CN113703482B - Task planning method based on simplified attention network in large-scale unmanned aerial vehicle cluster - Google Patents
Task planning method based on simplified attention network in large-scale unmanned aerial vehicle cluster Download PDFInfo
- Publication number
- CN113703482B CN113703482B CN202111006869.3A CN202111006869A CN113703482B CN 113703482 B CN113703482 B CN 113703482B CN 202111006869 A CN202111006869 A CN 202111006869A CN 113703482 B CN113703482 B CN 113703482B
- Authority
- CN
- China
- Prior art keywords
- unmanned aerial
- task
- aerial vehicle
- simplified
- attention network
- 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
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000004364 calculation method Methods 0.000 claims description 17
- 238000013528 artificial neural network Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 10
- 238000005311 autocorrelation function Methods 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 9
- 238000005314 correlation function Methods 0.000 claims description 5
- 238000011478 gradient descent method Methods 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 2
- 239000000284 extract Substances 0.000 abstract 1
- 238000004891 communication Methods 0.000 description 3
- 230000004927 fusion Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Automation & Control Theory (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a task planning method based on a simplified attention network in a large-scale unmanned aerial vehicle cluster, which mainly solves the task resource planning problem in the large-scale unmanned aerial vehicle cluster, and the implementation steps of the invention are as follows: 1. constructing and generating a sample set; 2. constructing a simplified attention network; 3. training a simplified attention network; 4. and planning the unmanned aerial vehicles which execute tasks in the large unmanned aerial vehicle cluster. The invention quickly and accurately extracts the high-dimensional characteristics of the unmanned aerial vehicle in the large unmanned aerial vehicle cluster through the constructed simplified attention model, aims at effectively utilizing various task resources carried in the large unmanned aerial vehicle cluster, optimizes the utilization rate of the task resources in the large unmanned aerial vehicle cluster, and effectively solves the problem of task resource planning of the large unmanned aerial vehicle cluster.
Description
Technical Field
The invention belongs to the technical field of communication, and further relates to a task planning method based on a simplified attention network in a large unmanned aerial vehicle cluster in the technical field of unmanned aerial vehicle task planning. The invention can be applied to a large unmanned aerial vehicle cluster which is formed by at least 100 unmanned aerial vehicles with different attributes, realizes the scheduling of task resources, meets the requirements of task targets on various task resources, and effectively distributes the task resources carried by the unmanned aerial vehicle cluster in real time.
Background
Unmanned Aerial Vehicle (UAV) cluster is composed of low-cost UAVs with various functions and different structures, and different UAVs can carry different types of airborne equipment, so that the UAV cluster has the advantages of low task execution cost, high task execution efficiency and the like. The existing unmanned aerial vehicle cluster task allocation methods mainly comprise two methods, one is a heuristic task allocation algorithm based on iterative optimization, and the other is a rapid task allocation algorithm based on reinforcement learning. The heuristic task allocation method is long in operation time and cannot meet the real-time requirement of task planning when facing the problem of task allocation of large unmanned aerial vehicle clusters, and the machine learning task planning algorithm is poor in robustness when different task planning problems are processed although the overall operation time of the method is short.
The patent document of Nanjing post and telecommunications university 'A unmanned aerial vehicle mission planning method based on a self-organizing neural network' (application number: 201711495472.9 application date: 2018.06.15 application publication number: CN 108170147A) discloses an unmanned aerial vehicle mission planning method based on a self-organizing neural network. The method comprises the following specific steps: initializing a self-organizing network, including initializing the number of nodes of a neural network, initializing a node arrangement form and the like; the second step: selecting winning nodes in the self-organizing network according to a set algorithm rule; the third step: constructing a winning node win field by taking the winning node as a center; the fourth step: updating network parameters of the self-organizing network according to the current iteration times, the constructed winning node win area, the weight updating rate in the winning node and the like; the fifth step: and when an emergency situation of task planning is faced, setting a dynamic response rule, and updating the network parameters again. The method has the disadvantages that the number of unmanned aerial vehicles in the task planning method is small, and the number of iterations required by the method is too large when the problem of large unmanned aerial vehicle cluster task allocation is solved, so that the method cannot be applied to the problem of large unmanned aerial vehicle cluster task allocation.
The patent document 'unmanned aerial vehicle task cooperative allocation method based on Q learning' (application number: 202010612864.4 application date: 2020.06.30 application publication number: CN 111736461A) applied by the university of electronic science and technology of Xian discloses an unmanned aerial vehicle task cooperative allocation method based on Q learning. The method comprises the following specific steps: initializing network parameters according to the unmanned aerial vehicle information and the task information; the second step is that: establishing a sensitive unmanned aerial vehicle and indirect unmanned aerial vehicle alliance, and establishing a communication rule between the sensitive unmanned aerial vehicle and the indirect unmanned aerial vehicle; the third step: updating Q learning model parameters according to the communication rule; the fourth step: and acquiring a task distribution result. The method has the defects that the structure for constructing the model is complex, and the calculation amount of the whole model is large, so that the method cannot be applied to the problem of task allocation of large unmanned aerial vehicle clusters.
Disclosure of Invention
The invention aims to provide a task planning method based on simplified attention in a large unmanned aerial vehicle cluster aiming at the defects of the prior art, and is used for solving the problems of complex network structure and large calculation amount in a large unmanned aerial vehicle cluster task allocation method.
The specific idea for realizing the invention is as follows: and (3) adopting the constructed and trained simplified attention network to extract the high-dimensional characteristics of the unmanned aerial vehicles in the large-scale unmanned aerial vehicle cluster in real time, and iterating a plurality of unmanned aerial vehicles by using a probability formula to execute a task target. In a feature extraction network of a simplified attention network, the constructed autocorrelation coefficient layer establishes the association between each unmanned aerial vehicle feature and the unmanned aerial vehicle, the autocorrelation coefficient is used for representing the degree of association between each unmanned aerial vehicle feature and the unmanned aerial vehicle, the constructed mutual correlation coefficient layer establishes the association between each unmanned aerial vehicle feature and an unmanned aerial vehicle cluster, the cross correlation coefficient is used for representing the degree of association between each unmanned aerial vehicle feature and the unmanned aerial vehicle cluster, the autocorrelation coefficient and the cross correlation coefficient are obtained by combining the characteristics of each unmanned aerial vehicle, the high-dimensional features of each unmanned aerial vehicle are extracted in a fusion manner, the problem that the high-dimensional features of the unmanned aerial vehicles can be accurately obtained only by associating one unmanned aerial vehicle with all unmanned aerial vehicles in the network in the prior art is solved, and the association times of each unmanned aerial vehicle are reduced; the invention updates the parameters of the whole network by using a gradient descent method, solves the task planning problem of the large-scale unmanned aerial vehicle cluster in real time, and overcomes the problem of high calculation complexity in the network in the prior art;
the steps of the invention comprise:
step 1, generating a sample set:
generating a 10000 groups sample set; each group of sample sets comprises a large unmanned aerial vehicle cluster formed by at least 100 unmanned aerial vehicles and a task target executed by the large unmanned aerial vehicle cluster;
step 2, constructing a simplified attention network:
(2a) constructing a feature extraction network consisting of an input layer, an autocorrelation coefficient calculation layer, a cross correlation coefficient calculation layer and an output layer, setting the input dimensionality of the input layer to be 128, and setting the output dimensionality to be 256; setting the input dimensionalities of an autocorrelation system calculation layer, a cross correlation coefficient calculation layer and an output layer to be 256, and setting the output dimensionalities to be 256; the autocorrelation coefficient calculation layer is realized by an autocorrelation function, and the cross-correlation coefficient calculation layer is realized by a cross-correlation function;
(2b) setting parameters of a fully-connected neural network and an LSTM network;
setting the input dimension of the fully-connected neural network to be 8 and the output dimension to be 128; setting the input dimension of the LSTM network to be 256 and the output dimension to be 512;
(2c) cascading a feature extraction network, a fully-connected neural network and an LSTM (Long-Short Term Memory) network into a simplified attention network;
step 3, training a simplified attention network:
inputting the sample set into a simplified attention network, and iteratively updating the parameters of the simplified attention network by using a gradient descent method until the loss function value is converged to obtain a trained simplified attention network;
step 4, planning unmanned aerial vehicle for executing tasks in large-scale unmanned aerial vehicle cluster
(4a) Inputting the type and quantity of task resources carried by each unmanned aerial vehicle in a large unmanned aerial vehicle cluster to be planned and the type and quantity of task resources required by a task target into a trained simplified attention network, and outputting high-dimensional characteristics of each unmanned aerial vehicle;
(4b) calculating the probability of each unmanned aerial vehicle in the large-scale unmanned aerial vehicle cluster to execute the task target by using a probability formula, and selecting the unmanned aerial vehicle corresponding to the maximum probability as the unmanned aerial vehicle executing the task target;
(4c) inputting the types and the number of the task resources of the remaining unmanned aerial vehicles in the large unmanned aerial vehicle cluster to be planned into the trained simplified attention network to obtain the high-dimensional characteristics of each remaining unmanned aerial vehicle in the large unmanned aerial vehicle cluster to be planned, and selecting the unmanned aerial vehicle for executing the task by adopting the method same as the step (4 b);
(4d) judging whether the task resource types carried by all the unmanned aerial vehicles for executing the task target meet the constraint condition, if so, executing the step 5, otherwise, executing the step (4c)
And 5, taking the selected multiple unmanned aerial vehicles for executing the task targets as a planning scheme.
Compared with the prior art, the invention has the following advantages:
firstly, the invention adopts the simplified attention network to carry out task planning on the large-scale unmanned aerial vehicle group, and overcomes the problem that algorithm convergence is realized by increasing a large number of algorithm iteration times along with the increase of the number of unmanned aerial vehicles when the task planning problem of the large-scale unmanned aerial vehicle group is processed in the prior art, so that the invention has the characteristics of strong algorithm robustness and high task planning speed, is beneficial to reducing the time of task planning of the large-scale unmanned aerial vehicle group, and optimizes the resource utilization rate of the task planning of the large-scale unmanned aerial vehicle group.
Secondly, when the simplified attention network is constructed to extract the features of each unmanned aerial vehicle in the large unmanned aerial vehicle cluster, the high-dimensional features of each unmanned aerial vehicle are extracted by quickly fusing the autocorrelation coefficient and the cross correlation coefficient, so that the problem that the high-dimensional features of the unmanned aerial vehicles can be accurately extracted only by a neural network with a complex structure in the prior art is solved, and the unmanned aerial vehicle feature extraction method has the characteristics of simple network structure and accurate unmanned aerial vehicle feature extraction.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The invention is described in further detail below with reference to fig. 1 and the examples.
Step 1, generating a sample set:
in the embodiment of the invention, a 10000 groups of sample sets are generated; each group of sample set comprises a large unmanned aerial vehicle cluster formed by 100 unmanned aerial vehicles and a task target executed by the large unmanned aerial vehicle cluster.
In the embodiment of the invention, the types of the task resources carried by each unmanned aerial vehicle in the large unmanned aerial vehicle cluster are 5, and the value range of the quantity of each task resource is [0,1 ].
The types of the task resources required by the task target executed by the large unmanned aerial vehicle cluster are 5, and the value range of the quantity of each task resource is [2,5 ].
Step 2, constructing a simplified attention network:
building a feature extraction network consisting of an input layer, an autocorrelation coefficient calculation layer, a cross-correlation coefficient calculation layer and an output layer, setting the input dimension of the input layer to be 128, and setting the output dimension to be 256; setting the input dimensionality of the hidden layer, the autocorrelation system calculating layer, the cross correlation coefficient calculating layer and the output layer to be 256, and setting the output dimensionality to be 256; the autocorrelation calculating layer is implemented by an autocorrelation function, and the cross-correlation coefficient calculating layer is implemented by a cross-correlation function.
The autocorrelation function is as follows:
wherein λ is i An autocorrelation function, h, representing the ith training data in a reduced attention network i Initial characteristics, l, representing the ith training data q And l k Parameter representing a reduced attention network, d hi Represents h i Relu represents the linear rectification activation function operation, and N represents the total number of training data in each set of samples.
The cross-correlation function is as follows:
wherein, beta i An autocorrelation function of the ith training data in the reduced attention network is shown.
The fusion characteristics of the training data are calculated as follows:
wherein H i Represents a fusion feature of the ith training data in the reduced attention network, l v Representing parameters of a simplified attention network.
Setting parameters of a fully-connected neural network and an LSTM network;
setting the input dimension of the fully-connected neural network to be 8 and the output dimension to be 128; the LSTM network input dimension is set to 256 and the output dimension is set to 512.
And cascading the feature extraction network, the fully-connected neural network and the LSTM (Long-Short Term Memory) network into a simplified attention network.
Step 3, training a simplified attention network:
and inputting the sample set into the simplified attention network, and iteratively updating the parameters of the simplified attention network by using a gradient descent method until the loss function value is converged to obtain the trained simplified attention network.
The loss function is as follows:
wherein F represents a loss function value, n represents the total number of the unmanned aerial vehicles executing the task, Σ represents summation operation, g represents the number of the unmanned aerial vehicles executing the task, and the value range of g is [1, n ]]S represents the total number of task resource categories required by the task object, m represents the number of task resource categories required by the task object, a g,m The number of the m-th task resources carried by the g-th task execution unmanned aerial vehicle is shown.
And 4, planning the unmanned aerial vehicle for executing the task in the large unmanned aerial vehicle cluster.
Inputting the types and the quantity of task resources carried by each unmanned aerial vehicle in a large unmanned aerial vehicle cluster to be planned and the types and the quantity of the task resources required by a task target into a trained simplified attention network, and outputting the high-dimensional characteristics of each unmanned aerial vehicle.
In the embodiment of the invention, the number of the unmanned aerial vehicles in the large unmanned aerial vehicle cluster to be planned is 100, the types of the task resources carried by each unmanned aerial vehicle are 5, and the value range of the number of each task resource is [0,1 ].
In the embodiment of the invention, the types of task resources required by the task targets to be executed in the large unmanned aerial vehicle cluster to be planned are 5, and the number of each type of task resources is 2.
(4b) And calculating the probability of each unmanned aerial vehicle in the large-scale unmanned aerial vehicle cluster to execute the task target by using a probability formula, and selecting the unmanned aerial vehicle corresponding to the maximum probability as the unmanned aerial vehicle executing the task target.
The probability formula is as follows:
wherein p is i The probability of the ith unmanned aerial vehicle in the large unmanned aerial vehicle cluster to be planned to execute the task target is represented, Exp represents exponential operation with a natural constant e as a base, H i And the high-dimensional characteristics of the ith frame in the large unmanned aerial vehicle cluster to be planned are shown.
The probability formula is used for calculating that the probabilities that 10 unmanned aerial vehicles numbered 8, 25, 36, 45, 19, 89, 76, 84, 89 and 92 execute the task target are the maximum, and the probabilities are 0.8, 0.65, 0.89, 0.98, 0.78, 0.86, 0.71, 0.89, 0.93 and 0.83 respectively.
The types of task resources carried by 10 unmanned persons executing the task are 5, which are equal to the types of the task resources, and the total amount of the carried task resources is 12, which is larger than the total amount of the task resource requirements.
And 5, taking the selected 10 unmanned aerial vehicles executing the task targets as a task planning scheme of the large unmanned aerial vehicle cluster.
Claims (8)
1. A task planning method based on a simplified attention network in a large unmanned aerial vehicle cluster is characterized in that the simplified attention network for extracting high-dimensional features of an unmanned aerial vehicle is constructed, the simplified attention network is trained by using a gradient descent method, and parameter updating of the whole network is realized; the planning method comprises the following steps:
step 1, generating a sample set:
generating a 10000 groups sample set; each group of sample sets comprises a large unmanned aerial vehicle cluster formed by at least 100 unmanned aerial vehicles and a task target executed by the large unmanned aerial vehicle cluster;
step 2, constructing a simplified attention network:
(2a) constructing a feature extraction network consisting of an input layer, an autocorrelation coefficient calculation layer, a cross correlation coefficient calculation layer and an output layer, setting the input dimensionality of the input layer to be 128, and setting the output dimensionality to be 256; setting the input dimensionalities of an autocorrelation system calculation layer, a cross correlation coefficient calculation layer and an output layer to be 256, and setting the output dimensionalities to be 256; the autocorrelation coefficient calculation layer is realized by an autocorrelation function, and the cross-correlation coefficient calculation layer is realized by a cross-correlation function;
(2b) setting parameters of a fully-connected neural network and an LSTM network;
setting the input dimension of the fully-connected neural network to be 8 and setting the output dimension to be 128; setting the input dimension of the LSTM network to be 256 and the output dimension to be 512;
(2c) cascading a feature extraction network, a fully-connected neural network and an LSTM (Long-Short Term Memory) network into a simplified attention network;
step 3, training a simplified attention network:
inputting the sample set into a simplified attention network, and iteratively updating the parameters of the simplified attention network by using a gradient descent method until a loss function value is converged to obtain a trained simplified attention network;
step 4, planning unmanned aerial vehicles for executing tasks in the large unmanned aerial vehicle cluster:
(4a) inputting the type and quantity of task resources carried by each unmanned aerial vehicle in a large unmanned aerial vehicle cluster to be planned and the type and quantity of task resources required by a task target into a trained simplified attention network, and outputting high-dimensional characteristics of each unmanned aerial vehicle;
(4b) calculating the probability of each unmanned aerial vehicle in the large-scale unmanned aerial vehicle cluster to execute the task target by using a probability formula, and selecting the unmanned aerial vehicle corresponding to the maximum probability as the unmanned aerial vehicle executing the task target;
(4c) inputting the types and the number of the task resources of the remaining unmanned aerial vehicles in the large unmanned aerial vehicle cluster to be planned into the trained simplified attention network to obtain the high-dimensional characteristics of each remaining unmanned aerial vehicle in the large unmanned aerial vehicle cluster to be planned, and selecting the unmanned aerial vehicle for executing the task by adopting the method same as the step (4 b);
(4d) judging whether the task resource types carried by all the unmanned aerial vehicles for executing the task target meet the constraint condition, if so, executing the step 5, otherwise, executing the step (4c)
And 5, taking the selected multiple unmanned aerial vehicles for executing the task targets as a planning scheme.
2. The method according to claim 1, wherein the large drone cluster comprising at least 100 drones in step 1 includes the number of drones in the large drone cluster, and the type and number of task resources carried by each drone.
3. The method for task planning in a large unmanned aerial vehicle cluster based on a simplified attention network as claimed in claim 1, wherein the task objectives executed by the large unmanned aerial vehicle cluster in step 1 include the types of resources required by the task objectives and the amount of resources required by the task objectives.
4. The method for reduced attention network based mission planning in a large fleet of unmanned aerial vehicles according to claim 1, wherein said autocorrelation function in step (2b) is as follows:
wherein λ is i An autocorrelation function, h, representing the ith training data in a reduced attention network i Initial characteristics, l, representing the ith training data q And l k Parameter representing a reduced attention network, d hi Denotes h i And Relu represents the linear rectification activation function operation, and N represents the total number of training data in each set of samples.
5. The method for reduced attention network based mission planning in a large fleet of unmanned aerial vehicles according to claim 4, wherein said cross-correlation function in step (2b) is as follows:
wherein, beta i Representing the autocorrelation function of the ith training data in the reduced attention network.
6. The method for reduced attention network based mission planning in a large fleet of unmanned aerial vehicles according to claim 1, wherein said penalty function in step 3 is as follows:
wherein F represents a loss function value, n represents the total number of the unmanned aerial vehicles executing the task, Σ represents summation operation, g represents the number of the unmanned aerial vehicles executing the task, and the value range of g is [1, n ]]S represents the total number of task resource categories required by the task object, m represents the number of task resource categories required by the task object, a g,m The number of the m-th task resources carried by the g-th task execution unmanned aerial vehicle is shown.
7. The method for reduced attention network based mission planning in a large fleet of unmanned aerial vehicles according to claim 6, wherein said probability formula in step (4b) is as follows:
wherein p is i Expressing the probability of the ith unmanned aerial vehicle in the large-scale unmanned aerial vehicle cluster to be planned to execute the task target, Exp expressing exponential operation with a natural constant e as the base, H i And the high-dimensional characteristics of the ith frame in the large unmanned aerial vehicle cluster to be planned are shown.
8. The method for simplified attention network based mission planning in a large fleet of unmanned aerial vehicles according to claim 1, wherein said constraint in step (4d) is a condition that satisfies the following conditions simultaneously:
the method comprises the following steps that 1, the total amount of task resources carried by a plurality of unmanned aerial vehicles executing task targets is selected to be larger than the total amount of task resources required by the task targets;
and 2, selecting the task resource types carried by the multiple unmanned aerial vehicles executing the task targets to be equal to the task resource types required by the task targets.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111006869.3A CN113703482B (en) | 2021-08-30 | 2021-08-30 | Task planning method based on simplified attention network in large-scale unmanned aerial vehicle cluster |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111006869.3A CN113703482B (en) | 2021-08-30 | 2021-08-30 | Task planning method based on simplified attention network in large-scale unmanned aerial vehicle cluster |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113703482A CN113703482A (en) | 2021-11-26 |
CN113703482B true CN113703482B (en) | 2022-08-12 |
Family
ID=78657012
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111006869.3A Active CN113703482B (en) | 2021-08-30 | 2021-08-30 | Task planning method based on simplified attention network in large-scale unmanned aerial vehicle cluster |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113703482B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115507852B (en) * | 2022-09-07 | 2023-11-03 | 广东工业大学 | Multi-unmanned aerial vehicle path planning method based on blockchain and enhanced attention learning |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AU2020101435A4 (en) * | 2020-07-21 | 2020-08-27 | Southwest University | A panoramic vision system based on the uav platform |
CN112256056A (en) * | 2020-10-19 | 2021-01-22 | 中山大学 | Unmanned aerial vehicle control method and system based on multi-agent deep reinforcement learning |
CN113286275A (en) * | 2021-04-23 | 2021-08-20 | 南京大学 | Unmanned aerial vehicle cluster efficient communication method based on multi-agent reinforcement learning |
-
2021
- 2021-08-30 CN CN202111006869.3A patent/CN113703482B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AU2020101435A4 (en) * | 2020-07-21 | 2020-08-27 | Southwest University | A panoramic vision system based on the uav platform |
CN112256056A (en) * | 2020-10-19 | 2021-01-22 | 中山大学 | Unmanned aerial vehicle control method and system based on multi-agent deep reinforcement learning |
CN113286275A (en) * | 2021-04-23 | 2021-08-20 | 南京大学 | Unmanned aerial vehicle cluster efficient communication method based on multi-agent reinforcement learning |
Non-Patent Citations (5)
Title |
---|
Image Deraining for UAV USING Split Attention Based Recursive Network;FENG Yidan.etc;《Transactions of Nanjing University of Aeronautics and Astronautics》;20200815;全文 * |
基于深度强化学习的群体对抗策略研究;刘强等;《智能计算机与应用》;20200501(第05期);全文 * |
面向体系仿真的智能无人机集群作战建模总体框架研究;邹立岩.etc;《军事运筹与系统工程》;20210330;全文 * |
面向多机协同的Att-MADDPG围捕控制方法设计;刘峰等;《空军工程大学学报(自然科学版)》;20210625;全文 * |
面向无人机的轻量级Siamese注意力网络目标跟踪;崔洲涓等;《光学学报》;20200707;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN113703482A (en) | 2021-11-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111537945B (en) | Intelligent ammeter fault diagnosis method and equipment based on federal learning | |
CN109508812A (en) | A kind of aircraft Trajectory Prediction method based on profound memory network | |
CN112200391B (en) | Power distribution network edge side load prediction method based on k-nearest neighbor mutual information feature simplification | |
CN112685504B (en) | Production process-oriented distributed migration chart learning method | |
CN111897353A (en) | Aircraft maneuvering trajectory prediction method based on GRU | |
CN109255726A (en) | A kind of ultra-short term wind power prediction method of Hybrid Intelligent Technology | |
CN113485409B (en) | Geographic fairness-oriented unmanned aerial vehicle path planning and distribution method and system | |
CN113703482B (en) | Task planning method based on simplified attention network in large-scale unmanned aerial vehicle cluster | |
CN113706151A (en) | Data processing method and device, computer equipment and storage medium | |
CN113902131B (en) | Updating method of node model for resisting discrimination propagation in federal learning | |
CN109144099B (en) | Fast evaluation method for unmanned aerial vehicle group action scheme based on convolutional neural network | |
CN112766603A (en) | Traffic flow prediction method, system, computer device and storage medium | |
CN107194415A (en) | Peak clustering method based on Laplace centrality | |
CN114897155A (en) | Integrated model data-free compression method for satellite | |
CN116095089A (en) | Remote sensing satellite data processing method and system | |
CN116226689A (en) | Power distribution network typical operation scene generation method based on Gaussian mixture model | |
CN113312177A (en) | Wireless edge computing system and optimization method based on federal learning | |
CN116911459A (en) | Multi-input multi-output ultra-short-term power load prediction method suitable for virtual power plant | |
Wang et al. | Coupled alternating neural networks for solving multi-population high-dimensional mean-field games with stochasticity | |
CN116244484B (en) | Federal cross-modal retrieval method and system for unbalanced data | |
CN116957304B (en) | Unmanned aerial vehicle group collaborative task allocation method and system | |
CN111783688B (en) | Remote sensing image scene classification method based on convolutional neural network | |
CN116450704A (en) | Automatic generation method and generation device of flow model | |
CN113919188B (en) | Relay unmanned aerial vehicle path planning method based on context-MAB | |
CN110163249B (en) | Base station classification identification method and system based on user parameter characteristics |
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 |