j Unmanned plane U through non-deadlock sequencing j The task sequence becomesCorresponding unmanned aerial vehicle U i The remaining task sequence is-> Unmanned plane U i Execution of task sequence S i Alpha, beta are weight coefficients in formula (1), S i,n Representing a task sequence S i N-th task of->Representing a task sequence S i The cost of each task is accumulated; />The cost of three parts is: the approximate duration cost, the waiting duration cost and the execution duration cost are calculated by the formula (5);
the method for generating the bidding document comprises the following steps: unmanned plane U j Sequentially to S i All tasks in the system calculate the overall cost change before and after negotiation, select the task minimizing the overall cost as the bidding task, and issue the bid value bid j,i Maximum value of the overall cost change;
bid j,i ＝ΔC(T m* ) (10)
step 2.7: unmanned plane U of bidder i Judging whether to bid up; unmanned plane U i If the winning bid information is received, updating the task sequence of the winning bid information; if the task sequence is not marked, the original task sequence is maintained;
step 2.8: if bidder unmanned plane U i After receiving the algorithm convergence information, terminating the non-deadlock contract network (DF-CNP) algorithm and the unmanned plane U i Executing tasks according to the current allocation results; if the signer update information is received, the number of the signer unmanned aerial vehicle is backward and forward; returning to step 2.2.
2. The multi-machine distributed time-series task allocation method based on the non-deadlock contract net algorithm as set forth in claim 1, wherein: and step three, according to the non-deadlock time sequence task distribution result of the multi-unmanned aerial vehicle output in the step two, the multi-unmanned aerial vehicle executes corresponding reconnaissance, striking and evaluation time sequence tasks, and the total execution time length sum and the overall task completion time of the multi-unmanned aerial vehicle tasks are effectively shortened.
3. The multi-machine distributed time sequence task allocation method based on the non-deadlock contract net algorithm as set forth in claim 2, wherein: the task category of the detection, striking and evaluation task in the corresponding detection, striking and evaluation time sequence task can be deleted or expanded according to the actual task requirement.
4. The multi-machine distributed time-series task allocation method based on the non-deadlock contract net algorithm according to claim 3, wherein the method is characterized in that: the unmanned aerial vehicle selection mechanism of the signer is improved as follows: all unmanned aerial vehicles are used as signer unmanned aerial vehicles according to the numbering sequence of the unmanned aerial vehicles; the algorithm convergence condition is improved as follows: after all unmanned aerial vehicles serve as signer unmanned aerial vehicles, algorithm cost is not reduced any more, and algorithm convergence conditions are met.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110861873.1A CN113671987B (en) | 2021-07-29 | 2021-07-29 | Multi-machine distributed time sequence task allocation method based on non-deadlock contract net algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110861873.1A CN113671987B (en) | 2021-07-29 | 2021-07-29 | Multi-machine distributed time sequence task allocation method based on non-deadlock contract net algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113671987A CN113671987A (en) | 2021-11-19 |
| CN113671987B true CN113671987B (en) | 2024-02-02 |
Family
ID=78540645
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110861873.1A Active CN113671987B (en) | 2021-07-29 | 2021-07-29 | Multi-machine distributed time sequence task allocation method based on non-deadlock contract net algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113671987B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114200963B (en) * | 2022-02-17 | 2022-05-10 | 佛山科学技术学院 | Unmanned aerial vehicle autonomous mission planning method and device under dynamic environment and storage medium |
| CN114545975B (en) * | 2022-03-08 | 2023-03-17 | 大连理工大学 | Multi-unmanned aerial vehicle system task allocation method integrating multi-target evolution algorithm and contract network algorithm |
| CN115113626B (en) * | 2022-07-20 | 2023-04-11 | 威康(深圳)智能有限公司 | Robot distributed control system and method thereof |
| CN115525068B (en) * | 2022-10-29 | 2023-03-28 | 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 iterative optimization |
| CN115963852B (en) * | 2022-11-21 | 2023-09-12 | 北京航空航天大学 | Unmanned aerial vehicle cluster construction method based on negotiation mechanism |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105139069A (en) * | 2015-07-30 | 2015-12-09 | 江苏科技大学 | Method for improving contract net based on particle swarm algorithm |
| CN107330560A (en) * | 2017-07-04 | 2017-11-07 | 北京理工大学 | A kind of multitask coordinated distribution method of isomery aircraft for considering temporal constraint |
| CN107515618A (en) * | 2017-09-05 | 2017-12-26 | 北京理工大学 | A kind of isomery unmanned plane cotasking distribution method for considering time window |
| CN111144784A (en) * | 2019-12-31 | 2020-05-12 | 中国电子科技集团公司信息科学研究院 | Task allocation method and system for manned/unmanned cooperative formation system |
| AU2020101065A4 (en) * | 2020-06-19 | 2020-07-23 | Hubei University Of Technology | Method for scheduling UAVs based on chaotic adaptive firefly algorithm |
| CN113009934A (en) * | 2021-03-24 | 2021-06-22 | 西北工业大学 | Multi-unmanned aerial vehicle task dynamic allocation method based on improved particle swarm optimization |
-
2021
- 2021-07-29 CN CN202110861873.1A patent/CN113671987B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105139069A (en) * | 2015-07-30 | 2015-12-09 | 江苏科技大学 | Method for improving contract net based on particle swarm algorithm |
| CN107330560A (en) * | 2017-07-04 | 2017-11-07 | 北京理工大学 | A kind of multitask coordinated distribution method of isomery aircraft for considering temporal constraint |
| CN107515618A (en) * | 2017-09-05 | 2017-12-26 | 北京理工大学 | A kind of isomery unmanned plane cotasking distribution method for considering time window |
| CN111144784A (en) * | 2019-12-31 | 2020-05-12 | 中国电子科技集团公司信息科学研究院 | Task allocation method and system for manned/unmanned cooperative formation system |
| AU2020101065A4 (en) * | 2020-06-19 | 2020-07-23 | Hubei University Of Technology | Method for scheduling UAVs based on chaotic adaptive firefly algorithm |
| CN113009934A (en) * | 2021-03-24 | 2021-06-22 | 西北工业大学 | Multi-unmanned aerial vehicle task dynamic allocation method based on improved particle swarm optimization |
Non-Patent Citations (2)
| Title |
|---|
| 保性的Petri网细化运算在多主体协作中的应用;方欢;;安徽理工大学学报(自然科学版);第29卷(第02期);50-56 * |
| 多机器人路径规划研究方法;张亚鸣;雷小宇;杨胜跃;樊晓平;瞿志华;贾占朝;;计算机应用研究;第25卷(第09期);2566-2569 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113671987A (en) | 2021-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113671987B (en) | Multi-machine distributed time sequence task allocation method based on non-deadlock contract net algorithm | |
| CN110134146B (en) | Distributed multi-unmanned aerial vehicle task allocation method under uncertain environment | |
| Muller et al. | A generalized training approach for multiagent learning | |
| CN106990792B (en) | Multi-unmanned aerial vehicle collaborative time sequence coupling task allocation method based on hybrid gravity search algorithm | |
| CN105302153B (en) | The planing method for the task of beating is examined in the collaboration of isomery multiple no-manned plane | |
| Dai et al. | Multi-robot dynamic task allocation for exploration and destruction | |
| CN107886201B (en) | Multi-objective optimization method and device for multi-unmanned aerial vehicle task allocation | |
| CN109669452A (en) | A kind of cloud robot task dispatching method and system based on parallel intensified learning | |
| CN111784211B (en) | Cluster-based group multitask allocation method and storage medium | |
| CN114281104B (en) | Multi-unmanned aerial vehicle cooperative regulation and control method based on improved ant colony algorithm | |
| CN114594794A (en) | Multi-machine collaborative task planning method considering subsystem execution capacity | |
| CN105094970B (en) | The method of more times scheduling models of task can be divided under a kind of solution distributed system | |
| CN112013829A (en) | Multi-UAV/UGV (unmanned aerial vehicle/user generated Unit) cooperative long-term operation path planning method based on multi-objective optimization | |
| Agarwal et al. | Non-additive multi-objective robot coalition formation | |
| CN114415735A (en) | Dynamic environment-oriented multi-unmanned aerial vehicle distributed intelligent task allocation method | |
| CN114154819A (en) | Task similarity-based multi-AGV distributed scheduling method and system | |
| Sallam et al. | A hybrid differential evolution with cuckoo search for solving resource constrained project scheduling problems | |
| CN109636709A (en) | A kind of figure calculation method suitable for heterogeneous platform | |
| CN115022231B (en) | Optimal path planning method and system based on deep reinforcement learning | |
| CN115689184A (en) | Distributed task allocation method based on consensus binding algorithm | |
| CN115494873A (en) | Heterogeneous multi-unmanned aerial vehicle cooperative task allocation method based on Monte Carlo tree search architecture under time sequence constraint | |
| CN116089083A (en) | Multi-target data center resource scheduling method | |
| CN115273502A (en) | Traffic signal cooperative control method | |
| CN115150335A (en) | Optimal flow segmentation method and system based on deep reinforcement learning | |
| CN114610058A (en) | Multi-task distributed task allocation method for cluster aircraft |
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 |