Unmanned aerial vehicle cluster scheduling method based on communication coupling

Abstract

The invention discloses an unmanned aerial vehicle cluster scheduling method based on communication coupling, and the method comprises the steps: S1, dynamically constructing a task chain through a hierarchical search mechanism, dividing cluster nodes into four types: a reconnaissance node, a decision node, an execution node and a target node, setting a link selection rule, and generating a candidate OODA ring; s2, optimizing task chains based on an intelligent screening strategy, wherein the task chains are screened according to task priorities, combined score optimization and collaborative decomposition and order reduction are carried out; s3, establishing a communication and task coupling model, and realizing resource allocation through a three-layer architecture of a task layer directed flow network, a communication layer undirected topology network and a structure layer physical location network; and S4, executing a decision by adopting a hybrid optimization algorithm. According to the method, the configuration of the cluster task capability and the optimization of cluster communication are combined, so that the cluster is comprehensively optimal in communication and task capability when executing the task.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of unmanned aerial vehicle cluster scheduling, and particularly relates to an unmanned aerial vehicle cluster scheduling method based on communication coupling. BACKGROUND

[0002] An unmanned aerial vehicle cluster is a large-scale cluster system, which adopts a large number of (more than 10 orders of magnitude) low-cost unmanned aerial vehicle platforms. The basis of the unmanned aerial vehicle cluster is its communication, with inter-unmanned aerial vehicle cooperation as the core. Through the application of swarm intelligence technology, the unmanned aerial vehicle cluster develops in the direction of autonomous control, self-organization and self-adaptation. At the same time, with the development of actual application requirements, the unmanned aerial vehicle load is also constantly enriched, and the load modularization feature is increasingly prominent.

[0003] In the process of performing complex tasks such as fire rescue and atmospheric monitoring, the unmanned cluster is often affected by external environments such as terrain and obstacles, and is also limited by its own task capacity, resulting in disconnection or even failure of the communication link of the unmanned cluster, and further causing changes in the cluster topology structure, so that the cluster network is divided into several parts due to the inability to communicate with each other, and finally there is a risk of being unable to complete the expected task. Topology control is a common means to solve related problems in the unmanned cluster network. Topology control technology can prolong the life cycle of the overall network, so that the network has good overall performance. Existing topology control methods include power control algorithms, network node position movement and increasing network nodes, and the purpose of these methods is to improve the overall communication performance of the network. SUMMARY

[0004] To solve the above technical problems, the present application provides an unmanned aerial vehicle cluster scheduling method based on communication coupling, which combines the configuration of cluster task capacity and the optimization of cluster communication, so that the cluster is in the comprehensive optimal state of communication and task capacity when performing tasks.

[0005] To achieve the above purpose, the present application provides an unmanned aerial vehicle cluster scheduling method based on communication coupling, comprising:

[0006] S1, dynamically constructing a task chain through a hierarchical search mechanism, including dividing the cluster nodes into four types of reconnaissance nodes, decision nodes, execution nodes and target nodes, setting link selection rules and generating candidate OODA rings;

[0007] S2, optimizing the task chain based on an intelligent screening strategy, including three-stage processing of single-chain screening according to task priority, combination score optimization and cooperative decomposition and reduction;

[0008] S3, establishing a communication and task coupling model, and realizing resource allocation through a three-layer architecture of a task layer directed flow network, a communication layer undirected topology network and a structure layer physical position network;

[0009] S4, a hybrid optimization algorithm is used to execute decision-making, including generating an initial solution based on a maximum flow algorithm, local search by an improved simulated annealing algorithm and global optimization by multi-objective weight fusion.

[0010] The present application has the technical effect that the present application discloses a UAV cluster scheduling method based on communication coupling, the scheduling considers the influence of the reliability of cluster communication on task capability, and the reliability of UAV cluster task scheduling is improved. BRIEF DESCRIPTION OF DRAWINGS

[0011] The accompanying drawings, which form a part of the present application, are intended to provide further understanding of the present application and serve to explain the present application together with the specification. The illustrative embodiments of the present application and their description serve to explain the present application. In the drawings:

[0012] Figure 1 A task chain decision-making method flowchart of the embodiment of the present application;

[0013] Figure 2 A communication task coupling decision-making flowchart of the embodiment of the present application;

[0014] Figure 3 A function correlation diagram of the embodiment of the present application;

[0015] Figure 4 A communication task mapping flowchart of the embodiment of the present application;

[0016] Figure 5 A communication task coupling decision-making flowchart of the embodiment of the present application;

[0017] Figure 6 A task layer flow network diagram of the embodiment of the present application;

[0018] Figure 7 A communication task bandwidth allocation scheme flowchart of the embodiment of the present application;

[0019] Figure 8 A cluster disaster relief simulation scenario diagram of the embodiment of the present application;

[0020] Figure 9 A diagram showing the convergence of different algorithms with experiments of the embodiment of the present application. DETAILED DESCRIPTION

[0021] It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0022] It is noted that the steps shown in the flowcharts of the drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowcharts, in some cases the steps shown or described can be executed in an order different from that shown here.

[0023] The embodiment provides a UAV cluster scheduling method based on communication coupling, comprising:

[0024] S1, dynamically constructing a task chain through a hierarchical search mechanism, including dividing cluster nodes into four types of reconnaissance nodes, decision nodes, execution nodes and target nodes, setting link selection rules and generating candidate OODA rings;

[0025] S2, optimizing the task chain based on an intelligent screening strategy, including three-stage processing of screening single chains according to task priority, combined score optimization and cooperative decomposition and reduction;

[0026] S3, establishing a communication and task coupling model, and realizing resource allocation through a three-layer architecture of a task layer directed flow network, a communication layer undirected topological network and a structure layer physical position network;

[0027] S4, executing decisions by using a hybrid optimization algorithm, including generating an initial solution based on a maximum flow algorithm, local search by an improved simulated annealing algorithm and global optimization by multi-objective weight fusion.

[0028] For a UAV cluster, the current decision-making method is usually centralized decision-making, that is, unified command by a certain leader node, so in the embodiment, only the case of cooperative sensing and cooperative execution tasks is considered for task allocation method design. In order to simulate cooperative reconnaissance, cooperative execution and other task scenarios at the same time, considering the overhead and communication difficulty of multi-machine cooperation, the selection of the link selection link follows the following rules:

[0029] (1) For each target node T i , at most in two rings C i1 , C i2 , each target T i is at least in one ring, and the node inflow f is greater than or equal to the task demand M.

[0030] (2) For each reconnaissance node a i , at most in one ring C i .

[0031] (3) For each decision node b i , at most in two rings C i1 , C i2 , but the two rings need to have the same task target T i .

[0032] (4) For each execution node c i , at most in two rings C i1 ,C i2 , but the two rings need to have the same task target T i .

[0033] Suppose there are n tasks, each with different importance, and they are assigned to the UAV cluster. The following is its task link search process:

[0034] 1. First, all single loop configuration combinations that meet the task capability requirements need to be searched. In the search process, the following principles will be followed:

[0035] First, according to the importance of the task target, start from the highest importance target, and search all OODA rings in a depth-first manner.

[0036] For the task loop searched, it is necessary to check whether each node meets the traffic requirements of the task capability. In the case of meeting the task capability requirements, this method will record the k1 single chains with the highest task capability traffic. Here k1 is actually the smaller of the set search upper bound and the number of feasible solutions.

[0037] In the k1 cases corresponding to the k1 single chains, the corresponding nodes or traffic in the original network will be removed. Specifically, if the reconnaissance or execution node is used, the node will be removed; if the decision node is used, the traffic will be reduced. At the same time, based on the structure layer network, there is a mutual exclusion relationship between the task traffic of several different task capability nodes corresponding to the same structure node.

[0038] 2. After completing the first step, calculate the score of the single chain combination based on the above results. The score index is the scheduling target, and the link combination with the best score is selected.

[0039] 3. If the single chain combination covering all targets cannot be searched in the first step, the situation of cooperative sensing and decision-making will be searched. In the case of considering cooperation, a task t i with a large overall capability requirement is divided into two sub-tasks t i1 and t i2 , which require less task capability. Corresponding to the task nodes that need to be split in the network, each node's task capability is reduced while creating another node with the same task relationship. After modifying the original task layer network through this change, the DFS traversal gives a feasible OODA ring.

[0040] 4. Output the final result covering all target nodes, and the overall algorithm flow is shown in Figure 1 .

[0041] The simulated annealing algorithm mainly includes the following steps:

[0042] (1) Initialization: Select an initial solution and give a higher "temperature" value.

[0043] (2) Iterative process: a. Randomly select a candidate solution in the neighborhood of the current solution. b. Calculate the objective function value (usually the cost function or energy function) of the current solution and the candidate solution. c. If the candidate solution is better, accept it as the new current solution. d. If the candidate solution is not as good as the current solution, there is still a certain probability of accepting it as the new current solution, and this probability is related to the temperature. The higher the temperature, the greater the probability of accepting a poor solution. e. Reduce the temperature (usually according to a certain cooling plan, such as linear cooling, exponential cooling, etc.).

[0044] (3) Termination condition: When the temperature is reduced to below a certain threshold, or after a sufficient number of iterations, the algorithm terminates.

[0045] The key to the simulated annealing algorithm is its ability to accept inferior solutions in the early stage to escape from local optimal solutions. As the temperature continues to decrease, the algorithm gradually reduces the probability of accepting inferior solutions, thus stabilizing. First, generate the OODA loop in the cluster task network through network structure traversal, and then select the optimal task combination through the simulated annealing algorithm, thereby realizing rapid decision-making of the cluster.

[0046] Communication is an important element in the process of unmanned aerial vehicle cluster task, and the communication structure directly affects the coordination relationship of the unmanned aerial vehicle cluster. Limited communication resources always restrict the information transmission and reaction speed of the cluster. How to set a reasonable communication structure and optimize the cluster task structure to improve the reliability of the cluster task is a problem to be solved. At the same time, for different task links, the cluster needs to allocate different types of task resources. How to allocate limited communication resources to form a reasonable communication link to match the allocated tasks is also a problem to be solved. Finally, how to integrate the elements of unmanned aerial vehicle cluster in the communication and task two levels to form a comprehensive decision-making of unmanned aerial vehicle cluster considering the coupling of communication and task is also a problem to be solved. The overall research content is shown in Figure 2 .

[0047] To solve the above problems, first, a model of unmanned aerial vehicle cluster communication network is established according to the key elements of unmanned aerial vehicle cluster communication network in the task process; then the coupling relationship between the task load layer and the communication layer of the unmanned aerial vehicle cluster is analyzed, and a coupling model of the communication layer and the task layer of the unmanned aerial vehicle cluster is established; finally, based on the task and communication coupling model, a task decision-making method of unmanned aerial vehicle cluster considering the communication link of unmanned aerial vehicle cluster is designed to realize the optimal task structure decision-making of unmanned aerial vehicle cluster considering the communication factors.

[0048] Unmanned aerial vehicle cluster task decision-making model considering communication coupling

[0049] Decision elements for UAV swarm communication

[0050] Communication links in a UAV swarm are the foundation for the cooperation among different UAVs. In the decision-making of a UAV swarm task, communication conditions play an important role. It is directly related to the transmission, sharing, and efficiency of cooperation among UAVs. In a UAV swarm, each UAV needs to exchange information such as position, speed, and target state in real time in order to make collaborative decisions and actions. If the communication conditions are not good, the transmission of information may be interfered or delayed, which may cause the UAVs to fail to obtain the necessary information in time, and thus affect the decision-making efficiency and task execution effect of the entire swarm.

[0051] In order to make decisions for the task of a UAV swarm, it is necessary to evaluate the state of the UAV swarm during the task execution, including communication relationships, spatial position relationships, and task relationships. For these elements, existing research uses a three-layer complex network model. Specifically, the three-layer complex network model models the UAV swarm as a three-layer structure including a communication layer network, a collective layer network, and a task layer network.

[0052] Based on the three-layer complex network model of a UAV swarm, the smooth communication links between different UAVs are the premise for the cooperation of different payloads between UAV platforms. However, due to the different payloads carried by each platform in the UAV swarm, the task capabilities of each platform are different, and their communication needs are also different. Therefore, a reasonable communication strategy can maximize the communication needs of different UAVs and optimize the performance of the entire swarm. By considering the coupling effect of task capability, communication resources can be allocated according to the task demand and capability level of each UAV. In this way, UAVs with high task capability can obtain more communication resources to better complete tasks, while UAVs with low task capability can obtain appropriate communication resources to avoid resource waste.

[0053] Topology is an important research content of UAV swarm communication network problems. A reasonable topology can prolong the life cycle of the overall network and improve the performance of the network. Power control algorithm is suitable for the whole process from the initial state to the task process of the swarm, and is a widely studied method. The current research on UAV swarm communication topology and power control lacks consideration of the task characteristics and task capability of the UAV swarm, and the communication decision cannot fully meet the task demand. Therefore, in the coupling decision of UAV swarm task communication, it is of certain research value to allocate communication power according to the overall task demand.

[0054] In the three-layer network of UAV swarm, the spatial orientation directly affects the communication quality, the task demand and the ability are transmitted in the link through the information flow, the communication layer topology relationship is closely coupled with the spatial relationship of the UAV swarm node and the task demand, and therefore the communication layer topology control technology of the UAV swarm is the key of the three-layer topology control of the UAV swarm. The power allocation problem of the communication layer needs to consider the spatial distance between the nodes and meet the task ability demand of the swarm. The coupling relationship between the communication and the task of the swarm can be represented as Figure 3 .

[0055] The above relationship is called the function-related relationship of the UAV swarm communication, that is, the ability of the UAV swarm task layer node depends on the existence of the transmission of the UAV swarm communication layer node, and the loss of the UAV swarm communication layer node will directly lead to the failure of the corresponding UAV swarm task layer node.

[0056] Meanwhile, considering the characteristics of the UAV swarm communication network, the communication network connection and the task network connection of the UAV swarm are not one-to-one corresponding in topology. In the task network, according to the OODA relationship, the UAV task link is composed of links and sub-networks according to its ability; and in the communication network, the cooperative relationship between the task abilities does not directly correspond to the transmission relationship of the information. The transmission of the task information may have a certain relay, so as to realize the connection through a certain relay node, thereby improving the communication coverage and the communication quality. The transmission times of the information are called hops, and by limiting the maximum transmission times of the information, i.e., the maximum hop number of the information, the information transmission of the swarm can be limited. The nodes directly connected in the task layer may be connected in the same row layer through a limited number of relays, and the generation process of the route after the task decision is as shown in Figure 4 .

[0057] UAV swarm communication model

[0058] The UAV swarm communication refers to the information exchange mode between each UAV node in the UAV swarm and between the UAV and the ground control station. From the form, the communication technology of the UAV swarm includes self-organizing network (Ad Hoc), multi-network cooperation, and wireless self-organizing network technology. Regardless of which communication technology is adopted, the common elements in the current UAV swarm communication research include the communication bandwidth, the communication distance, the communication topology structure, the information sharing ability, and the reliability of the communication. Among them, the most important element of the swarm communication for the task decision is the bandwidth, the structure, and the delay of the communication.

[0059] The bandwidth limits the amount and speed of information that can be shared between the UAVs. In the case of limited bandwidth, the UAV needs to compress or select the most critical information for transmission. In the research in the latter part of the embodiment, how to allocate the limited bandwidth resources to adapt to each typical task is a problem to be solved.

[0060] The communication topology in a swarm, such as star-shaped, mesh-shaped, etc., affects the flow and processing of information, and different topologies have a direct impact on the efficiency and effectiveness of task decision-making. Therefore, by studying the task decision-making model of task and communication coupling, selecting the appropriate communication topology before the task starts is also an aspect of exploring the coupling relationship of communication tasks.

[0061] Communication distance limitations are mainly based on the comprehensive influence of topology structure and signal quality. In the process of the task, the spatial position of the swarm changes dynamically. The communication link decided at the beginning of the task must maintain its quality throughout the task process. Therefore, the task decision of the swarm must consider the spatial activity range of its structure.

[0062] It is crucial to study the significance of communication latency in UAV swarm communication, as it directly affects the efficiency of task execution, the responsiveness of the system, the formulation of distributed decision-making, the synchronization and coordination of swarm actions, and the optimization of resource allocation. Excessive communication latency can lead to delayed system responses, reduced decision-making quality, increased synchronization errors, and impact on network stability and fault tolerance, even affecting operational safety and user experience. In addition, the study of communication latency contributes to the development of technical standards and protocols, serving as a key indicator for evaluating the performance of different communication protocols and strategies, and plays an important role in determining the feasibility of specific communication technologies in practical applications. Therefore, a deep understanding of the characteristics of communication latency is crucial for designing efficient, reliable, and safe UAV swarm task decision-making.

[0063] (1) Channel Model

[0064] In the process of task execution of a UAV swarm, in the free space model of communication, there is a certain relationship between communication distance and transmission power. Generally, the longer the communication distance, the greater the signal transmission power needed to overcome propagation loss and interference. Specifically, the relationship between signal transmission power P and communication distance d can be expressed as:

[0065]

[0066] where P0 is the transmission power at a distance of 0, d is the communication distance, and α is an index related to the free space model. This relationship shows that as the communication distance increases, the transmission power needs to increase proportionally.

[0067] In the free space model, the exponent a is usually taken as 2, because according to the geometric optics theory, the attenuation in air is proportional to the square of the distance. This value is based on the idealized straight-line transmission assumption, ignoring factors such as multipath propagation and scattering. However, in actual free space communication systems, due to the influence of environmental factors, the actual value of a may vary depending on the specific situation. In addition, some existing technologies mention that the value of a is 4, which is mainly for free space communication systems in the high millimeter wave frequency band, considering the greater path loss caused by higher frequencies. However, overall, the value of a is between 2 and 4, and the specific value needs to be evaluated and selected according to the actual situation.

[0068] Assuming that the minimum acceptable received power of the communication network during the UAV cluster task process is P r , then the required transmission power Pt can be represented as:

[0069]

[0070] Assuming that during the task process, a certain UAV with sensing capability has a sensing range of R s , according to relevant literature, the communication range R c of such UAVs is generally required to be:

[0071] R c > = 2R s .

[0072] The signal and interference to noise ratio (SINR) of UAV communication can be represented as follows:

[0073]

[0074] where I is the interference of the single-hop neighbor node pair to this link at the receiving node.

[0075] The bit error rate of the system link ij can be given by the following formula:

[0076]

[0077] The bit error rate can reflect the connectivity of a communication link to some extent. The higher the bit error rate, the lower the efficiency of information transmission through this link, and the lower the connectivity of the communication link. Therefore, the bit error rate is used to calculate the communication quality of this link, which is calculated as follows:

[0078] P = 1 - PER.

[0079] (2) Communication bandwidth model

[0080] The communication bandwidth of a UAV swarm is a key indicator of the data exchange capacity in a UAV swarm system, which directly affects the capacity and efficiency of data transmission, determines the amount of data that can be transmitted per unit of time, and thus affects the efficiency of communication and the ability of multi-user access. The size of the bandwidth is related to the quality of service, especially for tasks with real-time requirements such as video transmission, which requires high bandwidth to ensure the smoothness and quality of information. At the same time, effective allocation and management of bandwidth is crucial for coordinating swarm actions, dealing with interference and noise, adapting to environmental changes, and meeting different task requirements. The actual available bandwidth is limited by multiple factors such as communication technology, hardware capability, spectrum policy, etc., and increasing bandwidth may involve a trade-off between cost and benefit. Therefore, reasonable planning and optimization of bandwidth use are the basis for ensuring the efficient and reliable operation of the UAV swarm communication system.

[0081] The communication bandwidth model of a UAV swarm usually takes into account the following factors: the number of UAVs, the bandwidth required by each UAV, the communication distance, signal attenuation, interference, bandwidth allocation strategy, and network topology structure. Here, a simplified assumption is adopted to construct the communication bandwidth model parameters as follows:

[0082] B total : Total available bandwidth.

[0083] B requiredi : Bandwidth required by the ith UAV.

[0084] The bandwidth allocation strategy of the swarm as a whole can be represented by the following formula:

[0085]

[0086] where the bandwidth required by each UAV B required is determined by the type of task segment being performed on the UAV and the amount of data required, and the data specifications for typical scenarios will be given in the simulation experiments.

[0087] (3) Communication delay model

[0088] Researching the communication latency of UAV swarm is crucial for ensuring the efficiency and accuracy of task decision-making, as in a swarm environment, the actions of each UAV depend on real-time or near real-time data exchange. Communication latency affects the freshness of information, which in turn affects the quality of decisions made based on that information. In emergency situations such as search and rescue missions, quick decision-making and response can mean the difference between success and failure. Additionally, latency can affect the synchronization and coordination capabilities of UAVs in the swarm, especially when precise formation or cooperative execution of complex tasks is required. Network delays can also cause delays in the execution of control commands, increasing the risk and uncertainty of task execution. Therefore, understanding and optimizing communication latency helps improve the response speed, reliability of task execution, and stability of the entire system of a UAV swarm. At the same time, this also helps in designing UAV swarm communication protocols and network architectures to consider performance requirements and constraints in practical applications, ensuring that the UAV swarm can effectively complete tasks in various environments and conditions. A communication latency model typically considers propagation delay, transmission delay, processing delay, and queuing delay. Propagation delay is determined by the speed of signal propagation in space and the distance between UAVs, transmission delay is related to the size of the data packet and the data rate of the communication link, processing delay is related to the processing capacity of network nodes, and queuing delay is related to network traffic and queue management strategies. The sum of these delays is the total communication latency, which provides a basis for evaluating and optimizing UAV swarm communication.

[0089] Assuming that the spatial distance is relatively close during the task process, a relatively simple communication transmission latency model is adopted, which only considers transmission latency during communication. The formula is as follows:

[0090] D = L / R;

[0091] where D is the size of the delay, L is the data volume of the task, and R is the data transmission rate. Assuming that the communication mode of the UAV swarm assumes a time division orthogonal form, multiple tasks are transmitted simultaneously, so queuing delay is not considered at this time. In order to study the interaction between communication and task, the demand for task data is mainly considered in the index, so transmission delay is selected as the representative element of communication latency.

[0092] The communication transmission rate R follows the following formula:

[0093]

[0094] where W is the bandwidth obtained by the communication UAV, P is the signal power at the UAV receiving the signal, and σ is the environmental noise.

[0095] (4) Communication power task capability correlation model

[0096] In the communication process, task information transmission and task state maintenance need to continuously consume the power of the UAV cluster. For each task, according to the OODA loop, it can be decomposed into a data type communication and an instruction type communication. Therefore, the execution time of a task can be decomposed into four parts, namely T1 forward time, T2 perception transmission time, T3 decision time and T4 attack time. In the whole process, the power consumption of the UAV is the flight consumption of the whole time period and the communication consumption in T2 and T3 time periods.

[0097] In a task, the total energy consumption of a UAV i is:

[0098] E i =P i (T1+T2+T3+T4)+P i2 T2+P i3 T3;

[0099] Where P i2 is the transmission power of UAV i in the task perception transmission process, and P i3 is the transmission power of UAV i in the task decision transmission process.

[0100] UAV cluster task scheduling method considering communication condition coupling

[0101] UAV cluster communication bandwidth constraint

[0102] UAV cluster communication bandwidth constraint is one of the important challenges faced in the current development of UAV technology. With the expansion of the size of the UAV cluster and the growth of communication demand, the allocation and management of communication bandwidth become particularly important. In the UAV cluster, each UAV needs to communicate with other UAVs and ground control stations in real time to share data, coordinate tasks and ensure flight safety. However, the communication bandwidth is limited and cannot meet the demand of all UAVs transmitting large amounts of data at the same time. Therefore, how to reasonably allocate and manage communication bandwidth to ensure smooth and efficient communication of the UAV cluster has become a problem to be solved.

[0103] The control algorithm of communication bandwidth constraint includes two parts: on the one hand, the dynamic bandwidth allocation algorithm adjusts the allocation of communication bandwidth in real time according to the communication demand and priority of UAVs. On the other hand, multi-hop communication and cooperative communication technologies can be used to expand the communication range and enhance the communication capacity.

[0104] The established decision model first considers the bandwidth constraint of UAV cluster communication under multi-task parallelism. Whether the bandwidth is sufficient depends on the specific task situation. In the scene design, typical tasks will be designed, and based on these typical tasks, several communication constraints will be determined.

[0105] The UAV cluster communication task can be divided into two categories: command type demand and task type demand. In these two categories, the configuration of communication quality and communication bandwidth plays a key role.

[0106] The command type demand mainly focuses on communication quality. This kind of demand includes real-time command transmission between UAVs, data sharing, and bidirectional communication with the ground command system. High communication quality can ensure that UAVs accurately and quickly receive and execute instructions, improving overall combat effectiveness. In order to meet this demand, the UAV cluster should choose a high-reliability task link to ensure the stability and security of the communication link.

[0107] The task type demand focuses on communication bandwidth. This kind of demand includes data transmission between UAVs, remote monitoring, and a large amount of data exchange with the ground command system. High communication bandwidth can reduce the information transmission delay between UAVs, improve the flexibility and response speed of task execution. There are two basic ways of bandwidth allocation: time division multiplexing and frequency division multiplexing. Frequency division multiplexing refers to allocating bandwidth into several intervals according to frequency, and multiple UAVs simultaneously use it for communication. Time division multiplexing refers to using UAVs in the same frequency range, and according to their own time slots, they take turns to communicate.

[0108] Assuming that the total bandwidth of the system is B, the task communication is divided into information and command two stages, and assuming that the UAV cluster will generate multiple clusters according to the structure, each cluster uses frequency division multiplexing to allocate bandwidth and communicates at the same time, and each cluster uses time division multiplexing to take turns to communicate. For a task containing multiple task targets, assuming that the UAV cluster forms m clusters to execute the task. The communication spectrum satisfies the following formula:

[0109]

[0110] In the overall task, the communication time T2 of a cluster M satisfies the following formula:

[0111] T 2M =∑t 2m ;

[0112] That is, the communication time of a cluster is the sum of the time of each communication in the cluster. In a task, for m clusters, the communication time T2 is the maximum communication time in each cluster.

[0113] Inter-node task communication quality

[0114] If a task information is forwarded for multiple times, i.e. multi-hop transmission, a large task delay will be generated in the process of UAV cluster task execution. Therefore, in general UAV cluster communication decision, the number of hops of information transmission is usually restricted. In order to ensure the high reliability of communication between any two communication nodes, the maximum number of hops of communication is selected as 2 in the embodiment. That is, in the OODA loop, the cooperative information among the perception, decision and execution loads is communicated through at most one relay node, which corresponds to the complex network model. In the network layer of the UAV cluster, the task link between the two task loads of the task layer network must have a communication layer link corresponding to it with a maximum length of 2. According to this rule, there are three communication situations: 1) for two task loads belonging to the same body layer, the communication hop number is 0, and the communication reliability is 1. 2) for two task loads belonging to different body layers, when they directly communicate, the task reliability is the reliability of the direct link between the two communication nodes. 3) when two task loads belonging to different body layers communicate through a relay, the communication reliability of the task information is given by the task reliability of 2-hop communication. In summary, the task communication reliability between the task loads c i and c j can be expressed as the following formula:

[0115]

[0116] wherein hc is the communication hop number, P ij is the task load c i and c j corresponding communication node direct task communication reliability, K is the set of common adjacent nodes of the task loads c i and c j corresponding communication nodes. The overall reliability of the communication network is considered as the series connection of all link reliabilities, i.e.

[0117] P total = ΠP i ;

[0118] wherein P i is the reliability of the ith communication link in the network.

[0119] Task capability constraint

[0120] The network capability of the cluster task load layer is affected by the corresponding communication nodes, so the final performance of the network capability depends on the interaction between the maximum capability that the task load can theoretically exert and the communication capability. For the task load layer network, it has been modeled as a flow network, so the potential task capability can be given by the maximum flow algorithm.

[0121] In the task layer flow network model, the flow direction of the task capacity is from the target node to the perception node, from the perception node to the decision node, from the decision node to the execution node, and from the execution node back to the target node to form a closed loop. The task capacity of the task chain flowing between the nodes can be simply obtained based on the constraint condition of the flow network traffic function.

[0122] However, in the process of performing tasks, there are often multiple task targets in the cluster, and the number of task chains increases with the increase of the number of task targets. When the same node is reused in different task chains, the task chains will cross each other, gradually forming a complex network structure. Simple calculation cannot solve the task capacity of a group of task chains forming a network structure. Here, the relatively mature maximum flow algorithm is applied to calculate the task capacity of the overall task network. For the calculation of the maximum flow of the network, the maximum flow calculation process in the network is as follows:

[0123] 1) Add a source point and a sink point of the flow network in the network, connect them to the target nodes at both ends of the network expansion diagram respectively. Initialize the traffic of the network to 0.

[0124] 2) Search a path in a breadth-first manner, take the minimum capacity of the edge between each point and the previous point as the traffic between the two points, and add the traffic value to the network flow.

[0125] 3) Trace the path in reverse, subtract the traffic value of the path from the capacity of the edge between the two points to establish a reverse edge, and update the entire network.

[0126] 4) Repeat steps 2) and 3) for the updated network until no augmented path can be found from the source point to the sink point. At this time, the network flow is the maximum flow.

[0127] Let the maximum flow of the network be C, which corresponds to the potential capacity of the task layer. Then the comprehensive index of the cluster network structure considering the task capacity and the communication quality can be represented as:

[0128] C combine = C*P.

[0129] Instruction transmission constraints

[0130] For each task chain, there are two main instructions, i.e., the information transmission from the perception load to the decision load and the information transmission from the decision load to the execution load.

[0131] The task information of the load transfer from the perception load decision is mainly data type information, including the position and state of the target object, and the more detailed and accurate the target information is, the more beneficial to the decision result it should be in theory. However, in the information transmission process, the communication delay condition requirement is low, but the data volume to be transmitted is large. This transmission corresponds to the business type scene, and the communication link of the unmanned aerial vehicle cluster serving this load relationship mainly requires sufficient communication bandwidth, and the time delay is mainly generated in the communication data transmission process, that is, the larger the data volume is, the slower the transmission rate is. For this business type information transmission, the data size is also closely related to the accuracy requirement of perception in the task process, and for high-value task targets, high-precision perception load is usually allocated for close reconnaissance, so as to provide accurate criteria for decision-making, which also directly leads to a large transmission data volume.

[0132] Accordingly, it is assumed that for business data, the data volume to be transmitted by perception depends on the perception ability of the unmanned aerial vehicle load performing the perception task. After the ability comprehensive sampling rate resolution and other ability indexes are evaluated, the result is that the perception ability is C a The perception node considers that the data volume to be transmitted in one task decision process is:

[0133] L=KC a ;

[0134] K is a constant in the task scene.

[0135] For a plurality of tasks T1, T2...T i , it is assumed that the communication information is transmitted in the network at the same time, and the total bandwidth of the system is B. The bandwidth allocated for each task is B1, B2...B i . The system needs to process all data transmission tasks within a limited time.

[0136] For task T i , let its task data volume be L i , and the transmission time t i it needs is:

[0137]

[0138] The data transmission needs to be completed within the specified total task time τ, and needs to meet the following constraints:

[0139] max{t1, t2...t i}<τ.

[0140] The transmission of decision to execution link is generally instruction type information, the characteristics of this kind of information is small data, high priority, high delay requirement, high precision requirement. In the transmission of this kind of information, the main delay comes from the relay jump process, in order to ensure that the task execution command can be correctly issued, set the minimum acceptable communication link quality R m , and limit the maximum number of instruction transfer to 2.

[0141] Cluster task decision algorithm considering the coupling of communication and task capability

[0142] The cluster task decision considering the coupling of communication and task capability needs to integrate the above communication and task model coupling elements. Specifically, first, generate a number of task chain decisions in the task layer according to the task value and task capability, then search for feasible communication decisions in these task decisions according to the coupling relationship between communication and task, and finally output the result. Specifically, in the communication search, the bandwidth allocation feasibility of data type communication needs to be solved first, then the optimization of the quality of instruction type communication is solved, and finally the composite decision of communication and task is formed. The specific process is as follows Figure 5 as shown.

[0143] (1) Task index

[0144] According to the constructed task communication reliability and task capability constraint, combined with the task capability constraint condition, the comprehensive decision optimization target is constructed. From the task layer, the existing task layer comprehensive evaluation function has considered the reliability, task cost and task time of the load, and its formula is:

[0145] B = ω1R + ω2P + ω3T;

[0146] Where R is the task layer reliability index, P is the task layer cost index, and T is the task layer time index. Based on the above comprehensive evaluation function, the unmanned aerial vehicle cluster can realize the optimal decision for the task layer. In order to calculate the above index, the following process can be carried out.

[0147] First, a three-layer network structure including communication layer, collective layer and structure layer is constructed, in which the task layer is a directed flow network, and the structure layer is an undirected network. The communication layer and the structure layer nodes are in one-to-one correspondence. Let the total network topology be G = {G1, G2, G3, V} including three-layer network topology and interlayer relationship V. The internal topology of the task layer network G3 is given by the OODA relationship and the task capability range of action, the internal topology of the body layer G2 is given by the spatial position, and the communication layer topology G3 is given by the initially set communication relationship. In order to construct the directed flow network of the task layer, a virtual source point and a virtual sink point need to be added in the task layer. The structure diagram of a task layer network is shown in Figure 6 as shown.

[0148] For any selected task link combination in the network, based on the maximum flow algorithm, the maximum capacity of the task layer is calculated as C.

[0149] According to the task chain capacity calculation results, for a fixed task layer structure, under the task capacity constraint, the task capacity of the unmanned aerial vehicle cluster can be calculated. The task value of the corresponding link combination {z1, z2...} is calculated, and the results are sorted in descending order according to the task value B.

[0150] (3) Data link decision based on communication bandwidth allocation

[0151] The data link is the communication signal of the perception payload of the unmanned aerial vehicle cluster. In order to make such signals meet the task requirements, it is necessary to allocate communication bandwidth within the unmanned aerial vehicle cluster. Communication bandwidth allocation is a complex process that involves efficient and fair use of network resources. From the task layer, the bandwidth allocation mainly considers whether the total bandwidth of the selected network structure as a whole meets the task requirements.

[0152] In order to generate the required data link and verify whether it meets the task requirements, for a given number of task link combinations {z i} according to the task value sorting, from z1 to solve the corresponding communication link and bandwidth allocation:

[0153] Before bandwidth allocation, first search for communication links. According to the task link, the corresponding communication network nodes can be mapped, and the nodes can be connected through 0 to 2-hop communication. In order to verify the existence of such connection, the simulated annealing algorithm is used to solve the minimum value of the shortest path of all service type communication and the corresponding communication structure. If the solution exists, further consider bandwidth allocation.

[0154] As Figure 7 shown, a relatively simple bandwidth allocation method, dynamic bandwidth allocation method with priority, is used, and the process is as follows:

[0155] 1. Initialization:

[0156] Sort all data communication links of the unmanned aerial vehicle cluster in descending order according to the priority P of the target. Initialize bandwidth allocation, which will divide 50% of the total bandwidth equally among all links, and reserve 50% of the total bandwidth.

[0157] 2. Bandwidth allocation:

[0158] For each link, if the currently allocated bandwidth is less than the required bandwidth, calculate the currently allocatable bandwidth, and allocate the bandwidth in the remaining bandwidth according to the importance ratio.

[0159] 3. Bandwidth adjustment:

[0160] The remaining bandwidth is allocated to high priority services according to the current allocation ratio.

[0161] 4. Bandwidth borrowing:

[0162] For each service request, check if the minimum bandwidth requirement is met. If not, borrow bandwidth from low priority tasks.

[0163] 5. Iterative optimization:

[0164] Repeat steps 2-4 until all link requirements are met or there is no more bandwidth to allocate.

[0165] 6. For cases where there is no feasible solution, replace the communication link and repeat steps 1-5.

[0166] If the constraints are not met for all potential communication links mapped to the current task, consider replacing the task decision scheme z.

[0167] Finally, according to the communication delay model, the communication delay of the data link mainly considers the transmission delay, including one or two times of forwarding delay, so the final decision should also meet the delay constraint, i.e. for the task link decision z i The communication delay of each service link must also be within the maximum delay τ allowed by the task. The set of service communication links corresponding to the task link decision that meets the requirements is denoted as E = {E1, E2,... E n}.

[0168] (2) Instruction link decision with communication delay constraints

[0169] Based on the allocation of data communication links, for scheme z i , the instruction communication link needs to be solved. Instruction communication link is small data, low delay, high quality signal. In order to meet this demand, environmental noise, interference and communication hop delay need to be considered.

[0170] First, make the following assumptions: 1. For the transmission of instruction information, the bandwidth occupation is not considered. 2. For the transmission of instruction information, the delay needs to consider the forwarding from the communication. 3. For the transmission of instruction information, the interference from the communication itself needs to be considered. 4. In a task, the data volume of each instruction information is the same.

[0171] The above assumption is based on the following facts: first, the instruction type information has less data volume, and does not occupy too much communication resource in the transmission process, and has high priority. Therefore, it is assumed that the instruction information uses the reserved fixed bandwidth, does not participate in the bandwidth allocation process of the service type information, and the bandwidth occupation is the same. Second, because the data volume is small, the time delay of data transmission is considered to be small in the communication time delay in the task decision, and the time delay of the forwarding process is mainly considered. Finally, because the instruction type link has high requirements on communication quality, the internal communication interference cannot be ignored.

[0172] In the case of simultaneous communication, the signal-to-noise interference ratio SINR can be expressed as the following formula:

[0173]

[0174] Accordingly, the bit error rate of each communication link can be further calculated to represent the communication reliability. At the same time, in order to maximize the task reliability in the cluster task execution process, the coupling between the communication reliability and the cluster task capability needs to be considered. In view of this requirement, the embodiment proposes a link search method using simulated annealing algorithm.

[0175] Based on the bandwidth allocation and link solving method proposed in the embodiment, the task decision z i , the mapping of the communication layer service link {E1, E2...E n} is generated.

[0176] In the case of selecting a group of instruction links M={M1, M2...M n}, it is assumed that for a task target Target i , the corresponding communication chain is E i and M i , and the communication reliability of this task is the series reliability of the communication link:

[0177] R Targeti =E i M i ;

[0178] Assuming that the importance of this target is w i , the index corresponding to the task capability is:

[0179]

[0180] In order to maximize the cluster task reliability, the selected communication link should satisfy the following results:

[0181] J=max∑J i .

[0182] The optimal combination can be solved by the simulated annealing algorithm. The flow is as follows:

[0183] 1) Determine the initial solution, set the initial temperature as T, and the cooling rate as α

[0184] 2) For the current solution, randomly replace one of the instruction links, calculate the difference ΔE between the replaced ∑J and the current solution i .

[0185] 3) If ΔE>0, accept the new solution, if ΔE<0, accept the new solution with a probability of e .

[0186] 4) Update the temperature T as αT .

[0187] 5) Repeat the processes of 2) to 4) until the temperature reaches the set minimum temperature.

[0188] The communication link generated by replacement needs to meet the communication delay constraint, according to the task condition, the maximum delay T max of task transmission is determined. max For each task target, according to its importance, the maximum transmission delay T max may be τ or 2τ. τ is the maximum time for one transmission in the unmanned aerial vehicle cluster communication, that is, for an instruction type information, at most two-hop forwarding.

[0189] In the search process, the maximum number of iterations is set to 500. If 500 iterations are performed without obtaining a combination of instruction links that meets the constraint, the scheme z i is replaced until the above requirements are met.

[0190] (4) Comprehensive communication capability index

[0191] Through the above solving, the total delay T of unmanned aerial vehicle cluster communication and the communication network reliability J under the fixed decision can be obtained, and by weighting and adding the two, the overall communication evaluation index M of the unmanned aerial vehicle cluster in one decision is obtained.

[0192] M=ω1T+ω2J;

[0193] And through the comprehensive of the task index B and the communication index M, the comprehensive benefit function V of the decision is obtained:

[0194] V=φ1B+φ2M.

[0195] An unmanned aerial vehicle cluster fire rescue scene case of the application, and the effectiveness of the established model and the proposed method is verified through numerical simulation. Netlogo is used to simulate the fire rescue scene.

[0196] There are 10 flammable areas in a certain forest, in order to timely respond to forest fires and reduce the harm of fire, people prepared a UAV cluster for fire rescue. The specific configuration of the UAV cluster is: a number of reconnaissance UAVs, a number of decision-making UAVs, and a number of execution UAVs.

[0197] The simulation adopts the multi-agent simulation software Netlogo, and the simulation scene is a square world of 200*200 cells. In this scene, the UAV carrying the sensing payload is set to blue, the UAV with the decision-making payload is set to green, and the UAV with the execution payload is set to red. The initial moment is located on the left side of the scene, and 10 immobile task targets are set on the right side of the screen, which are black in color. The 10 targets have different importance. During the task process, the UAV cluster first forms a formation to fly to the target area. When reaching the target area, the sensing UAV first conducts reconnaissance on the target, and the data is returned to the control node through the communication link. Finally, the control node distributes command information to command the relevant execution UAV to go to the target location and perform the material dropping task. This experiment aims to compare the influence of different decision-making methods on the final task result and the influence of different communication structures on the decision-making situation during the task process.

[0198] The initial and task execution process scenes are shown in Figure 8

[0199] Experimental scheme: (1) Comparison of different scale control structures

[0200] In order to compare the running situation of the algorithm under different scales, A, B, and C three UAV cluster quantity configurations are set, and two communication structures are set for comparison experiments. The ABC three UAV cluster quantity configurations are shown in Table 1.

[0201] Table 1

[0202]

[0203] In order to compare the communication adaptability of the UAV cluster under different structures to the proposed algorithm, the topology based on node degree and the communication structure based on the navigation method are compared. Under the control structure based on node degree, each UAV has connections with several UAVs of the same type in addition to the same connections as the task structure, so that the total degree of the node in the communication layer is 10, while the communication structure based on the navigation method only has the communication structure overlapping with the task layer.

[0204] (2) Comparison of different algorithms

[0205] ​To compare the differences of different algorithms in this task scenario, different algorithms are used to solve this scenario, and the final results are compared. Here, the present application selects the traversal algorithm, PSO algorithm and the comprehensive optimization algorithm proposed by the present application, and compares their running conditions in the scene.

[0206] Experimental results

[0207] (1) Comparison of different scale control structures

[0208] In the above scenario, the task communication coupling algorithm is used, and under the condition of setting the algorithm to run 1000 iterations, the experimental results are obtained by taking the average value of multiple experiments for different configurations and communication structures as shown in Table 2.

[0209] Table 2

[0210]

[0211] Under different task configurations and different communication topologies, the proposed algorithm gives a feasible solution. It can be seen that as the task input increases, the decision benefit objective function gradually reaches its upper limit, but in this process, the topology based on node degree control improves the communication relay capability by increasing the connection between nodes of the same type, and produces a decision scheme with higher task reliability.

[0212] (2) Comparison of different algorithms

[0213] Uniformly select unmanned aerial vehicle configuration B as the experimental configuration, and each algorithm is iterated 100 times. The convergence of different algorithms with experiments is shown in Table 3. Figure 9 According to the experimental results, for the proposed scenario and unmanned aerial vehicle configuration B, the proposed algorithm and the PSO algorithm both obtain good convergence effect. Although the PSO algorithm converges faster, it quickly falls into local optimum, and the final objective function value is slightly lower than the proposed task communication coupling algorithm. The traversal algorithm still needs a large number of iterations after 100 iterations to approach the optimal result. Accordingly, the algorithm proposed by the present application has achieved good results in the comprehensive practice and optimization effect for the preset scenario.

[0214] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any changes or replacements within the technical scope disclosed by the present application can be easily thought of by those skilled in the art, and should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A UAV cluster scheduling method based on communication coupling, characterized in that: include: S1. Dynamically construct task chains through a hierarchical search mechanism, including dividing cluster nodes into four categories: reconnaissance nodes, decision nodes, execution nodes, and target nodes, setting link selection rules and generating candidate OODA loops; S2. Optimize task chains based on intelligent screening strategies, including three-stage processing: screening single chains by task priority, combination scoring optimization, and collaborative decomposition and reduction; S3. Establish a communication and task coupling model and implement resource allocation through a three-layer architecture consisting of a directed flow network at the task layer, an undirected topology network at the communication layer, and a physical location network at the structure layer. S4. Use hybrid optimization algorithms to execute decisions, including generating initial solutions based on the maximum flow algorithm, improving the simulated annealing algorithm for local search, and global optimization based on multi-objective weight fusion.

2. The method for scheduling drone clusters based on communication coupling according to claim 1, characterized in that: The hierarchical search mechanism in S1 includes: layering the reconnaissance nodes, decision nodes, execution nodes, and target nodes according to their functions, setting the link selection rule for each target node to loop level ≤ 2 and node traffic ≥ task requirements, and generating candidate OODA rings of single rings or collaborative rings through depth-first search.

3. The method for scheduling drone clusters based on communication coupling according to claim 1, wherein: The process of the intelligent screening strategy in S2 includes: screening the optimal single link based on the task priority k1, weighted optimization of candidate links through a combined scoring formula, and then decomposing the collaborative ring into single links to reduce the task dimension.

4. The method for dispatching drone clusters based on communication coupling according to claim 1, wherein: The process of the communication and task coupling model in S3 includes: the task layer constructs a directed flow network to represent the OODA loop, the communication layer constructs an undirected topology network to manage bandwidth and latency, and the structure layer maps spatial relationships through a physical location network.

5. The method for dispatching drone clusters based on communication coupling according to claim 1, wherein: The process of the communication layer undirected topology network includes: dynamically adjusting the link bandwidth based on the bandwidth allocation model, and optimizing the instruction transmission path through the bit error rate minimization formula.

6. The method for dispatching drone clusters based on communication coupling according to claim 1, wherein: The process of the hybrid optimization algorithm in S4 includes: using the maximum flow algorithm to generate an initial task allocation plan, performing local search by improving the simulated annealing algorithm, and completing global optimization by combining multi-objective weights.

7. The method for dispatching a drone cluster based on communication coupling according to claim 1, wherein: The process of improving the simulated annealing algorithm includes: taking bandwidth utilization and delay jitter as energy functions, and controlling the solution space search range through the temperature decay rate.

8. The method for dispatching drone clusters based on communication coupling according to claim 1, wherein: The multi-objective weight fusion process in S4 includes: weighting and summing the communication quality index and the task efficiency index, and balancing resource allocation conflicts by dynamically adjusting the weight coefficients.

Images