Cluster task capability evaluation method, device and system, and storage medium

Abstract

The application discloses a cluster task capability evaluation method, device and system and a storage medium. The method proposes a cluster structure negative entropy model to describe the influence degree of a cluster static structure on cluster task capability. The correlation between single machine behavior decision and the cluster structure negative entropy model is considered, and then a cluster dynamic behavior negative entropy task capability evaluation model based on behavior decision is proposed to realize the purpose of evaluating cluster task capability. In view of the proposed cluster task capability measurement index, the application analyzes the influence factor sensitivity of the index from the angles of a cluster control structure, a cluster size, a task node ratio and a cluster link capability change, and verifies the correctness and accuracy of the index in describing the cluster task capability.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) swarm technology, and in particular to a method, apparatus, system, and storage medium for assessing swarm mission capabilities. Background Technology

[0002] In the field of complex system evaluation and prediction techniques, commonly used methods include Bayesian theory, DS evidence theory, Bayesian network models, Copula functions, machine learning, general generating function methods, bond graph methods, comprehensive evaluation methods based on information commutation, decision graph methods, simulation methods, and other related methods. These methods can be used to evaluate systems and components within a cluster. However, evaluating and predicting the mission capabilities of a cluster system requires addressing the cluster's behavior in conjunction with its systemic characteristics. With a clearer understanding of cluster equipment systems and mission processes, previously proposed cluster mission architectures are no longer adequate to reflect the complexity, connectivity, and networking characteristics of cluster systems. To overcome this limitation, Boyd proposed the OODA model, which decomposes the process of UAV cluster mission execution into four main stages: observation, judgment, decision-making, and action. Based on this, Cares established a static model, abstracting military equipment into nodes of these four stages and representing information, materials, and energy transfer during the mission process using edges, forming a preliminary complex network. Subsequently, research based on complex network methods and OODA loop descriptions of task processes has focused on evaluating cluster task capabilities and performance. However, while this approach considers and details the dynamic processes of cluster tasks, the impact of inter-entity interactions on the overall collaborative system remains unresolved. To address this, Tolk et al. improved upon Cares' static model, but their improvement did not fully consider the impact of node interactions on task capabilities. This is because current research on task capability evaluation is largely based on traditional indicator systems and decomposed using the analytic hierarchy process (AHP). While the task processes are comprehensive, the indicator system is static and independent. This approach essentially ignores the impact of inter-entity interactions on the overall collaborative task system capability. Therefore, moving away from existing traditional indicator systems and proposing task capability evaluation indicators that consider the dynamic task behavior of clusters is the primary task of cluster task capability evaluation.

[0003] Unmanned aerial vehicle (UAV) swarms, characterized by their numerous internal nodes, complex network structure, and diverse information transmission paths, are often used to execute complex and ever-changing mission commands. Their macroscopic state during missions is determined by node attributes and network conditions. Therefore, applying entropy theory to assess the mission capabilities of UAV swarms is highly adaptable. However, current applications of entropy theory in swarm assessment primarily address the uncertainty of node capabilities, considering only node attribute parameters and neglecting node mission behavior decisions. Therefore, current assessment methods are not entirely applicable to this context. Summary of the Invention

[0004] This invention addresses the aforementioned problems in the prior art. Therefore, there is a need for a method, apparatus, system, and storage medium for evaluating cluster task capabilities. Considering the direct or indirect interactions between unit nodes within a cluster, and the characteristic that cluster behavior is formed by the convergence and fusion of numerous internal node behaviors, this invention utilizes the "negative entropy principle" to standardize the description of the cluster task network structure. Approaching the issue from the perspective of network topology edges and nodes, it proposes a quantitative evaluation method for cluster task capabilities. Furthermore, to quantitatively describe the impact of node task behavior decisions on cluster task capabilities, considering the correlation between cluster single-machine behavior decisions and the cluster structure negative entropy model, a cluster dynamic behavior negative entropy task capability evaluation model based on behavior decisions is proposed.

[0005] According to a first aspect of the present invention, a method for evaluating cluster task capabilities is provided, the method comprising:

[0006] A negative entropy model of the cluster static structure is constructed to describe the degree of influence of the cluster static structure on the cluster task capability. The negative entropy model of the cluster static structure is expressed as follows:

[0007]

[0008] Where NEG represents the negative entropy of the cluster's static structure, and P ij Represents any two nodes V i and node V j The probability of timely transmission of task instructions between nodes, where n represents the total number of nodes;

[0009] Based on the correlation between cluster single-machine behavior decision-making and the cluster static structure negative entropy model, a cluster task capability evaluation model based on dynamic behavioral negative entropy is established to evaluate cluster task capabilities. The cluster task capability evaluation model based on dynamic behavioral negative entropy is expressed as follows:

[0010]

[0011] Where t represents the task time, t+τ represents the future task time, and NGE(t) represents the negative entropy of the cluster static structure at time t.

[0012] Furthermore, a negative entropy model of the cluster's static structure is constructed using the following method:

[0013] The cluster task network is represented by an undirected network graph, in which node V i If a connection to node V can be established through any node and edge... j The communication connection is established, thus determining node V. i and node V j There exists a path d ij Path d ij The calculation formula is:

[0014]

[0015] Where {k1,k2,…,k n} is node V i To node V j Intermediate nodes in the communication path From intermediate node k1 to node V i The accompanying weight, For intermediate node k j To node k n The accompanying weight;

[0016] According to node V i and node V j Path d between ij Determine node V i and node V j The shortest path length rij between r ij =min(d ij );

[0017] Let D be the sum of communication paths between all pairs of nodes in the cluster network topology. Then, for any two nodes V... i and node V j The probability of timely transmission of task instructions between them, P ij The following expression exists:

[0018]

[0019] Based on the negative entropy calculation formula in statistical physics, the probability of timeliness in task instruction transmission between any two nodes is accumulated to construct a negative entropy model of the cluster static structure.

[0020] Furthermore, based on the formula for calculating negative entropy and the Shannon entropy representation, the formula for calculating negative entropy under statistical physics is determined as follows:

[0021]

[0022] Where, p i This represents the probability of sample i appearing.

[0023] Furthermore, the Shannon entropy representation is expressed as follows:

[0024]

[0025] Where i labels all possible samples in the probability space, and p i K represents the probability of sample i occurring, and K is a constant related to the choice of unit.

[0026] Furthermore, a cluster task capability evaluation model based on dynamic behavioral negative entropy is established using the following method:

[0027] By incorporating the current task time t and the future task time t+τ into the impact of single-machine decision-making on cluster task capabilities, a cluster task capability evaluation model based on dynamic behavioral negative entropy is obtained.

[0028] Furthermore, the impact of the single-machine decision-making on the cluster task capability includes:

[0029] 1) The spatial position of a single machine is relatively stationary relative to the cluster center, that is, the single machine maintains its current behavioral state. At this time, the structural negative entropy of the single machine to the cluster remains unchanged.

[0030] 2) The spatial position of a single machine moves relative to the cluster center or the performance parameters of the single machine change, and at this time the absolute value of the structural negative entropy of the single machine on the cluster increases;

[0031] 3) The spatial position of a single machine moves relative to the cluster center or the performance parameters of the single machine change, and at this time the absolute value of the structural negative entropy of the single machine on the cluster is reduced.

[0032] Furthermore, the cluster task capability evaluation model based on dynamic behavioral negative entropy has the following two properties:

[0033] (1) NGE(t) has negative qualitative property, that is, NGE(t) < 0, where the negative sign represents the process of the cluster actively resisting the increase of entropy in the static structural state, and the absolute value |NGE(t)| represents the magnitude of the task capability in the state.

[0034] (2) The comparison of cluster dynamic behavior negative entropy models is only for a single task, and the cluster state capabilities under different task conditions are not comparable.

[0035] According to a second technical solution of the present invention, a cluster task capability evaluation device is provided, the device comprising:

[0036] The first model construction module is configured to construct a cluster static structure negative entropy model, which describes the impact of the cluster static structure on the cluster's task capabilities. The cluster static structure negative entropy model is expressed as follows:

[0037]

[0038] Where NEG represents the negative entropy of the cluster's static structure, and P ij Represents any two nodes V i and node V j The probability of timely transmission of task instructions between nodes, where n represents the total number of nodes;

[0039] The second model construction module is configured to establish a cluster task capability evaluation model based on dynamic behavioral negative entropy based on the correlation between cluster single-machine behavior decisions and the cluster static structure negative entropy model, so as to achieve the evaluation of cluster task capabilities. The cluster task capability evaluation model based on dynamic behavioral negative entropy is expressed as follows:

[0040]

[0041] Where t represents the task time, t+τ represents the future task time, and NGE(t) represents the negative entropy of the cluster static structure at time t.

[0042] Furthermore, the first model building module is further configured to build a cluster static structure negative entropy model by means of the following method:

[0043] The cluster task network is represented by an undirected network graph, in which node V i If a connection to node V can be established through any node and edge... j The communication connection is established, thus determining node V. i and node V j There exists a path d ij Path d ij The calculation formula is:

[0044]

[0045] Where {k1,k2,…,k n} is node V i To node V j Intermediate nodes in the communication path From intermediate node k1 to node V i The accompanying weight, For intermediate node k j To node k n The accompanying weight;

[0046] According to node V i and node V jPath d between ij Determine node V i and node V j The shortest path length rij between r ij =min(d ij );

[0047] Let D be the sum of communication paths between all pairs of nodes in the cluster network topology. Then, for any two nodes V... i and node V j The probability of timely transmission of task instructions between them, P ij The following expression exists:

[0048]

[0049] Based on the negative entropy calculation formula in statistical physics, the probability of timeliness in task instruction transmission between any two nodes is accumulated to construct a negative entropy model of the cluster static structure.

[0050] Furthermore, the first model building module is further configured to determine the negative entropy calculation formula under statistical physics based on the negative entropy calculation formula and the Shannon entropy representation, expressed as:

[0051]

[0052] Where, p i This represents the probability of sample i appearing.

[0053] Furthermore, the Shannon entropy representation is expressed as follows:

[0054]

[0055] Where i labels all possible samples in the probability space, and p i K represents the probability of sample i occurring, and K is a constant related to the choice of unit.

[0056] Furthermore, the second model building module is further configured to establish a cluster task capability evaluation model based on dynamic behavioral negative entropy using the following method:

[0057] By incorporating the current task time t and the future task time t+τ into the impact of single-machine decision-making on cluster task capabilities, a cluster task capability evaluation model based on dynamic behavioral negative entropy is obtained.

[0058] Furthermore, the impact of the single-machine decision-making on the cluster task capability includes:

[0059] 1) The spatial position of a single machine is relatively stationary relative to the cluster center, that is, the single machine maintains its current behavioral state. At this time, the structural negative entropy of the single machine to the cluster remains unchanged.

[0060] 2) The spatial position of a single machine moves relative to the cluster center or the performance parameters of the single machine change, and at this time the absolute value of the structural negative entropy of the single machine on the cluster increases;

[0061] 3) The spatial position of a single machine moves relative to the cluster center or the performance parameters of the single machine change, and at this time the absolute value of the structural negative entropy of the single machine on the cluster is reduced.

[0062] Furthermore, the cluster task capability evaluation model based on dynamic behavioral negative entropy has the following two properties:

[0063] (1) NGE(t) has negative qualitative property, that is, NGE(t) < 0, where the negative sign represents the process of the cluster actively resisting the increase of entropy in the static structural state, and the absolute value |NGE(t)| represents the magnitude of the task capability in the state.

[0064] (2) The comparison of cluster dynamic behavior negative entropy models is only for a single task, and the cluster state capabilities under different task conditions are not comparable.

[0065] According to a third technical solution of the present invention, a cluster task capability evaluation system is provided, the system comprising: a memory for storing computer programs; and a processor for executing the computer programs to implement the method described above.

[0066] According to a fourth technical solution of the present invention, a non-transitory computer-readable storage medium storing instructions is provided, which, when executed by a processor, performs the method described above.

[0067] The cluster task capability evaluation method, apparatus, system, and storage medium according to various embodiments of the present invention have at least the following technical effects:

[0068] (1) Based on the results of static analysis of complex networks, this invention proposes to use the negative entropy index of cluster dynamic behavior to measure the correlation between cluster active behavior and cluster task capability, and to quantify the impact of behavioral decisions on cluster task results.

[0069] (2) Regarding the proposed cluster task capability measurement index, this invention analyzes the sensitivity of the influencing factors of the index from the perspectives of cluster control structure, cluster size, task node ratio and cluster link capability change, and verifies the correctness and accuracy of the index in describing the cluster task capability.

[0070] (3) The task capability assessment model proposed in this invention can combine dynamic simulation and static structural analysis models, with each module interacting with the others. Using data as a medium, it forms a cluster task status assessment method based on the construction and deduction of dynamic relationships of system task behavior. This provides a certain reference and application foundation for subsequent intelligent optimization research on cluster task capabilities and cluster-related research. Attached Figure Description

[0071] In drawings that are not necessarily drawn to scale, the same reference numerals may describe similar parts in different views. The same reference numerals with or without letter suffixes may indicate different instances of similar parts. The drawings generally illustrate various embodiments by way of example rather than limitation and, together with the description and claims, serve to explain embodiments of the invention. Where appropriate, the same reference numerals are used in all drawings to refer to the same or similar parts. Such embodiments are illustrative and not intended to be exhaustive or exclusive embodiments of the apparatus or method.

[0072] Figure 1 A flowchart of a cluster task capability evaluation method according to an embodiment of the present invention is shown.

[0073] Figure 2 A comparison and verification diagram of the negative entropy behavior of different basic control models according to an embodiment of the present invention is shown.

[0074] Figure 3 The diagram illustrates the impact of cluster size on three basic control structures according to embodiments of the present invention, where a) the negative entropy value |NEG| of cluster dynamic behavior varies with cluster size under different control modes; and b) the percentage increase in the relative increment of negative entropy value |NEG| of cluster dynamic behavior varies with cluster size under different control modes.

[0075] Figure 4 A diagram illustrating the influence of task node allocation ratios according to an embodiment of the present invention is shown.

[0076] Figure 5 A diagram illustrating the impact of link task capabilities according to an embodiment of the present invention is shown. Detailed Implementation

[0077] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and specific examples, but this is not intended to limit the present invention. If there is no necessary sequential relationship between the various steps described herein, the order in which they are described as examples should not be considered a limitation. Those skilled in the art should understand that the order can be adjusted, as long as it does not disrupt the logical consistency between them and render the entire process impossible.

[0078] Example 1: Cluster Task Capability Evaluation Method

[0079] This invention provides a method for evaluating cluster task capabilities. Please refer to [link / reference]. Figure 1This is a flowchart of a cluster task capability evaluation method according to an embodiment of the present invention. The method includes steps S100-S200, which are described in detail below:

[0080] Step S100: Construct a negative entropy model of the cluster static structure to describe the degree of influence of the cluster static structure on the cluster's task capabilities.

[0081] A systems description based on reliability entropy theory is highly feasible for describing the complexity of the internal interaction structure of UAV swarms. Due to the real-time nature of command transmission and the variability of complex mission environments, the swarm system and its surrounding environment constitute an isolated system, without considering the introduction of swarm nodes or external human intervention.

[0082] For UAV swarms performing swarm tasks, their proactive response to the degree of chaos in their own system is directly manifested in the transmission of task link information, i.e., the composition of the swarm network structure. Therefore, this embodiment uses the "negative entropy principle" to standardize the description of the swarm task network structure, approaching it from the perspective of network topology edges and nodes, and initially achieving a quantitative assessment of the swarm task capabilities.

[0083] The specific construction principle and process of the cluster static structure negative entropy model are as follows:

[0084] Since the cluster task network structure is an element that directly reflects the cluster's proactive task capability, a cluster structure negative entropy model is proposed to describe the degree of influence of the cluster's static structure on the cluster's task capability.

[0085] The statistical physics community has rigorously derived an equivalent expression for the Boltzmann formula:

[0086]

[0087] Where i denotes all possible microstates, p i This represents the probability of microstate i occurring.

[0088] With the application of statistical physics to entropy theory, the concept of "Shannon entropy," used to characterize information content, was introduced, establishing the important position of entropy theory in information theory. The "Shannon entropy" (also known as "information entropy") is represented as follows:

[0089]

[0090] Where i labels all possible samples in the probability space, p i This represents the probability of the sample appearing. K is a constant related to the choice of unit, and is usually set to K=1.

[0091] Based on the formula for calculating negative entropy and the Shannon entropy representation, a formula for calculating negative entropy based on statistical physics is proposed:

[0092]

[0093] Where p i This represents the probability of sample i appearing.

[0094] In the static model of a complex UAV swarm network G={V,E}, task commands are transmitted through swarm task network links. These links are intricate, and information transmission involves uncertainty in path selection. To describe this uncertain transmission probability, P is used. ij Indicates any two nodes V in the system i and node V j The probability of timely transmission of task instructions between them is used to establish a negative entropy model of the static structure of the UAV swarm. The following section discusses P. ij The introduction and definition of the concepts are derived in detail.

[0095] Cluster task networks generally have bidirectional communication capabilities and are typically represented in the form of an undirected network graph in cluster topology networks, where node V is defined. i If a connection to node V can be established through any node and edge... j The communication connection is called node V. i and node V j It can communicate, that is, node V i and node V j There exists a path d ij To characterize the differences in actual task capabilities between each pair of nodes' path communication capabilities, a path distance with attached weight w is introduced as a reference for calculating the communication path. In this case:

[0096]

[0097] Where {k1,k2,…,k n} is node V i To node V j Intermediate nodes in the communication path.

[0098] Without considering the latency of information processing by cluster nodes, the timeliness of information transmission and feedback between individual machines mainly depends on the shortest path length r between any two nodes in the complex network topology G. ij r ij =min(d ij ).

[0099] To describe the probability of selecting the shortest path for information transmission between any two points in the cluster, D is defined as the sum of communication paths between all pairs of nodes in the cluster network topology. In this case, for any two nodes V... i and node Vj The probability of timely transmission of task instructions between them, P ij The following expression exists:

[0100]

[0101] Based on the formula for calculating negative entropy in statistical physics, by summing the negative entropy contributions of each node pair in the cluster network, we can obtain the formula for calculating the negative entropy (NEG) of the cluster's static structure:

[0102]

[0103] Based on the formula for calculating the negative entropy of this structure, the analysis and evaluation of the static structural state of the cluster can be achieved.

[0104] Step S200: Based on the correlation between cluster single-machine behavior decision-making and the cluster static structure negative entropy model, establish a cluster task capability evaluation model based on dynamic behavior negative entropy to achieve cluster task capability evaluation.

[0105] The static structure negative entropy model of a cluster can reasonably and accurately evaluate the task capability of a cluster in a static state. However, the changes in task capability during the execution of a UAV cluster task always depend on the interaction of the behavioral decisions of the numerous individual nodes in the cluster. At the same time, the cluster control structure does not change constantly. Simply establishing the correlation between the static structure model and the cluster task capability does not have the ability to evaluate dynamic task capability. Therefore, this embodiment considers the correlation between the cluster's individual behavioral decisions and the cluster structure negative entropy model, and proposes a cluster dynamic behavioral negative entropy task capability evaluation model based on behavioral decisions.

[0106] Specifically, referring to the static structural negative entropy model of a cluster, the dynamic behavioral negative entropy model of a cluster, considering behavioral decisions, is mainly used to describe the cluster task capability indicators under dynamic task scenario changes. According to the principle that single-machine decisions in an unmanned cluster affect the spatial location changes and performance parameters of a single machine, which in turn affect the task network link transmission capacity and the task link weight value w, ultimately affecting the static structural negative entropy model of the cluster. Based on the above influence process, the cluster structural negative entropy model caused by single-machine decisions can undergo the following three changes:

[0107] (1) The spatial position of a single machine is relatively stationary relative to the cluster center, that is, the single machine maintains its current behavior state. At this time, the structural negative entropy of the single machine to the cluster remains unchanged.

[0108] (2) The spatial position of a single machine moves relative to the cluster center or the performance parameters of the single machine change, and at this time the absolute value of the structural negative entropy of the single machine on the cluster increases, that is, the task capability of the cluster is improved in the active state.

[0109] (3) The spatial position of a single machine moves relative to the cluster center or the performance parameters of the single machine change. At this time, the absolute value of the structural negative entropy of the single machine on the cluster is reduced, that is, the task capability of the cluster decreases in the active state.

[0110] The impact of the above three types of single-machine decisions on cluster task capabilities is incorporated into the task time parameter, meaning that the current cluster task capability reflects the influencing factors of future cluster behavioral decisions. Therefore, by introducing the current task time t and the future task time t+τ into the cluster static structural negative entropy model, we obtain a cluster dynamic behavioral negative entropy model that considers behavioral decisions:

[0111]

[0112] As can be seen from the construction process of the cluster dynamic behavior negative entropy model, the model has two basic properties:

[0113] (1) NEG has negative qualitative properties, that is, NEG < 0, where the negative sign represents the process of the cluster actively resisting entropy increase in this static structural state, and the absolute value |NEG| represents the magnitude of the task capability in this state.

[0114] (2) NEG is relative, that is, the comparison of cluster dynamic behavior negative entropy models is only for a single task, and the cluster state capabilities under different task conditions are not comparable.

[0115] The cluster dynamic behavior negative entropy model that considers behavioral decision-making is a prerequisite for realizing subsequent behavioral decision optimization. That is, it establishes the correlation between behavioral decisions and changes in cluster task capabilities, providing a model basis for the realization of subsequent dynamic task optimization decision-making process.

[0116] Based on the aforementioned understanding of cluster behavior-structure-task multi-scale, during cluster task execution, different cluster control structures, cluster size, cluster task node ratio, and single-machine task capability indicators in the cluster will all have a certain impact on the overall task capability of the cluster. To ensure the rationality of the task capability evaluation index of the cluster dynamic behavior negative entropy model proposed in this paper, the following analysis and verification of the influencing factors of the model parameters of the model will be carried out in conjunction with Examples 2 to 5.

[0117] Example 2: Analysis of the Influence of Cluster Control Structure

[0118] The control structure model of unmanned aerial vehicle (UAV) swarms is a key factor reflecting the complexity and intelligence of the swarm. Based on the development trend of autonomous UAV swarms, early structures were primarily based on simple control and aimed at achieving victory through sheer numbers – a "human wave tactic." This evolved into network-centric decision-making structures proposed in the early 20th century with the development of networking, and now to distributed decision-making structures emphasizing behavioral decision-making. Each shift in swarm model is accompanied by adjustments to the internal control structure. In terms of control structure development trends, there is a tendency to transition from a comprehensive single-machine control model to a swarm-centric control model, and ultimately to an intelligent control model.

[0119] Based on the above analysis and considering the development trend of swarm control structure models, this paper proposes three control modes to describe the UAV swarm control structure: a simple behavior-based control model, a distributed navigation-based control model, and an autonomous intelligent control model. In addition, there are many other complex control models, which are essentially evolutions of typical control models; these will not be discussed in detail in this paper.

[0120] Behavioral control model:

[0121] The behavioral control model, also known as the "pigeon flock control model," aligns with the basic control principles of most biological groups in nature. Based on the analysis and research of the basic structure of the activities of biological groups such as flocks of geese and pigeons, their control methods are mapped to unmanned aerial vehicle (UAV) swarm systems, enabling rapid control of the swarm system structure.

[0122] The main configuration of the behavioral control model is a cluster of individual drones connected in close proximity and randomly. In this control structure, all individual drones in the drone swarm system do not have a clear control hierarchy. They rely solely on close-range communication to form simple communication links with several surrounding drones, thus forming a communication network structure for the swarm system. This ensures the normal execution of basic tasks and swarm behavior and avoids collisions between swarms.

[0123] Assuming the network is represented as G = {V, E}, and since each node in the behavioral control model is interconnected with its nearest communicable node and this connection is uncertain, and assuming the node indices are arranged according to spatial location, the adjacency matrix of the behavioral control structure model is represented as follows:

[0124]

[0125] Among them, w ij This indicates that there is a weight for the control connection between nodes i and j.

[0126] Because behavioral control models lack distinct control hierarchies, the autonomy requirements for individual systems are generally low. Even under interference or disruption, swarm systems can quickly form a basic control network, making it the fundamental control structure model for UAV swarms. It is typically used for basic patrol and search missions. However, in behavioral control models, control links between nodes are generally established based on proximity, making these links susceptible to real-time spatial relationships and prone to instability. This makes them unsuitable for complex swarm tasks. Furthermore, the lack of a clear control hierarchy presents significant drawbacks in transmitting mission commands or behavioral change information. Therefore, behavioral control models are typically preferred for local control in complex missions.

[0127] Distributed pilot control model:

[0128] The distributed navigation control model is a control model proposed in recent years based on the widely accepted "wolf pack control model" and the concept of distributed warfare. The wolf pack control model is derived from the advantages of orderly and strict task execution based on the study of the group behavior of wolf packs in nature, while distributed control is the core idea of ​​the "mosaic warfare" concept, which has advantages such as operational timeliness and rapid command response.

[0129] In the distributed navigation control model, based on the concept of distributed operations, the cluster is divided into different sub-clusters according to homogeneity or heterogeneity, and each sub-cluster has the ability to autonomously execute tasks. Each sub-cluster typically has one or more single-machine nodes with strong control capabilities, one of which is selected as the central node, and all other nodes establish connections with the central node. Simultaneously, control links exist between the central nodes of different sub-clusters. This forms an overall distributed, locally centralized control mode to ensure stable transmission and rapid response of control commands.

[0130] Because the distributed navigation control model has distributed cluster sub-clusters, in the network layer G={V,E}, it is stipulated that... Represents all nodes in the i-th sub-cluster. This represents the central node of the sub-cluster. The adjacency matrix of this distributed control structure model is then represented as follows:

[0131]

[0132] Because the distributed navigation control model has a central node, the control structure connections within each layer exhibit a clear directionality, thus providing strong stability and ensuring the top-down transmission and execution of task commands. Furthermore, the presence of multiple quasi-central nodes ensures a degree of resilience in the event of cluster disruption. However, since the central node is responsible not only for internal control of sub-clusters but also for establishing control connections with other sub-cluster central nodes, this control mode places high demands on the autonomy of the cluster's drones (i.e., the central node drones).

[0133] Autonomous intelligent control model:

[0134] The autonomous intelligent control structure model represents a high level of swarm autonomy and intelligence. This model possesses advantages such as full connectivity, intelligence, and autonomous dynamic adjustment, making it a promising research subject for future military intelligence. This control model primarily describes the control process of rapidly and rationally planning the swarm's task execution control links to achieve optimal task performance under conditions of strong individual unit autonomy. Based on current swarm technology development, the autonomous intelligent control model proposed in this paper mainly considers a dynamic link planning model under a certain degree of individual unit task autonomy.

[0135] The autonomous intelligent control structure model requires full connectivity for the cluster control structure, meaning there are no hierarchical control relationships within the cluster, and control links can be quickly established between any individual nodes. During task execution, individual nodes can rapidly adjust the structure of the actual link according to the task's adaptability to achieve the optimal result of instruction execution.

[0136] Since the autonomous intelligent control structure model has a dynamically changing fully connected structure, its adjacency matrix can be represented as follows:

[0137]

[0138] Among them (w) ij ) indicates that there is a potential control connection between node i and node j.

[0139] The autonomous intelligent control structure model is dynamic and variable, and can provide an intrinsic control structure foundation for dynamic behavior models. Therefore, in the subsequent research of this paper, it is mainly applied to the real-time task allocation behavior and the reconstruction behavior of communication links, so as to improve the autonomy of cluster behavior.

[0140] As the autonomy of the swarm increases, different control structures of UAV swarms have different impacts on the swarm's task execution capabilities. The main reason for the impact of the swarm control structure on the swarm's task capabilities lies in the change of the swarm network structure link connection method.

[0141] To verify the impact of this factor on the actual task capability of the cluster, this embodiment takes a 50-node cluster as an example and compares and verifies the aforementioned basic control model. During the verification, two cases are compared: one with and without considering the cluster task network weight. In the case of considering the task network weight, it is assumed that the weight value is only related to the spatial location of the nodes. The verification results are as follows: Figure 2 As shown.

[0142] Depend on Figure 2 The comparative analysis of the structural negative entropy of the three basic models shown under different node scales yields the following results:

[0143] (1) Under the condition of the same number of nodes and node parameters, regardless of whether the weight value is considered, the behavioral negative entropy value |NEG| has the following relationship: |NEG| 自主智能 >|NEG| 分布式领航法 >|NEG| 行为法 ;

[0144] (2) Without considering weights or only associating the spatial location of nodes with weight values, the fully autonomous intelligent control model has a smaller improvement in task capability compared to the distributed navigation model, while the distributed navigation control model has a more significant improvement in task capability compared to the behavior model.

[0145] Based on the above results, there is a corresponding physical model explanation in the cluster task process. Regarding result (1), as can be seen from the analysis of the three models above, as can be seen from the influence of the cluster control structure on the transmission of task instructions, as the autonomous capability of the cluster UAV improves, its control structure link connection is closer to the communication between any nodes, that is, the node task instruction transmission capability is more efficient. Therefore, the corresponding basic control structure task capability relationship is: autonomous intelligent control model > distributed navigation method control model > behavior method control model.

[0146] Regarding result (2): the distributed navigation control model and the fully autonomous control model have a small difference in the shortest path of task execution (the shortest path of the distributed navigation method is generally 1 to 2 node pairs, while the shortest path of the fully autonomous model is 1 node pair). Therefore, the difference in task capability between the two models is small.

[0147] Example 3: Analysis of the Impact of Cluster Size

[0148] During cluster task execution, the overall size of the cluster is often a key factor affecting the success of the task. Generally, the larger the cluster size, the higher its task execution capability. Based on the cluster dynamic behavior negative entropy index proposed in this paper, the sensitivity of this index to changes in cluster size is analyzed below.

[0149] To verify the impact of cluster size under different control structures, this embodiment still focuses on three basic control structure modes, discussing the changes in index parameters during the process of cluster size ranging from 2 to 100 aircraft. The verification results are as follows: Figure 3 As shown.

[0150] Depend on Figure 3 The following results can be obtained:

[0151] (1) Under the same control structure mode, as the number of cluster nodes increases, the cluster's task capability index |NEG| continues to increase, but the rate of improvement of task capability continues to decrease.

[0152] (2) The sensitivity of the cluster dynamic behavior negative entropy index to cluster size is basically the same under different control modes.

[0153] The above results are reflected in the physical manifestation of actual cluster tasks as follows: During cluster task execution, as the cluster size increases, the complexity of the cluster task network increases under the same control model, the number of task instruction transmission links between individual nodes increases, and the corresponding task capability of the cluster network should also increase accordingly. As the system gradually expands, the number of system task links gradually increases from a few necessary links to a large number of redundant task links, resulting in a corresponding decrease in the rate of improvement of cluster task capability. Therefore, the cluster dynamic behavior negative entropy task capability evaluation index proposed in this paper has a certain degree of rationality and interpretability in its response to the impact of changes in cluster size on task capability.

[0154] Example 4: Analysis of the Influence of Cluster Task Node Allocation

[0155] Cluster task node allocation refers to the ratio of the number of nodes in a cluster carrying different task payloads when executing tasks, and it mainly affects the cluster task link structure model. Cluster tasks are always executed jointly by single machines carrying different task payloads in the form of task links. Typically, the main entities executing tasks include three types: reconnaissance nodes, decision nodes, and execution nodes.

[0156] This embodiment primarily analyzes the impact of different ratios of execution nodes and decision nodes on the overall task capability of the cluster. The cluster size is kept constant at 100 drones. The actual ratio of execution nodes to decision nodes (i.e., the number of execution nodes controlled by the decision nodes) is adjusted to obtain the changes in negative entropy of the cluster's dynamic behavior, as shown below. Figure 4 As shown.

[0157] Depend on Figure 4 The following results can be obtained:

[0158] (1) With the cluster link task capacity and cluster size remaining constant, as the number of execution nodes controlled by the decision node increases, the cluster task capacity shows a trend of first increasing and then decreasing, and there are extreme points (such as...). Figure 4 Chinese x ymax =13), the specific value of this extreme point is related to the task capacity of each link in the current cluster and the total cluster size;

[0159] (2) When the cluster node ratio is small, the cluster task capability changes drastically with the node ratio and has a clear increasing trend; when the cluster node ratio is large, the cluster task capability changes more gradually with the node ratio.

[0160] Based on the above results, in actual cluster tasks, when there are too many decision nodes in the cluster, each decision node controls fewer execution nodes, failing to fully utilize the decision-making capabilities of the cluster's decision nodes, and the task links in the cluster are relatively simple. Therefore, the overall task capability of the cluster tends to drop sharply. Conversely, when there are too few decision nodes in the cluster, the entire cluster tends to adopt a fully pilot-based control mode. When the decision-making capabilities of the decision nodes are sufficient, it has a significant advantage in task execution. However, when the decision-making capabilities of the decision nodes are insufficient to control too many execution nodes, it will also reduce the effective task links in the cluster, thereby leading to a decrease in the cluster's task capability.

[0161] Example 5: Analysis of the impact of cluster link task capabilities

[0162] During cluster task execution, the individual node's task capabilities can change due to factors such as the task environment, its own degradation, and external attacks. These changes directly affect the capabilities of the task chains involved in that node, potentially impacting the overall task capabilities of the cluster. To characterize the impact of this node capability change process on the proposed cluster dynamic behavior negative entropy model, this embodiment analyzes the influence of task chains.

[0163] Based on the aforementioned analysis of the impact, this embodiment indirectly or directly maps the changes in the overall cluster task capability caused by non-task link disruption and non-node failure due to the combined effects of various factors during the cluster task process to the changes in cluster task link capability, and analyzes the influencing factors of link task capability.

[0164] In a drone swarm, the link capability between two individual nodes is typically influenced by many factors, including node parameters, communication quality, and environmental interference. To facilitate the description of the impact of link capability on the overall swarm capability, this embodiment utilizes the concept of a weighting coefficient to analyze the link capability. The weighting coefficient is defined as a comprehensive factor influencing all the above factors, used to characterize the actual communication capability of the link at a given moment.

[0165] Taking a distributed drone swarm of 100 drones, divided into 5 sub-swarms (each sub-swarm consisting of 19 execution drones and 1 decision-making drone), as an example, by changing the task capabilities of all links directly participated in by one of the drones (either the decision-making drone or the execution drone), the change in the negative entropy model value of the swarm's dynamic behavior as the weight coefficient of that node's task participation link changes is as follows: Figure 5 As shown.

[0166] Depend on Figure 5 The following results can be obtained from the curve showing the change in the effect:

[0167] (1) Under the condition that other factors remain unchanged, an increase in the task capacity of a certain link usually leads to an increase in the overall task capacity of the cluster;

[0168] (2) The impact of changes in the task capabilities of different links on the overall task capability of the cluster varies. The task capabilities of critical links have a greater impact on the overall task capability of the cluster, while changes in the task capabilities of non-critical links have a smaller or no impact on the overall task capability of the cluster.

[0169] Furthermore, although exemplary embodiments have been described herein, their scope includes any and all embodiments based on the invention that have equivalent elements, modifications, omissions, combinations (e.g., schemes involving intersections of various embodiments), adaptations, or alterations. Elements in the claims will be interpreted broadly based on the language used in the claims and are not limited to the examples described in this specification or during the implementation of this application, and such examples will be interpreted as non-exclusive. Therefore, this specification and examples are intended to be considered illustrative only, and the true scope and spirit are indicated by the following claims and the full scope of their equivalents.

[0170] The above description is intended to be illustrative and not restrictive. For example, the above examples (or one or more of them) can be used in combination with each other. Other embodiments can be used by those skilled in the art when reading the above description. Furthermore, in the above detailed description, various features may be grouped together to simplify the invention. This should not be construed as an intention that a feature of an unclaimed invention is necessary for any claim. Rather, the subject matter of the invention may be less than all the features of a particular embodiment of the invention. Thus, the following claims are incorporated herein by reference as examples or embodiments, wherein each claim is an independent, separate embodiment, and these embodiments are contemplated to be combined with each other in various combinations or arrangements. The scope of the invention should be determined by reference to the appended claims and the full scope of their equivalents.

Claims

1. A method for evaluating the capabilities of cluster tasks, characterized in that, The method includes: A negative entropy model of the cluster static structure is constructed to describe the degree of influence of the cluster static structure on the cluster task capability. The negative entropy model of the cluster static structure is expressed as follows: in, NEG This indicates the negative entropy of the cluster's static structure. Represents any two nodes and nodes The probability of timeliness in the transmission of task instructions between them. n Indicates the total number of nodes; Based on the correlation between cluster single-machine behavior decision-making and the cluster static structure negative entropy model, a cluster task capability evaluation model based on dynamic behavioral negative entropy is established to evaluate cluster task capabilities. The cluster task capability evaluation model based on dynamic behavioral negative entropy is expressed as follows: in, t Indicates the task time. Indicates the timeframe for future tasks. NGE ( t )express t Negative entropy of the cluster's static structure at time t; The cluster static structure negative entropy model is constructed using the following method: The cluster task network is represented by an undirected network graph, in which nodes... If a connection can be established between any node and an edge, then... The communication connection is then determined by the node. and nodes Path exists ,path The calculation formula is: in, For nodes To the node Intermediate nodes in the communication path intermediate node k 1 to node The accompanying weight, intermediate node k j To the node k n The accompanying weight; According to the node and nodes Path between Determine the node and nodes Shortest path length between ,in ; Let D be the sum of communication paths between all pairs of nodes in the cluster network topology. Then, any two nodes... and nodes Timeliness probability of task instruction transmission between The following expression exists: Based on the negative entropy calculation formula in statistical physics, the probability of timeliness of task instruction transmission between any two nodes is accumulated to construct a negative entropy model of the cluster static structure. The negative entropy model of cluster structure caused by single-machine decision-making can change in the following three ways: 1) The spatial position of a single machine is relatively stationary relative to the cluster center, that is, the single machine maintains its current behavioral state. At this time, the structural negative entropy of the single machine to the cluster remains unchanged. 2) The spatial position of a single machine moves relative to the cluster center or the performance parameters of the single machine change, and at this time the absolute value of the structural negative entropy of the single machine on the cluster increases; 3) The spatial position of a single machine moves relative to the cluster center or the performance parameters of the single machine change, and at this time the absolute value of the structural negative entropy of the single machine on the cluster is reduced.

2. The method according to claim 1, characterized in that, Based on the formula for calculating negative entropy and the Shannon entropy representation, the formula for calculating negative entropy under statistical physics is determined as follows: in, Indicates sample The probability of its occurrence.

3. The method according to claim 2, characterized in that, The Shannon entropy is expressed in the form of: in, Label all possible samples in the probability space. Indicates sample i The probability of its occurrence, It is a constant related to the choice of unit.

4. The method according to claim 1, characterized in that, A cluster task capability evaluation model based on dynamic behavioral negative entropy is established using the following method: Incorporate the current task time into the impact of single-machine decision-making on cluster task capabilities. and future mission time A cluster task capability evaluation model based on dynamic behavioral negative entropy was obtained.

5. The method according to claim 4, characterized in that, The cluster task capability evaluation model based on dynamic behavioral negative entropy has the following two properties: (1) NGE ( t It has negative definiteness, that is NGE ( t ) < 0, where the negative sign represents the process by which the cluster actively resists entropy increase in this static structural state, and the absolute value | NGE ( t | Represents the magnitude of the task capability in this state; (2) The comparison of the cluster dynamic behavior negative entropy model is only for a single task, and the cluster state capabilities under different task conditions are not comparable.

6. A cluster task capability evaluation device, used to implement the method as described in any one of claims 1 to 5, characterized in that, The device includes: The first model construction module is configured to construct a cluster static structure negative entropy model, which describes the impact of the cluster static structure on the cluster's task capabilities. The cluster static structure negative entropy model is expressed as follows: in, NEG This indicates the negative entropy of the cluster's static structure. Represents any two nodes and nodes The probability of timeliness in the transmission of task instructions between them. n Indicates the total number of nodes; The second model construction module is configured to establish a cluster task capability evaluation model based on dynamic behavioral negative entropy based on the correlation between cluster single-machine behavior decisions and the cluster static structure negative entropy model, so as to achieve the evaluation of cluster task capabilities. The cluster task capability evaluation model based on dynamic behavioral negative entropy is expressed as follows: in, t Indicates the task time. Indicates the timeframe for future tasks. NGE ( t )express t Negative entropy of the static structure of the cluster at a given time.

7. A cluster task capability evaluation system, characterized in that: The system includes: Memory, used to store computer programs; A processor for executing the computer program to implement the method as described in any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, perform the method according to any one of claims 1 to 5.

Images