CN116860002B - Unmanned aerial vehicle cluster task resource scheduling method based on flow network model - Google Patents
Unmanned aerial vehicle cluster task resource scheduling method based on flow network model Download PDFInfo
- Publication number
- CN116860002B CN116860002B CN202310954303.6A CN202310954303A CN116860002B CN 116860002 B CN116860002 B CN 116860002B CN 202310954303 A CN202310954303 A CN 202310954303A CN 116860002 B CN116860002 B CN 116860002B
- Authority
- CN
- China
- Prior art keywords
- task
- nodes
- node
- network
- cluster
- 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 46
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 42
- 241000257303 Hymenoptera Species 0.000 claims description 33
- 230000008447 perception Effects 0.000 claims description 17
- 230000006870 function Effects 0.000 claims description 16
- 239000003016 pheromone Substances 0.000 claims description 16
- 238000013507 mapping Methods 0.000 claims description 15
- 238000005457 optimization Methods 0.000 claims description 12
- 238000004891 communication Methods 0.000 claims description 11
- 238000012546 transfer Methods 0.000 claims description 11
- 238000009826 distribution Methods 0.000 claims description 10
- 230000007704 transition Effects 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 7
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 6
- 238000012804 iterative process Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 4
- 230000002195 synergetic effect Effects 0.000 claims description 4
- 238000003860 storage Methods 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 14
- 238000010586 diagram Methods 0.000 description 3
- 230000008520 organization Effects 0.000 description 3
- 125000002015 acyclic group Chemical group 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008278 dynamic mechanism Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
Classifications
-
- 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 an unmanned aerial vehicle cluster task resource scheduling method based on a flow network model. And secondly, introducing a concept of task capability, endowing the task capability to a task layer network node as a node capability attribute, and converting the node capability attribute into a task link flow attribute. And in the unmanned plane cluster system level, the cluster is subjected to streaming processing on the completion condition of the task, and the streaming flow in the task layer network is the capacity flow of the cluster for completing the task. And finally, on the basis of effectively evaluating the cluster task capacity, an ant colony algorithm is improved to realize searching global optimum and fast convergence, and an optimum cluster resource scheduling strategy is realized.
Description
Technical Field
The invention belongs to the technical field of unmanned aerial vehicle clusters, and particularly relates to an unmanned aerial vehicle cluster task resource scheduling method based on a flow network model.
Background
Along with the development of unmanned aerial vehicle system autonomy, networking communication technology and group intelligent theory, unmanned aerial vehicle cluster appears to make up for the defect of stand-alone in aspects such as perception scope, survivability and environmental suitability. The unmanned aerial vehicle cluster performs tasks in a cooperative mode of a plurality of unmanned aerial vehicles, so that the single-machine capacity is improved, the task capacity of the cluster system is expanded, and the unmanned aerial vehicle cluster is widely applied to the fields of traffic, geological detection, rescue and the like. Meanwhile, the wide application field and the complex task scene also provide higher requirements on the task reliability level of the unmanned aerial vehicle cluster.
The task reliability of the current unmanned aerial vehicle cluster focuses on the design of the reliability of single machines, so the method for improving the task reliability of the cluster is to improve the task reliability of each single machine. However, the task reliability of such a system of unmanned aerial vehicle clusters is not only related to the reliability of the single machine, but also to the organization and cooperation between different single machines. Under different complex task environments, the task reliability level of the unmanned aerial vehicle cluster can be greatly different due to different organizations and resource utilization degrees of the task process. For example, in the process of executing tasks, the task reliability can be affected by the task sequence and the unmanned aerial vehicle distribution ratio when executing different tasks. Therefore, research is required to be conducted from the aspects of optimal resource scheduling strategies and the like so as to improve the task reliability level.
Due to the importance of unmanned aerial vehicle task allocation and resource scheduling and the nonlinearity of the calculation method thereof, intelligent algorithms are mainly used for optimizing in the current research. For example, sabit Poudel has comprehensively reviewed the main ideas, operational features, advantages and limitations of several task allocation algorithms designed for unmanned aerial vehicle networks to date, and compared the significant features and performance factors of these task allocation algorithms; the Duo Zheng converts the collaborative attack problem into two sub-problems of multi-objective task allocation and collaborative trajectory optimization, and allocates unmanned aerial vehicle tasks by using a multi-objective allocation algorithm based on a search path. Among various optimization algorithms, the ant colony algorithm belongs to one of widely used algorithms. Aiming at the problems of cloud computing resource allocation and scheduling efficiency, the Hongji Liu designs a cloud computing self-adaptive task scheduling algorithm based on an ant colony algorithm, and effectively shortens the optimizing time; lizhi Chen expresses the heterogeneous unmanned aerial vehicle collaborative task allocation problem as a constrained multi-objective optimization problem containing three optimization objectives, and solves the problem by using a multi-ant colony optimization algorithm.
The task reliability level of the unmanned aerial vehicle is often represented by evaluating the efficiency of the cluster, and because the complex network has good model description capability, in order to better evaluate the efficiency of the unmanned aerial vehicle cluster, the complex interaction and dynamic relationship in the unmanned aerial vehicle cluster system are researched, and a plurality of expert students introduce complex network theory, for example, lin Bingxuan and the like by means of the complex network theory, through constructing a complex network topology structure of the unmanned aerial vehicle of the bee colony, typical statistical characteristic parameters such as node degree, average path length, clustering coefficient and the like and anti-damage characteristics are analyzed, so that the purpose of evaluating the efficiency of the unmanned aerial vehicle of the bee colony is achieved; wang Ershen et al construct a cooperative network, a countermeasure network and a cooperative countermeasure network of both sides of the unmanned aerial vehicle cluster by applying a complex space network theory, and verify the effectiveness of the model; wang et al effectively evaluate unmanned aerial vehicle clusters by constructing a multi-layer complex network model according to cluster system composition and functions. Therefore, the complex network theory can have better modeling description capability for the unmanned aerial vehicle cluster. Meanwhile, in order to quantify the task capacity of the unmanned aerial vehicle under different decisions of the execution cluster, attribute values are required to be added on the basis of a complex network theory, so that a flow network model which is more in line with actual conditions is obtained. At present, a plurality of scholars combine related theories in other fields, for example, li Huahua utilizes the structural characteristics of a complex network theory task process network to obtain a process network with a non-scale characteristic, takes a process as a node, judges that a process resource conflict is a side, and establishes a 0-1 flow network model; the Hossain directly maps the Australian aviation network into a complex network model and analyzes the model, and establishes a flow network model by taking the passenger flow as an attribute value.
Disclosure of Invention
The present invention has been made to solve the above-mentioned problems occurring in the prior art. Therefore, an unmanned aerial vehicle cluster task resource scheduling method based on a flow network model is needed, in order to solve the problem that the task reliability of an unmanned aerial vehicle cluster is improved by improving the task reliability of a single machine and neglecting the organization cooperation relationship among different single machines at present, the thought of resource scheduling is mainly provided, and the task capacity of the unmanned aerial vehicle cluster is quantitatively analyzed by combining technical methods such as resource pooling, multi-layer complex network modeling, flow network model, intelligent algorithm and the like based on a complex network and a flow network theory, so that the aim of improving the task reliability is fulfilled.
According to a first technical scheme of the invention, an unmanned aerial vehicle cluster task resource scheduling method based on a flow network model is provided, and the method comprises the following steps: constructing a task network resource pool; constructing a task link; constructing a task layer network; establishing a flow network model considering task capacity flow; and optimizing the unmanned aerial vehicle cluster task resource scheduling by utilizing an ant colony algorithm.
Further, the constructing the task network resource pool includes:
and providing a unified mapping platform for various unmanned aerial vehicles, namely a task layer network, mapping physical equipment entities into the task layer network, abstracting each unmanned aerial vehicle entity into task load nodes expressed in a complex network, finally clustering the task load nodes according to node types, and virtualizing each cluster into a resource pool.
Further, unmanned aerial vehicles performing different functions in the unmanned aerial vehicle cluster system are classified into 3 types: a perception class, a decision class, and an execution class;
grouping the cluster system into three layers of networks G= { G 1 ,G 2 ,G 3 ' respectively the communication layer network G 1 Structural layer network G 2 And task layer network G 3 The corresponding unmanned aerial vehicle cluster forms a communication data chain, an unmanned aerial vehicle and a task load;
for the structural layer network G 2 According to the structural layer network G 2 Network G carrying task load category pair structural layer 2 Classifying and abstracting task load into a task layer network G 3 Is a node in (a);
based on OODA task node division, on task layer network G 3 Three kinds of resource pools are built, namely a class resource pool, a decision class resource pool and an execution class resource pool are perceived respectively, resource division is carried out according to task load nodes mapped in the class resource pools, the same kind of task loads are matched with the corresponding resource pools, and a clustered task layer network of the resource pools is built.
4. The method of claim 1, wherein the constructing a task link comprises:
there are 4 classes of contiguous relations in the generalized OODA model that consider the synergistic relationship: a sense intelligence sharing link (S-S), an intelligence uploading link (S-D), a command cooperative link (D-D) and a command down link (D-I);
based on the 4-class continuous edge relationship, abstract extraction is carried out on the continuous edge relationship of the cluster task load:
1) Initially, the method comprisesAnd (3) initializing: determining the node number and position distribution of a task layer network according to cluster load configuration, and assuming that n nodes are generated for the task layer network, wherein the number of task load types corresponds to the number of upper node resource pools and is 3, and the number of nodes under each load is n i (i=1, 2, 3), where n 1 +n 2 +n 3 ＝n；
2) And (3) connection:
a. the connection sequence of the task loads of various types is defined, and the following three types are adopted: s- & gt D- & gt I, S- & gt S- & gt D- & gt I and S- & gt D- & gt I;
b. the load numbers corresponding to the three kinds of task loads which are connected according to the free combination algorithm are respectively n 1 *n 2 *n 3 Totally n 1 *n 2 *n 3 A connectable mode;
c. the same kind of task loads are connected according to a free combination algorithm in the perception type resource pool and the decision type resource pool and correspond to the cooperative working mode of the cluster, and the load numbers corresponding to the two kinds of task loads are respectively n 1 、n 2 Respectively having n 1 ！、n 2 The following is carried out A connectable mode;
d. in all connections, heavy edges and self-loops cannot be arranged between two nodes;
3) Ending: and outputting the constructed task layer link after all nodes and edges in the network are generated.
5. The method of claim 4, wherein at task layer network G 3 The middle node has the task capability corresponding to the task load as the attribute value, namely C 3 ＝{C 31 ,C 32 ,…,C 3n And n is the number of nodes, and constructing a task layer network by the following method:
node V based on cluster organism layer network 2 Mapping is carried out, and task load is abstracted into a node V of a task layer network 3 And aggregate based on the idea of resource pooling;
the task layer network nodes are connected according to a task link construction method;
and the task capacity corresponding to the task load is given to the corresponding node.
Further, the establishing a flow network model considering task capability flow includes:
the task layer network is equivalent to a stream network G '= { V', E ', F' }, and is a communicating acyclic directed graph. Where V 'is the set of nodes of the network, E' = { E 1 ,e 2 ,…,e n And F' = { F) is the edge set of the network 1 ,f 2 ,…,f n -a flow function of the network, the flow function being a non-negative value defined on the edge set;
the flow of the capacity in the cluster network depends on the topological relation among task load nodes, the capacity flows from a sensing node to a decision node and then flows to an execution node, the capacities among different nodes flow simultaneously to form the task capacity flow of the whole cluster together, and f i (1.ltoreq.i.ltoreq.n) corresponds to the corresponding link e i Task capability flow on (1.ltoreq.i.ltoreq.n).
Further, the optimizing the unmanned aerial vehicle cluster task resource scheduling by using the ant colony algorithm includes:
the initial position distribution of ants is set to be all nodes with the ingress degree of 0, namely perception class nodes, analysis is carried out based on a task laminar flow network, corresponding network capacity flow is selected as node heuristic information, and a node transfer formula is established:
η i ＝[suc(V i )+1]·V i
wherein V is i Representing the task capabilities corresponding to the nodes, the sub (V i ) The number of subsequent nodes for node i;
in the iterative process of the algorithm, the node types to be traversed in the task link are divided into three types by combining the network characteristics of the task layer of the cluster and the task mode, and the three types of corresponding resource pools are stored in three types of sets for node traversal optimizing selection:
the sensing class nodes, namely nodes with the degree of entry of 0, are stored in a set T1;
the target node, namely the node with the output degree of 0, is stored in a set T2;
the rest nodes are decision class and execution class nodes and are stored in a set T3;
when the ants select nodes, the method comprises three steps:
1) When an ant transitions from a node in set T1 to a node in set T3 or set T1, the node selection probability is as follows:
wherein τ ij (t) represents the corresponding path pheromone value, eta from i to j at the corresponding time ij For heuristic information of corresponding paths from i to j, the influence degree of alpha and beta reaction pheromones and heuristic information on the selected paths of ants is called pheromones and heuristic factors, and T represents the paths which the ants do not pass through yet;
2) When the ants are transferred from the nodes in the set T3 to the nodes in the set T3, calculating according to the transfer probability;
3) When ants are transferred from the nodes in the set T3 to the nodes in the set T2, a nearby execution strategy is formulated according to the distance between the nodes and the execution range element parameters;
when ants are all at nodes in set T2, the traversal ends.
According to a second technical scheme of the invention, an unmanned aerial vehicle cluster task resource scheduling device based on a flow network model is provided, and the device comprises; a first building module configured to build a task network resource pool; a second construction module configured to construct a task link; a third building module configured to build a task layer network; a fourth building module configured to build a flow network model that accounts for task capability flows; and the optimization module is configured to optimize unmanned aerial vehicle cluster task resource scheduling by utilizing an ant colony algorithm.
Further, the first build module is further configured to:
and providing a unified mapping platform for various unmanned aerial vehicles, namely a task layer network, mapping physical equipment entities into the task layer network, abstracting each unmanned aerial vehicle entity into task load nodes expressed in a complex network, finally clustering the task load nodes according to node types, and virtualizing each cluster into a resource pool.
According to a third aspect of the present invention, there is provided a readable storage medium storing one or more programs executable by one or more processors to implement the method as described above.
The invention has at least the following beneficial effects:
(1) The method is characterized in that a cluster analysis method based on a task process is effectively combined with a complex network theory, cluster macro system features and unmanned aerial vehicle individual task capability factors are fully fused, a cluster flow network model is constructed, a cluster task capability assessment method is provided based on the model and a networked cluster task capability index system, and the problem that the traditional assessment method ignores the influence of an organization scheduling relationship in an unmanned aerial vehicle cluster on task reliability is solved;
(2) The intelligent optimization algorithm is improved based on the cluster network model, the state transition rule and the parallel strategy of the algorithm are optimized for the complex and changeable task process and the complex and dynamic mechanism characteristics of the cluster, the rapid optimizing iteration of the task strategy is realized, the optimal task scheduling strategy suitable for the task scene can be rapidly found, the cluster task capacity is exerted to the maximum extent, and the task loss is reduced.
Drawings
FIG. 1 illustrates a clustered resource pooling schematic according to an embodiment of the invention;
FIG. 2 illustrates a task layer network link schematic according to an embodiment of the present invention;
FIG. 3 illustrates a task network traffic schematic according to an embodiment of the present invention;
FIG. 4 shows a resource scheduling model schematic according to an embodiment of the invention;
FIG. 5 illustrates a parallel policy diagram incorporating a task layer network in accordance with an embodiment of the present invention;
FIG. 6 illustrates a modified algorithm flow diagram according to an embodiment of the invention;
fig. 7 shows a block diagram of an unmanned aerial vehicle cluster task resource scheduling device based on a flow network model according to an embodiment of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the drawings and detailed description to enable those skilled in the art to better understand the technical scheme of the present invention. Embodiments of the present invention will be described in further detail below with reference to the drawings and specific examples, but not by way of limitation. The order in which the steps are described herein by way of example should not be construed as limiting if there is no necessity for a relationship between each other, and it should be understood by those skilled in the art that the steps may be sequentially modified without disrupting the logic of each other so that the overall process is not realized.
The embodiment of the invention provides an unmanned aerial vehicle cluster task resource scheduling method based on a flow network model, which comprises the following steps:
step 1: and constructing a task network resource pool.
The unmanned aerial vehicle cluster resource pool is an aggregation scheduling system which is used for abstracting and virtualizing task load entities. Firstly, a unified mapping platform, namely a task layer network, is required to be provided for various unmanned aerial vehicle devices, physical device entities are mapped to access the task layer network, all unmanned aerial vehicle entities are abstracted into task load nodes which can be represented in a complex network, and finally, the task load nodes are clustered according to node types and all clusters are virtualized into a resource pool.
By means of the task process OODA analysis, we divide the unmanned aerial vehicles in the unmanned aerial vehicle cluster system that perform different functions into 3 classes: a perception class (S), a decision class (D) and an execution class (I). According to the composition and the functions of the cluster system, the cluster system is divided into three layers of networks G= { G 1 ,G 2 ,G 3 ' respectively the communication layer network G 1 Structural layer network G 2 And task layer network G 3 And the corresponding unmanned aerial vehicle cluster forms a communication data chain, an unmanned aerial vehicle and a task load.
Thus, for unmanned aerial vehicle cluster body layer G 2 In other words, can carry task loads according to the task loadClassifying the task load into task layers G by category 3 Is included in the node (a). Meanwhile, three types of resource pools, namely a perception type resource pool, a decision type resource pool and an execution type resource pool, can be constructed in a task layer based on OODA task node division, and the resource division is carried out according to the task load nodes mapped in the task layer, the same type of task load is matched with the corresponding resource pool, and a clustered task layer network with the resource pool is constructed.
Step 2: and constructing a task link.
For the standard OODA model, there are 2 classes of connection relations: an information uploading link (S-D) and an instruction down link (D-I); there are 4 classes of edges in the generalized OODA model that consider the synergistic relationship: the system comprises a sensing information sharing link (S-S), an information uploading link (S-D), a command cooperative link (D-D) and a command down link (D-I).
Based on the 4 kinds of continuous edge relations, abstract extraction is carried out on the continuous edge relations of the cluster task load, and the continuous edge construction rule is as follows:
1. initializing: determining the node number and position distribution of a task layer network according to cluster load configuration, and assuming that n nodes are generated for the task layer network, wherein the number of task load types corresponds to the number of upper node resource pools and is 3, and the number of nodes under each load is n i (i=1, 2, 3), where n 1 +n 2 +n 3 ＝n。
2. And (3) connection:
a. the connection sequence of the task loads of various types is defined, and the following three types are adopted: s- & gt D- & gt I, S- & gt S- & gt D- & gt I and S- & gt D- & gt I;
b. different task resource pools, namely different kinds of task loads are connected according to a free combination algorithm, namely the load numbers corresponding to the three kinds of task loads are respectively n 1 *n 2 *n 3 Then there is n in total 1 *n 2 *n 3 A connectable mode;
c. and the same kind of task loads are also connected according to a free combination algorithm in the perception class resource pool and the decision class resource pool, and the cooperative work modes of the clusters are corresponding. The load numbers corresponding to the two task loads are respectively n 1 、n 2 Then respectively have n 1 ！、n 2 The following is carried out A connectable mode;
d. all connections require that there be no heavy edges between the two nodes, nor no self-loops.
3. Ending: when all nodes and edges in the network are generated, the constructed task layer link model is output as shown in fig. 2, wherein the broken line represents the connectable task link.
Step 3: and constructing a task layer network.
The unmanned aerial vehicle load plays a role in the task process, namely the performance of the task capacity, so that the unmanned aerial vehicle load is in the task layer network model G 3 The middle node needs to have the task capability corresponding to the task load as the attribute value, namely C 3 ＝{C 31 ,C 32 ,…,C 3n And n is the number of nodes.
Specifically, the task layer network model building process can be described as follows:
1. node V based on cluster organism layer network 2 Mapping is carried out, and task load is abstracted into a node V of a task layer network 3 And aggregate based on the idea of resource pooling;
2. the task layer network nodes are connected according to the construction rule of the task links;
3. and the task capacity corresponding to the task load is given to the corresponding node.
Step 4: a flow network model is built that considers the task capability flow.
In the actual task process, the capacity exertion of each node in the task link not only depends on the capacity value of the node, but also is limited by the capacities of other nodes in the same link, so that the capacity exertion of the node cannot be completely realized. For a directed capacity network, it is generally subjected to a streaming process, i.e., converting the node capability attribute into a task link traffic attribute. And in the unmanned plane cluster system level, the cluster is subjected to streaming processing on the completion condition of the task, and the streaming flow in the task layer network is the capacity flow of the cluster for completing the task.
Thus, the task layer network model can be equivalently a flow network G '= { V', E ', F' }, which is a directed graph of connected loop-free.Where V 'is the set of nodes of the network, E' = { E 1 ,e 2 ,…,e n And F' = { F) is the edge set of the network 1 ,f 2 ,…,f n And is the traffic function of the network, which is a non-negative value defined on the edge set. The flow of the capacity in the cluster network depends on the topological relation among task load nodes, the capacity flows from a sensing node to a decision node and then flows to an execution node, the capacities among different nodes flow simultaneously to form the task capacity flow of the whole cluster together, and f i (1.ltoreq.i.ltoreq.n) corresponds to the corresponding link e i The task capacity flow on (1.ltoreq.i.ltoreq.n) is shown in FIG. 3.
Step 5: and optimizing the unmanned aerial vehicle clustering capacity resource scheduling method by using an ant colony algorithm.
Corresponding to a real task scene, task nodes are added on the basis of a task laminar flow network to form a complete OODA task link consisting of perception, decision, execution and target, as shown in fig. 4, wherein the connection mode of the task link determines the network structure and the overall task flow condition of the unmanned aerial vehicle cluster, and is also the network expression form of a cluster resource scheduling strategy. In addition, according to the condition that the unmanned aerial vehicle perception system can detect under the actual condition, the task capacity corresponding to the target node is regarded as two parts to constitute: the execution distance and the emergency level represent the distance of the target node from the carrying execution load unmanned aerial vehicle and the emergency degree of the target node in the whole task respectively.
According to the ant colony algorithm, the selection of the path is equivalent to the selection of each task chain in the unmanned aerial vehicle cluster task layer network, and the state transition probability of the algorithm is the probability of selecting the next node, so that the resource scheduling problem is converted into the traversing problem of the network nodes. And considering the capability characteristics in the network we consider optimizing the transfer rules based on the capability flow relationships in the task links.
Firstly, the initial position distribution of ants is set to be all nodes with the ingress degree of 0, namely perception nodes, analysis is carried out based on a task laminar flow network, corresponding network capacity flow is selected as node heuristic information, and a node transfer formula is established:
η i ＝[suc(V i )+1]·V i
wherein V is i And representing the task capacity corresponding to the node. Suc (V) i ) Is the number of subsequent nodes of node i. Suc (V) i ) Modified as [ suc (V) i )+1]When no subsequent node exists, the phenomenon that the node transition probability is 0 and the next iteration transition can not be started is avoided.
In the iterative process of the algorithm, the node types to be traversed in the task link are divided into three types by combining the network characteristics of the task layer of the cluster and the task mode, and the three types of corresponding resource pools are stored in three types of sets for node traversal optimizing selection:
1. the sensing class nodes, namely nodes with the degree of entry of 0, are stored in a set T1;
2. the target node, namely the node with the output degree of 0, is stored in a set T2;
3. the rest nodes are decision class and execution class nodes and are stored in a set T3.
Therefore, when the ants perform node selection, three steps are also divided:
1. when an ant transitions from a node in set T1 to a node in set T3 or set T1, the node selection probability is as follows:
wherein τ ij (t) represents the corresponding path pheromone value, eta from i to j at the corresponding time ij For heuristic information of corresponding paths from i to j, the influence degree of alpha and beta reaction pheromones and heuristic information on the selected paths of ants is called pheromones and heuristic factors, and T represents the paths which the ants do not pass through yet;
2. when the ants are transferred from the nodes in the set T3 to the nodes in the set T3, calculating according to the transfer probability;
3. when the ants are transferred from the nodes in the set T3 to the nodes in the set T2, a nearby execution strategy is formulated according to the distance between the nodes and the execution range element parameters.
When ants are all at nodes in set T2, the traversal ends. Traversing the unmanned aerial vehicle cluster task layer network model through an ant colony algorithm, calculating the task reliability according to the link connection relation at the moment, evaluating by taking the task reliability as an objective function, and judging the advantages and disadvantages of the iteration result.
The ant colony algorithm can be divided into a plurality of groups due to the parallelism of the ant colony algorithm, meanwhile, a model is optimized and solved, different ant groups are not mutually interfered, and information interaction is carried out through updating of pheromones, so that the algorithm operation speed is improved.
The embodiment of the invention also provides an unmanned aerial vehicle cluster task resource scheduling device based on the flow network model, as shown in fig. 7, the device 700 comprises;
a first building module 701 configured to build a task network resource pool;
a second building block 702 configured to build a task link;
a third building module 703 configured to build a task layer network;
a fourth building module 704 configured to build a flow network model that accounts for task capability flows;
and the optimization module 705 is configured to optimize unmanned aerial vehicle cluster task resource scheduling by utilizing an ant colony algorithm.
In some embodiments, the first build module is further configured to:
and providing a unified mapping platform for various unmanned aerial vehicles, namely a task layer network, mapping physical equipment entities into the task layer network, abstracting each unmanned aerial vehicle entity into task load nodes expressed in a complex network, finally clustering the task load nodes according to node types, and virtualizing each cluster into a resource pool.
In some embodiments, the first build module is further configured to:
unmanned aerial vehicles performing different functions in an unmanned aerial vehicle cluster system are classified into 3 types: a perception class, a decision class, and an execution class;
grouping the cluster system into three layers of networks G= { G 1 ,G 2 ,G 3 ' respectively the communication layer network G 1 Structural layer network G 2 And task layer network G 3 The corresponding unmanned aerial vehicle cluster forms a communication data chain, an unmanned aerial vehicle and a task load;
for the structural layer network G 2 According to the structural layer network G 2 Network G carrying task load category pair structural layer 2 Classifying and abstracting task load into a task layer network G 3 Is a node in (a);
based on OODA task node division, on task layer network G 3 Three kinds of resource pools are built, namely a class resource pool, a decision class resource pool and an execution class resource pool are perceived respectively, resource division is carried out according to task load nodes mapped in the class resource pools, the same kind of task loads are matched with the corresponding resource pools, and a clustered task layer network of the resource pools is built.
In some embodiments, the second build module is further configured to:
there are 4 classes of contiguous relations in the generalized OODA model that consider the synergistic relationship: a sense intelligence sharing link (S-S), an intelligence uploading link (S-D), a command cooperative link (D-D) and a command down link (D-I);
based on the 4-class continuous edge relationship, abstract extraction is carried out on the continuous edge relationship of the cluster task load:
1) Initializing: determining the node number and position distribution of a task layer network according to cluster load configuration, and assuming that n nodes are generated for the task layer network, wherein the number of task load types corresponds to the number of upper node resource pools and is 3, and the number of nodes under each load is n i (i=1, 2, 3), where n 1 +n 2 +n 3 ＝n；
2) And (3) connection:
a. the connection sequence of the task loads of various types is defined, and the following three types are adopted: s- & gt D- & gt I, S- & gt S- & gt D- & gt I and S- & gt D- & gt I;
b. the load numbers corresponding to the three kinds of task loads which are connected according to the free combination algorithm are respectively n 1 *n 2 *n 3 Totally n 1 *n 2 *n 3 A connectable mode;
c. the same kind of task loads are connected according to a free combination algorithm in the perception type resource pool and the decision type resource pool and correspond to the cooperative working mode of the cluster, and the load numbers corresponding to the two kinds of task loads are respectively n 1 、n 2 Respectively having n 1 ！、n 2 The following is carried out A connectable mode;
d. in all connections, heavy edges and self-loops cannot be arranged between two nodes;
3) Ending: and outputting the constructed task layer link after all nodes and edges in the network are generated.
In some embodiments, at task layer network G 3 The middle node has the task capability corresponding to the task load as the attribute value, namely C 3 ＝{C 31 ,C 32 ,…,C 3n -n is the number of nodes, the third building block being further configured to:
node V based on cluster organism layer network 2 Mapping is carried out, and task load is abstracted into a node V of a task layer network 3 And aggregate based on the idea of resource pooling;
the task layer network nodes are connected according to a task link construction method;
and the task capacity corresponding to the task load is given to the corresponding node.
In some embodiments, the fourth building module is further configured to:
the task layer network is equivalent to a stream network G '= { V', E ', F' }, and is a communicating acyclic directed graph. Where V 'is the set of nodes of the network, E' = { E 1 ,e 2 ,…,e n And F' = { F) is the edge set of the network 1 ,f 2 ,…,f n -a flow function of the network, the flow function being a non-negative value defined on the edge set;
the flow of the capability in the cluster network depends on the topological relation among task load nodes, the capability flows from the sensing node to the decision node and then flows to the execution node, and the capability among different nodesSimultaneously flowing to jointly form a task capacity flow of the whole cluster, f i (1.ltoreq.i.ltoreq.n) corresponds to the corresponding link e i Task capability flow on (1.ltoreq.i.ltoreq.n).
In some embodiments, the optimization module is further configured to:
the initial position distribution of ants is set to be all nodes with the ingress degree of 0, namely perception class nodes, analysis is carried out based on a task laminar flow network, corresponding network capacity flow is selected as node heuristic information, and a node transfer formula is established:
η i ＝[suc(V i )+1]·V i
wherein V is i Representing the task capabilities corresponding to the nodes, the sub (V i ) The number of subsequent nodes for node i;
in the iterative process of the algorithm, the node types to be traversed in the task link are divided into three types by combining the network characteristics of the task layer of the cluster and the task mode, and the three types of corresponding resource pools are stored in three types of sets for node traversal optimizing selection:
the sensing class nodes, namely nodes with the degree of entry of 0, are stored in a set T1;
the target node, namely the node with the output degree of 0, is stored in a set T2;
the rest nodes are decision class and execution class nodes and are stored in a set T3;
when the ants select nodes, the method comprises three steps:
1) When an ant transitions from a node in set T1 to a node in set T3 or set T1, the node selection probability is as follows:
wherein τ ij (t) represents the corresponding path pheromone value, eta from i to j at the corresponding time ij For heuristic information of corresponding paths from i to j, alpha and beta reflect influence degree of pheromone and heuristic information on ant selection path, namely pheromone and heuristic factor, T represents antA path that has not yet been traversed;
2) When the ants are transferred from the nodes in the set T3 to the nodes in the set T3, calculating according to the transfer probability;
3) When ants are transferred from the nodes in the set T3 to the nodes in the set T2, a nearby execution strategy is formulated according to the distance between the nodes and the execution range element parameters;
when ants are all at nodes in set T2, the traversal ends.
It should be noted that, the apparatus described in this embodiment and the method described in the foregoing belong to the same technical idea, which can achieve the same technical effect, and are not described here again.
Furthermore, although exemplary embodiments have been described herein, the scope thereof includes any and all embodiments having equivalent elements, modifications, omissions, combinations (e.g., of the various embodiments across), adaptations or alterations as pertains to the present invention. Elements in the claims are to be construed broadly based on the language employed in the claims and are not limited to examples described in the present specification or during the practice of the present application, which examples are to be construed as non-exclusive. It is intended, therefore, that the specification and examples be considered as exemplary only, with a true scope and spirit being indicated by the following claims and their full scope of equivalents.
The above description is intended to be illustrative and not restrictive. For example, the above-described examples (or one or more aspects thereof) may be used in combination with each other. For example, other embodiments may be used by those of ordinary skill in the art upon reading the above description. In addition, in the above detailed description, various features may be grouped together to streamline the invention. This is not to be interpreted as an intention that the features of the claimed invention are essential to any of the claims. Rather, inventive subject matter may lie in less than all features of a particular inventive embodiment. Thus, the following claims are hereby incorporated into the detailed description as examples or embodiments, with each claim standing on its own as a separate embodiment, and it is contemplated that these embodiments may be combined with one another in various combinations or permutations. The scope of the invention should be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.
Claims (8)
1. The unmanned aerial vehicle cluster task resource scheduling method based on the flow network model is characterized by comprising the following steps of:
constructing a task network resource pool;
constructing a task link;
constructing a task layer network;
establishing a flow network model considering task capacity flow;
optimizing unmanned aerial vehicle cluster task resource scheduling by utilizing an ant colony algorithm;
the establishing a flow network model considering task capability flow comprises the following steps:
the task layer network is equivalent to a stream network G '= { V', E ', F' }, which is a directed graph with communication and no loop, wherein V 'is a node set of the network, and E' = { E 1 ,e 2 ,…,e n And F' = { F) is the edge set of the network 1 ,f 2 ,…,f n -a flow function of the network, the flow function being a non-negative value defined on the edge set;
the flow of the capacity in the cluster network depends on the topological relation among task load nodes, the capacity flows from a sensing node to a decision node and then flows to an execution node, the capacities among different nodes flow simultaneously to form the task capacity flow of the whole cluster together, and f i (1.ltoreq.i.ltoreq.n) corresponds to the corresponding link e i (1.ltoreq.i.ltoreq.n);
the optimization of unmanned aerial vehicle cluster task resource scheduling by using the ant colony algorithm comprises the following steps:
the initial position distribution of ants is set to be all nodes with the ingress degree of 0, namely perception class nodes, analysis is carried out based on a task laminar flow network, corresponding network capacity flow is selected as node heuristic information, and a node transfer formula is established:
η i ＝[suc(V i )+1]·V i
wherein V is i Representing the task capabilities corresponding to the node,suc(V i ) The number of subsequent nodes for node i;
in the iterative process of the algorithm, the node types to be traversed in the task link are divided into three types by combining the network characteristics of the task layer of the cluster and the task mode, and the three types of corresponding resource pools are stored in three types of sets for node traversal optimizing selection:
the sensing class nodes, namely nodes with the degree of entry of 0, are stored in a set T1;
the target node, namely the node with the output degree of 0, is stored in a set T2;
the rest nodes are decision class and execution class nodes and are stored in a set T3;
when the ants select nodes, the method comprises three steps:
1) When an ant transitions from a node in set T1 to a node in set T3 or set T1, the node selection probability is as follows:
wherein τ ij (t) represents the corresponding path pheromone value, eta from i to j at the corresponding time ij For heuristic information of corresponding paths from i to j, the influence degree of alpha and beta reaction pheromones and heuristic information on the selected paths of ants is called pheromones and heuristic factors, and T represents the paths which the ants do not pass through yet;
2) When the ants are transferred from the nodes in the set T3 to the nodes in the set T3, calculating according to the transfer probability;
3) When ants are transferred from the nodes in the set T3 to the nodes in the set T2, a nearby execution strategy is formulated according to the distance between the nodes and the execution range element parameters;
when ants are all at nodes in set T2, the traversal ends.
2. The method of claim 1, wherein the building a task network resource pool comprises:
and providing a unified mapping platform for various unmanned aerial vehicles, namely a task layer network, mapping physical equipment entities into the task layer network, abstracting each unmanned aerial vehicle entity into task load nodes expressed in a complex network, finally clustering the task load nodes according to node types, and virtualizing each cluster into a resource pool.
3. The method according to claim 2, characterized in that the unmanned aerial vehicles performing different functions in the unmanned aerial vehicle cluster system are classified into 3 classes: a perception class, a decision class, and an execution class;
grouping the cluster system into three layers of networks G= { G 1 ,G 2 ,G 3 ' respectively the communication layer network G 1 Structural layer network G 2 And task layer network G 3 The corresponding unmanned aerial vehicle cluster forms a communication data chain, an unmanned aerial vehicle and a task load;
for the structural layer network G 2 According to the structural layer network G 2 Network G carrying task load category pair structural layer 2 Classifying and abstracting task load into a task layer network G 3 Is a node in (a);
based on OODA task node division, on task layer network G 3 Three kinds of resource pools are built, namely a class resource pool, a decision class resource pool and an execution class resource pool are perceived respectively, resource division is carried out according to task load nodes mapped in the class resource pools, the same kind of task loads are matched with the corresponding resource pools, and a clustered task layer network of the resource pools is built.
4. The method of claim 1, wherein the constructing a task link comprises:
there are 4 classes of contiguous relations in the generalized OODA model that consider the synergistic relationship: a sense intelligence sharing link (S-S), an intelligence uploading link (S-D), a command cooperative link (D-D) and a command down link (D-I);
based on the 4-class continuous edge relationship, abstract extraction is carried out on the continuous edge relationship of the cluster task load:
1) Initializing: determining task layers from cluster load configurationThe node number and position distribution of the network are assumed that n nodes are generated for the task layer network, the number of task load types corresponds to the number of upper node resource pools, 3 types are adopted, and the number of nodes under each load is n i (i=1, 2, 3), where n 1 +n 2 +n 3 ＝n；
2) And (3) connection:
a. the connection sequence of the task loads of various types is defined, and the following three types are adopted: s- & gt D- & gt I, S- & gt S- & gt D- & gt I and S- & gt D- & gt I;
b. the load numbers corresponding to the three kinds of task loads which are connected according to the free combination algorithm are respectively n 1 *n 2 *n 3 Totally n 1 *n 2 *n 3 A connectable mode;
c. the same kind of task loads are connected according to a free combination algorithm in the perception type resource pool and the decision type resource pool and correspond to the cooperative working mode of the cluster, and the load numbers corresponding to the two kinds of task loads are respectively n 1 、n 2 Respectively having n 1 ！、n 2 The following is carried out A connectable mode;
d. in all connections, heavy edges and self-loops cannot be arranged between two nodes;
3) Ending: and outputting the constructed task layer link after all nodes and edges in the network are generated.
5. The method of claim 4, wherein at task layer network G 3 The middle node has the task capability corresponding to the task load as the attribute value, namely C 3 ＝{C 31 ,C 32 ,…,C 3n And n is the number of nodes, and constructing a task layer network by the following method:
node V based on cluster organism layer network 2 Mapping is carried out, and task load is abstracted into a node V of a task layer network 3 And aggregate based on the idea of resource pooling;
the task layer network nodes are connected according to a task link construction method;
and the task capacity corresponding to the task load is given to the corresponding node.
6. An unmanned aerial vehicle cluster task resource scheduling device based on a flow network model is characterized by comprising;
a first building module configured to build a task network resource pool;
a second construction module configured to construct a task link;
a third building module configured to build a task layer network;
a fourth building module configured to build a flow network model that accounts for task capability flows;
the optimization module is configured to optimize unmanned aerial vehicle cluster task resource scheduling by utilizing an ant colony algorithm;
the fourth building block is further configured to:
the task layer network is equivalent to a stream network G '= { V', E ', F' }, which is a directed graph with communication and no loop, wherein V 'is a node set of the network, and E' = { E 1 ,e 2 ,…,e n And F' = { F) is the edge set of the network 1 ,f 2 ,…,f n -a flow function of the network, the flow function being a non-negative value defined on the edge set;
the flow of the capacity in the cluster network depends on the topological relation among task load nodes, the capacity flows from a sensing node to a decision node and then flows to an execution node, the capacities among different nodes flow simultaneously to form the task capacity flow of the whole cluster together, and f i (1.ltoreq.i.ltoreq.n) corresponds to the corresponding link e i (1.ltoreq.i.ltoreq.n);
the optimization module is further configured to:
the initial position distribution of ants is set to be all nodes with the ingress degree of 0, namely perception class nodes, analysis is carried out based on a task laminar flow network, corresponding network capacity flow is selected as node heuristic information, and a node transfer formula is established:
η i ＝[suc(V i )+1]·V i
wherein V is i Representing the task capabilities corresponding to the nodes, the sub (V i ) For node iSubsequent node numbers of (a);
in the iterative process of the algorithm, the node types to be traversed in the task link are divided into three types by combining the network characteristics of the task layer of the cluster and the task mode, and the three types of corresponding resource pools are stored in three types of sets for node traversal optimizing selection:
the sensing class nodes, namely nodes with the degree of entry of 0, are stored in a set T1;
the target node, namely the node with the output degree of 0, is stored in a set T2;
the rest nodes are decision class and execution class nodes and are stored in a set T3;
when the ants select nodes, the method comprises three steps:
1) When an ant transitions from a node in set T1 to a node in set T3 or set T1, the node selection probability is as follows:
wherein τ ij (t) represents the corresponding path pheromone value, eta from i to j at the corresponding time ij For heuristic information of corresponding paths from i to j, the influence degree of alpha and beta reaction pheromones and heuristic information on the selected paths of ants is called pheromones and heuristic factors, and T represents the paths which the ants do not pass through yet;
2) When the ants are transferred from the nodes in the set T3 to the nodes in the set T3, calculating according to the transfer probability;
3) When ants are transferred from the nodes in the set T3 to the nodes in the set T2, a nearby execution strategy is formulated according to the distance between the nodes and the execution range element parameters;
when ants are all at nodes in set T2, the traversal ends.
7. The apparatus of claim 6, wherein the first build module is further configured to:
and providing a unified mapping platform for various unmanned aerial vehicles, namely a task layer network, mapping physical equipment entities into the task layer network, abstracting each unmanned aerial vehicle entity into task load nodes expressed in a complex network, finally clustering the task load nodes according to node types, and virtualizing each cluster into a resource pool.
8. A readable storage medium storing one or more programs executable by one or more processors to implement the method of any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310954303.6A CN116860002B (en) | 2023-08-01 | 2023-08-01 | Unmanned aerial vehicle cluster task resource scheduling method based on flow network model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310954303.6A CN116860002B (en) | 2023-08-01 | 2023-08-01 | Unmanned aerial vehicle cluster task resource scheduling method based on flow network model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116860002A CN116860002A (en) | 2023-10-10 |
| CN116860002B true CN116860002B (en) | 2023-12-22 |
Family
ID=88232262
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310954303.6A Active CN116860002B (en) | 2023-08-01 | 2023-08-01 | Unmanned aerial vehicle cluster task resource scheduling method based on flow network model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116860002B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017080406A1 (en) * | 2015-11-11 | 2017-05-18 | 刘晓阳 | Mooring unmanned rotorcraft cluster platform system and liquid continuous spraying system |
| CN113759975A (en) * | 2021-09-15 | 2021-12-07 | 北京航空航天大学 | Task capability-based unmanned aerial vehicle cluster modeling method and model system |
| CN114281104A (en) * | 2021-12-16 | 2022-04-05 | 成都戎星科技有限公司 | Multi-unmanned-aerial-vehicle cooperative regulation and control method based on improved ant colony algorithm |
| CN114995521A (en) * | 2022-08-08 | 2022-09-02 | 中国科学院自动化研究所 | Multi-unmanned aerial vehicle distributed formation control method and device and electronic equipment |
| CN116016206A (en) * | 2022-12-07 | 2023-04-25 | 北京航空航天大学 | Elasticity evaluation method and system of unmanned cluster system in open environment |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11410561B2 (en) * | 2020-04-06 | 2022-08-09 | Honeywell International Inc. | Traffic management systems and methods for unmanned aerial vehicles |
-
2023
- 2023-08-01 CN CN202310954303.6A patent/CN116860002B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017080406A1 (en) * | 2015-11-11 | 2017-05-18 | 刘晓阳 | Mooring unmanned rotorcraft cluster platform system and liquid continuous spraying system |
| CN113759975A (en) * | 2021-09-15 | 2021-12-07 | 北京航空航天大学 | Task capability-based unmanned aerial vehicle cluster modeling method and model system |
| CN114281104A (en) * | 2021-12-16 | 2022-04-05 | 成都戎星科技有限公司 | Multi-unmanned-aerial-vehicle cooperative regulation and control method based on improved ant colony algorithm |
| CN114995521A (en) * | 2022-08-08 | 2022-09-02 | 中国科学院自动化研究所 | Multi-unmanned aerial vehicle distributed formation control method and device and electronic equipment |
| CN116016206A (en) * | 2022-12-07 | 2023-04-25 | 北京航空航天大学 | Elasticity evaluation method and system of unmanned cluster system in open environment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116860002A (en) | 2023-10-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113282368B (en) | Edge computing resource scheduling method for substation inspection | |
| CN112866015B (en) | Intelligent energy-saving control method based on data center network flow prediction and learning | |
| CN109818786A (en) | A kind of cloud data center applies the more optimal choosing methods in combination of resources path of appreciable distribution | |
| CN108563863B (en) | Energy consumption calculation and scheduling method for urban rail transit system | |
| CN116016206B (en) | Elasticity evaluation method and system of unmanned cluster system in open environment | |
| CN113285831A (en) | Network behavior knowledge intelligent learning method and device, computer equipment and storage medium | |
| Guo et al. | Traffic engineering in hybrid software defined network via reinforcement learning | |
| CN115828143A (en) | Node classification method for realizing heterogeneous primitive path aggregation based on graph convolution and self-attention mechanism | |
| Rahbari et al. | Fast and fair computation offloading management in a swarm of drones using a rating-based federated learning approach | |
| Goel et al. | Improved multi-ant-colony algorithm for solving multi-objective vehicle routing problems | |
| CN116860002B (en) | Unmanned aerial vehicle cluster task resource scheduling method based on flow network model | |
| Li et al. | Network topology optimization via deep reinforcement learning | |
| CN103559538A (en) | BP neural network structure optimizing method | |
| Chang et al. | Dynamic flow scheduling optimization based on intelligent control for digital twins | |
| Lin et al. | Towards zero touch networks: From the perspective of hierarchical language systems | |
| Mishra et al. | Leveraging augmented intelligence of things to enhance lifetime of UAV-enabled aerial networks | |
| Malandrino et al. | Energy-efficient Training of Distributed DNNs in the Mobile-edge-cloud Continuum | |
| Goudarzi et al. | A GA-based fuzzy rate allocation algorithm | |
| Yang et al. | Virtual network function placement based on differentiated weight graph convolutional neural network and maximal weight matching | |
| Xiang et al. | HDFS efficiency storage strategy for big data in smart city | |
| Gu et al. | A new perspective of link prediction in complex network for improving reliability | |
| Zhang et al. | Link Value Estimation Based Graph Attention Network for Link Prediction in Complex Networks | |
| Wen et al. | A dimensional learning squirrel search algorithm based on roulette strategy | |
| CN112925953B (en) | Dynamic network representation method and system | |
| CN111935006B (en) | Data transmission method, device, processing equipment and medium |
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 |