A method for evaluating the resilience of unmanned swarms based on task network efficiency

Abstract

This application proposes a method for assessing the resilience of unmanned swarms based on task network efficiency, belonging to the field of unmanned swarm technology. The method includes the following steps: constructing a task network model for the unmanned swarm, comprising a topological network structure consisting of a detection layer, a decision layer, an execution layer, and a task layer, with different types of nodes, all nodes communicating with each other, and multiple association patterns among nodes; designing node failure and recovery strategies based on the topological network structure; formulating unmanned swarm performance indicators based on the node failure and recovery strategies, and formulating unmanned swarm resilience indicators based on the dynamic changes of the unmanned swarm performance indicators; and using the unmanned swarm resilience indicators to assess the resilience of the unmanned swarm. This application can effectively restore the performance of unmanned swarms and significantly improve the resilience of the task network model.

Description

Technical Field

[0001] This application relates to the field of unmanned swarm technology, and in particular to a method for evaluating the resilience of unmanned swarms based on task network efficiency. Background Technology

[0002] Task network efficiency is a key indicator for evaluating the performance of a swarm's task network, directly reflecting the information coordination and rapid response capabilities among its units. A highly efficient swarm task network ensures the rapid transmission of instructions and information among multiple unmanned units, thereby significantly improving the overall task effectiveness of the swarm. Studying swarm task network efficiency to assess the resilience of swarms not only helps enhance their stability and reliability in complex environments but also demonstrates their efficient autonomous collaborative planning capabilities and superior dynamic adaptability.

[0003] The resilience of unmanned swarms refers to how, under specific mission conditions, unmanned swarms comprehensively utilize their swarm absorption capabilities to resist interference from internal failures and external abnormal factors, and use adaptive technology to reduce the impact of these interferences. Furthermore, they dynamically reorganize the structure of the unmanned swarm through swarm recovery capabilities to improve mission efficiency and ensure the successful completion of missions under different circumstances.

[0004] Resilience, as a comprehensive indicator of unmanned swarms, reflects their ability to resist, absorb, adapt to, and recover from disturbances. Currently, the academic community mainly classifies quantitative assessment methods for the resilience of unmanned swarms into three categories: resilience assessment methods based on index analysis, resilience assessment methods based on simulation, and resilience assessment methods based on performance curves. While these resilience assessment methods each have their own advantages, they also have certain problems, as detailed below:

[0005] First, based on the resilience assessment method of indicator analysis, a complete indicator system was constructed, standardized basic indicators were established and the indicators were aggregated. The resilience of unmanned swarms was measured by obtaining the final top-level evaluation value. However, this resilience assessment method often needs to be used in combination with other resilience assessment methods and cannot be assessed alone.

[0006] Secondly, simulation-based resilience assessment methods, within the framework of a mission simulation system, capture data generated during the simulation process and perform similar statistical analysis and evaluation calculations to derive the resilience value of the mission system. While this resilience assessment method can theoretically clearly reflect the dynamic evolution process of unmanned swarm missions, in practical applications it suffers from problems such as high complexity of the simulation system, difficulty in solving the problem, and poor interpretability of the internal working mechanism.

[0007] Finally, the resilience assessment method based on performance curves originated from the resilience triangle model proposed by Bruno and introduced the performance curve of the integral model to evaluate resilience. This is currently the most commonly used method for assessing the resilience of unmanned swarms in academia. However, this resilience assessment method is too complex in data acquisition and processing, resulting in high computational costs and low accuracy, thus limiting its applicability.

[0008] Therefore, it is necessary to propose a solution to improve one or more problems existing in the above-mentioned related technical solutions.

[0009] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0010] This application provides a method for evaluating the resilience of unmanned clusters based on task network efficiency, including the following steps:

[0011] A task network model for an unmanned cluster is constructed. The task network model includes a topological network structure consisting of a detection layer, a decision layer, an execution layer, and a task layer in sequence. The topological network structure includes different types of nodes, all of which are interconnected and have multiple association patterns.

[0012] Based on the described network topology, design node failure strategies and node recovery strategies;

[0013] Based on the node failure strategy and the node recovery strategy, formulate unmanned cluster performance indicators, and based on the dynamic changes of the unmanned cluster performance indicators, formulate unmanned cluster resilience indicators.

[0014] The unmanned swarm resilience index is used to assess the resilience of the unmanned swarm.

[0015] In an exemplary embodiment of this application, the types of nodes include: the detection layer includes multiple sensing nodes, the decision layer includes multiple decision nodes, the execution layer includes multiple execution nodes, and the task layer includes multiple target nodes;

[0016] The set of all the aforementioned sensing nodes is represented as: S = {S1, S2, ..., S...} i ,…,S j ,…,S I}, S I Indicates the Sth I There are 1 sensing nodes, where I represents the number of sensing nodes;

[0017] The set of all decision nodes is represented as: D = {D1, D2, ..., D} i ,…,Dj ,…,D J}, D J Indicates the Dth J There are 1 decision nodes, where J represents the number of decision nodes;

[0018] The set of all execution nodes is represented as: W = {W1, W2, ..., W...} i ,…,W j ,…,W N}, W N Indicates the Wth N There are N execution nodes, where N represents the number of execution nodes;

[0019] The set of all the target nodes is represented as: T = {T1, T2, ..., T} i ,…,T j ,…,T M}, T M Indicates the Tth M There are 1 target nodes, where M represents the number of target nodes.

[0020] In an exemplary embodiment of this application, the various association modes include:

[0021] T i →S i , indicating the Sth i The sensing node detected the Tth... i The target node is described, and from the Tth... i Relevant information was obtained at the target nodes;

[0022] S i →S j , indicating the Sth i The sensing node will obtain the relevant information and combine it with the Sth... j The sensing nodes exchange information;

[0023] S i →D i , indicating the Sth i The first sensing node will transmit the acquired relevant information to the Dth node. i The aforementioned decision nodes;

[0024] D i →D j , indicating the Dth i The decision node mentioned above starts from the Dth node. j The decision node receives the task instruction and implements the decision node D. i and the decision node D j Decision-making coordination between them;

[0025] Di →S i , indicating the Dth i The decision node transmits the task instruction to the Sth node. i The Sth sensing node makes the Sth i Each of the aforementioned sensing nodes adjusts its detection task;

[0026] D i →W i , indicating the Dth i The decision node transmits the task instruction to the Wth node. i The execution node mentioned above makes the Wth node... i Each execution node performs a specific task;

[0027] W i →T i , indicating the Wth i The execution node affects the Tth node according to the task instruction. i The target nodes.

[0028] In an exemplary embodiment of this application, the probe layer uses G S This indicates that the decision-making layer uses G. D This indicates that the execution layer uses G. W This indicates that the task layer uses G. T express;

[0029] All the association patterns between the detection layer and the decision layer are represented by E. SD E represents all the association patterns between the decision-making layer and the execution layer. DW This indicates that all the association patterns between the execution layer and the task layer are represented by E. WT The association patterns between the task layer and the probe layer are represented by F. TS express;

[0030] The expression for the task network model is:

[0031]

[0032] Among them, GA N G represents the task network model. TS This represents the detection and sensing network pointing from the task layer to the detection layer; G SS G represents an information-sharing network between probe layers. SD G represents the information transmission network from the detection layer to the decision-making layer; DS This represents the information feedback network from the decision-making level to the detection level; G DD G represents an information fusion network among decision-makers; DWThis represents the command network from the decision-making level to the execution level; G WT This represents the task allocation network that points from the execution layer to the task layer.

[0033] In an exemplary embodiment of this application, the node failure strategy includes a random failure strategy, a maximum degree failure strategy, and a maximum betweenness failure strategy; the maximum degree failure strategy and the maximum betweenness failure strategy are collectively referred to as the intentional failure strategy.

[0034] The random failure strategy is used to simulate the node failure of the task network model due to its own internal faults and external environmental interference; the intentional failure strategy is used to characterize the overall performance change of the task network model when it is subjected to various intentional failure factors.

[0035] The node recovery strategies include a random recovery strategy, a maximum degree recovery strategy, and a maximum betweenness recovery strategy;

[0036] The random recovery strategy means that at any given time, a set of failed nodes is constructed based on the failure information of all nodes stored in the task network model. The set of failed nodes includes multiple failed nodes, and an exponentially distributed random number is generated for each of the failed nodes. If the random number corresponding to a failed node is less than a preset simulation parameter, then the failed node is considered to have completed recovery at that time.

[0037] The maximum degree recovery strategy means sorting all the failed nodes in the failed node set in descending order of their degrees, selecting the failed node with the maximum degree for recovery; if the failed node is successfully recovered, the recovery steps continue to recover other failed nodes; if the failed node is not successfully recovered, the recovery process ends at that moment.

[0038] The maximum betweenness recovery strategy means sorting all the betweennesses generated by the failed nodes in the set of failed nodes in descending order, selecting the failed node corresponding to the largest betweenness to be recovered; if the failed node is successfully recovered, the recovery steps continue to recover other failed nodes; if the failed node is not successfully recovered, the recovery process ends at that moment.

[0039] In an exemplary embodiment of this application, the step of constructing the random failure strategy includes:

[0040] The failure status of all the nodes within a certain time period is probabilistically characterized using the exponential distribution sampling method.

[0041] The expression for the probability representation is:

[0042]

[0043] Among them, F(t; λ) represents the probability that a node operates normally from the initial moment to the failure moment, t represents the t-th moment within the task cycle, and λ represents the failure rate of nodes in the task network model;

[0044] Within a certain period of time, all the nodes operating normally generate first random numbers within [0, 1], and the first random numbers are represented by random1;

[0045] Compare all the first random numbers with F(t; λ) respectively. If random1 < F(t; λ), then the node is the failure node, and extract all the relevant information of each associated pattern of each failure node from the task network model for use after the node recovers;

[0046] The steps for constructing the maximum degree failure strategy include:

[0047] Use the exponential distribution sampling method to probabilistically characterize the failure situations of all the nodes within a certain period of time;

[0048] Within a certain period of time, all the nodes operating normally generate second random numbers within [0, 1], and the second random numbers are represented by random2;

[0049] Compare all the second random numbers with F(t; λ) respectively. If random2 < F(t; λ), then the node is the failure node, and extract all the relevant information of each associated pattern of each failure node from the task network model for use after the node recovers;

[0050] Select the node corresponding to the maximum degree among the remaining nodes, and iteratively execute the above two steps until the generated second random number is greater than or equal to F(t; λ), at which point the failure process of the node ends;

[0051] Among them, the degree represents the number of links between a certain node and all other nodes in the task network model;

[0052] The steps for constructing the maximum betweenness failure strategy include:

[0053] Use the exponential distribution sampling method to probabilistically characterize the failure situations of all the nodes within a certain period of time;

[0054] Within a certain period of time, all the nodes operating normally generate third random numbers within [0, 1], and the third random numbers are represented by random3;

[0055] Compare all the third random numbers with F(t; λ) respectively. If random3 < F(t; λ), the node is a failed node, and extract the relevant information of all the associated patterns of each failed node from the task network model for use after the node recovers;

[0056] Select the node corresponding to the largest betweenness among the remaining nodes, and iteratively execute the above two steps until the generated third random number is greater than or equal to F(t; λ), and the node failure process at this moment ends;

[0057] Among them, the betweenness represents the proportion of the number of paths passing through the node in all the shortest paths in the task network to the total number of shortest paths.

[0058] In an exemplary embodiment of the present application, the step of designing a node failure strategy and a node recovery strategy according to the topological network structure includes:

[0059] Determine the initial states of all the nodes in the task network model, initialize the network parameters of the task network model, and set the seed of the random number generator according to the design strategy;

[0060] At any moment, for all the nodes of the task network model, generate the node failure sequence and the node recovery sequence of all the nodes respectively according to the probability representation of the design strategy;

[0061] Select all the failed nodes and all the recovered nodes, update the states of all the failed nodes and all the recovered nodes according to the probability representation, and record the relevant information of all the failed nodes and all the recovered nodes.

[0062] In an exemplary embodiment of the present application, the step of formulating an unmanned cluster performance function according to the node failure strategy and the node recovery strategy, and formulating an unmanned cluster resilience index according to the dynamic change of the unmanned cluster performance function includes:

[0063] Select appropriate node failure strategy and node recovery strategy, and design relevant simulation experiments;

[0064] In each simulation experiment cycle, calculate the initial task network global efficiency according to the initial states of all the nodes in the task network model;

[0065] Calculate the task network global efficiency according to the node failure strategy and the node recovery strategy at each moment in the simulation experiment cycle;

[0066] The global efficiency of the task network at all times within the simulation experiment period is summed, and the sum of the efficiency is divided by the initial global efficiency of the task network, and then multiplied by time to obtain the unmanned swarm resilience index.

[0067] In an exemplary embodiment of this application, the expression for the global efficiency of the task network is:

[0068]

[0069] Among them, NE on (t) represents the global efficiency of the task network, and N represents the number of effective closed loops for all target nodes in the task network model; N r The number of target nodes is represented by t, and t represents time t within the task cycle. n The parameter representing the index of the target node. This represents the length of the shortest effective closed loop that passes through the target node.

[0070] In an exemplary embodiment of this application, the expression for the unmanned swarm resilience index is:

[0071]

[0072] Where R represents the unmanned swarm resilience index, t0 represents the initial time, and t f Let P(t) represent the time when the unmanned cluster recovers to a stable state, P(t) represent the performance of the unmanned cluster at time t within the task cycle, and P(t0) represent the initial performance of the unmanned cluster.

[0073] Beneficial effects:

[0074] This application provides a method for evaluating the resilience of unmanned swarms based on task network efficiency, which has at least the following beneficial effects:

[0075] (1) This application has achieved a complete evaluation system by constructing a task network model, node failure strategy, node recovery strategy, task network global efficiency and unmanned cluster resilience index, which can independently evaluate the resilience of unmanned clusters.

[0076] (2) This application has a complete and independent resilience assessment system that can clearly reflect the dynamic evolution process of the unmanned swarm mission. Furthermore, the simulation using random failure strategy and intentional failure strategy is not difficult and easy to solve.

[0077] (3) The unmanned swarm resilience assessment method proposed in this application has lower computational cost and higher accuracy compared with the current resilience assessment method based on performance curves in terms of data acquisition and processing. Attached Figure Description

[0078] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0079] Figure 1 This diagram illustrates the steps of an unmanned swarm resilience assessment method based on task network efficiency in an exemplary embodiment of this application.

[0080] Figure 2 This diagram illustrates the structure of the task network model in an exemplary embodiment of this application.

[0081] Figure 3 This diagram illustrates the classification of the entire assessment system for assessing the resilience of unmanned swarms as proposed in an exemplary embodiment of this application.

[0082] Figure 4 A schematic diagram showing the change in resilience performance of a task network model based on node failure strategy and node recovery strategy in an exemplary embodiment of this application;

[0083] Figure 5 A comparative schematic diagram of different failure-recovery strategies in simulation experiments of exemplary embodiments of this application is shown;

[0084] Figure 6 This diagram illustrates a comparison of different maximum failure-recovery strategies in simulation experiments of exemplary embodiments of this application. Detailed Implementation

[0085] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided to make this application more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0086] Furthermore, the accompanying drawings are merely illustrative of this application and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0087] This example implementation provides a method for assessing the resilience of unmanned clusters based on task network efficiency, such as... Figure 1 As shown, the evaluation method may include the following steps:

[0088] Step S101: Construct a task network model for the unmanned cluster. The task network model includes a topology network structure consisting of a detection layer, a decision layer, an execution layer, and a task layer. The topology network structure includes different types of nodes, all nodes are connected to each other, and all nodes have multiple association patterns.

[0089] Step S102: Based on the network topology, design node failure strategies and node recovery strategies.

[0090] Step S103: Based on the node failure strategy and node recovery strategy, formulate unmanned cluster performance indicators, and based on the dynamic changes of unmanned cluster performance indicators, formulate unmanned cluster resilience indicators.

[0091] Step S104: Use unmanned swarm resilience indicators to assess the resilience of the unmanned swarm.

[0092] This application proposes a method for evaluating the resilience of unmanned swarms based on task network efficiency, which has at least the following beneficial effects:

[0093] (1) This application has achieved a complete evaluation system by constructing a task network model, node failure strategy, node recovery strategy, task network global efficiency and unmanned cluster resilience index, which can independently evaluate the resilience of unmanned clusters.

[0094] (2) This application has a complete and independent resilience assessment system that can clearly reflect the dynamic evolution process of the unmanned swarm mission. Furthermore, the simulation using random failure strategy and intentional failure strategy is not difficult and easy to solve.

[0095] (3) The unmanned swarm resilience assessment method proposed in this application has lower computational cost and higher accuracy compared with the current resilience assessment method based on performance curves in terms of data acquisition and processing.

[0096] The following section will provide a more detailed explanation of the unmanned cluster resilience assessment method based on task network efficiency proposed in this example embodiment.

[0097] In step S101 of this embodiment, as follows Figure 2As shown, a task network model for an unmanned swarm is constructed. Based on the Observation-Orientation-Decision-Action (OODA) loop theory, the task network model is composed of a topological network consisting of a detection layer, a decision layer, an execution layer, and a task layer. The more types of OODA loops that the unmanned swarm's task network model can construct for a single target node, the more diverse the task methods the unmanned swarm possesses when influencing the target node, thus reflecting the unmanned swarm's relatively strong task execution capability for a particular target node.

[0098] As a large-scale and complex system, the relationships between each unit in an unmanned swarm are complex and diverse. In this embodiment, each unit in the unmanned swarm is regarded as a different node, and the various association patterns of interconnection and interoperability between units are regarded as the lines connecting the nodes, thus obtaining a multi-level, heterogeneous, directed task network topology of the unmanned swarm.

[0099] The task network model comprises a topological network structure consisting of a detection layer, a decision layer, an execution layer, and a task layer. This topological network structure includes different types of nodes, all of which are interconnected and exhibit various association patterns.

[0100] Furthermore, in this embodiment, the node types specifically include the following:

[0101] The detection layer contains multiple sensing nodes, and the set of all sensing nodes is represented as: S = {S1, S2, ..., S...} i ,…,S j ,…,S I}, S I Indicates the Sth I There are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 ...

[0102] The decision-making layer contains multiple decision nodes, and the set of all decision nodes is represented as: D = {D1, D2, ..., D...} i ,…,D j ,…,D J}, D J Indicates the Dth J There are 1 decision nodes, where J represents the number of decision nodes and D represents the set of decision nodes.

[0103] The execution layer contains multiple execution nodes, and the set of all execution nodes is represented as: W = {W1, W2, ..., W...} i ,…,W j ,…,W N}, W N Indicates the WthN There are N execution nodes, where N represents the number of execution nodes and W represents the set of execution nodes.

[0104] The task layer contains multiple target nodes, and the set of all target nodes is represented as: T = {T1, T2, ..., T} i ,…,T j ,…,T M}, T M Indicates the Tth M There are 3 target nodes, M represents the number of target nodes, and T represents the set of target nodes.

[0105] Furthermore, in this embodiment, all association patterns include:

[0106] T i →S i , indicating the Sth i The sensing node detected the Tth... i The target node, and from the Tth... i Relevant information was obtained at each target node;

[0107] S i →S j , indicating the Sth i The sensing node will obtain the relevant information and combine it with the Sth sensing node. j The sensing nodes exchange information;

[0108] S i →D i , indicating the Sth i The first sensing node will transmit the acquired relevant information to the Dth sensing node. i Each decision node;

[0109] D i →D j , indicating the Dth i The decision node starts from the Dth node. j Each decision node receives task instructions, enabling decision node D to... i and decision node D j Decision-making coordination between them;

[0110] D i →S i , indicating the Dth i The decision node transmits the task instructions to the Sth node. i The Sth sensing node makes the Sth... i Each sensing node adjusts its detection task;

[0111] D i →W i , indicating the Dth i The decision node transmits the task instructions to the Wth node.i The execution node makes the Wth node... i Each execution node performs a specific task;

[0112] W i →T i , indicating the Wth i The execution node affects the Tth node according to the task instructions. i Target nodes.

[0113] The interconnection patterns among all nodes in an unmanned swarm are diverse and can be adjusted according to specific task requirements and changes in the task environment to ensure efficient information transmission and reliable task completion. For example, depending on the needs, some nodes among all sensing nodes, decision-making nodes, and execution nodes can use wireless communication to improve the mobility and flexibility of the unmanned swarm; other nodes can rely on wired connections for more stable and secure data transmission.

[0114] Furthermore, in this embodiment, given that the multi-level coupled network of the unmanned swarm has a large number of heterogeneous nodes and various heterogeneous relationships, the use of a hypernetwork model can effectively and clearly represent the structural characteristics of the task network model of the unmanned swarm. Based on the association patterns of different layers of the unmanned swarm network, there is no direct communication relationship between the target node in the task layer and the execution node in the execution layer. Therefore, the corresponding intra-layer connections are empty sets. Utilizing hypernetwork theory, this embodiment performs high-level abstraction of the nodes and association patterns in each layer of the unmanned swarm network, thereby obtaining:

[0115] Detection layer using G S It is indicated that the decision-making level uses G D This indicates that the execution layer uses G. W This indicates that the task layer uses G. T Represented by E. All correlation patterns between the detection layer and the decision layer are represented by E. SD This indicates that all relationship patterns between the decision-making and execution levels are represented by E. DW This indicates that all association patterns between the execution layer and the task layer are represented by... EW T represents the association patterns between the mission layer and the probe layer, denoted by E. TS express.

[0116] Furthermore, the expression for this task network model is:

[0117]

[0118] Among them, G AN G represents the task network model. TS G represents the detection and sensing network pointing from the target layer to the detection layer; SS G represents an information-sharing network between probe layers.SD G represents the information transmission network from the detection layer to the decision-making layer; DS This represents the information feedback network from the decision-making level to the detection level; G DD G represents an information fusion network among decision-makers; DW This represents the command network from the decision-making level to the execution level; G WT This represents the task allocation network from the execution layer to the target layer.

[0119] In step S102 of this embodiment, as follows Figure 3 As shown, based on the topology of the task network model, node failure strategies and node recovery strategies are designed. Step S102 in this embodiment may include the following steps:

[0120] Sub-step S1021: Determine the initial state of all nodes in the task network model, initialize the network parameters of the task network model, and set the seed of the random number generator according to the design strategy.

[0121] Sub-step S1022: At any given time, for all nodes in the task network model, generate the node failure sequence and node recovery sequence for all nodes according to the probability representation of the design strategy.

[0122] Sub-step S1023: Select all failed nodes and all recovered nodes, update the status of all failed nodes and all recovered nodes according to the probability representation, and record the relevant information of all failed nodes and all recovered nodes.

[0123] Furthermore, the node failure strategies include random failure strategies and intentional failure strategies, among which the intentional failure strategies include maximum degree failure strategies and maximum betweenness failure strategies.

[0124] Random failure strategies can be used to simulate node failures in a task network model due to internal faults and external environmental interference. Deliberate failure strategies can be used to characterize the overall performance changes of a task network model when subjected to various deliberate failure factors.

[0125] Furthermore, the steps for constructing a random failure strategy include:

[0126] First, the failure status of all nodes within a certain time period is probabilistically characterized using the exponential distribution sampling method.

[0127] The expression for probability representation is:

[0128]

[0129] Among them, F(t; λ) represents the probability that a node operates normally from the initial moment to the failure moment, t represents the t-th moment within the task cycle, and λ represents the failure rate of nodes in the task network model.

[0130] Secondly, within a certain period of time, all nodes operating normally generate a first random number within [0, 1], and the first random number is represented by random1.

[0131] Finally, compare all the first random numbers with F(t; λ) respectively. If random1 < F(t; λ), then the node is a failed node, and extract the relevant information of all associated patterns of each failed node from the task network model respectively, to be used after the node recovers.

[0132] The steps to construct the maximum-degree failure strategy include:

[0133] Firstly, use the exponential distribution sampling method to probabilistically characterize the failure situations of all nodes within a certain period of time;

[0134] Within a certain period of time, all nodes operating normally generate a second random number within [0, 1], and the second random number is represented by random2;

[0135] Compare all the second random numbers with F(t; λ) respectively. If random2 < F(t; λ), then the node is a failed node, and extract the relevant information of all associated patterns of each failed node from the task network model respectively, to be used after the node recovers;

[0136] Select the node corresponding to the maximum degree among the remaining nodes, and iteratively execute the above two steps until the generated second random number is greater than or equal to F(t; λ), at which point the failure process of this node ends.

[0137] Here, the degree represents the number of connections of a certain node in the task network model with all other nodes.

[0138] Furthermore, the steps to construct the maximum-betweenness failure strategy include:

[0139] Use the exponential distribution sampling method to probabilistically characterize the failure situations of all nodes within a certain period of time;

[0140] Within a certain period of time, all nodes operating normally generate a third random number within [0, 1], and the third random number is represented by random3;

[0141] Compare all the third random numbers with F(t; λ) respectively. If random3 < F(t; λ), the node is a failed node, and extract the relevant information of all associated patterns of each failed node from the task network model respectively, to be used after the node is restored;

[0142] Select the node corresponding to the largest betweenness among the remaining nodes, and iteratively execute the above two steps until the generated third random number is greater than or equal to F(t; λ), and the node failure process at this moment ends;

[0143] Among them, the betweenness represents the proportion of the number of paths passing through a node in all the shortest paths in the task network model to the total number of the shortest paths.

[0144] In this embodiment, the betweenness refers to the node betweenness, which represents the proportion of the number of paths passing through a certain specific node in all the shortest paths in the task network model to the total number of the shortest paths. The maximum betweenness failure strategy is similar to the maximum degree failure strategy.

[0145] Furthermore, the constructed node recovery strategy is a specific manifestation of the adaptation and recovery ability of the task network model of the unmanned cluster. In this embodiment, the node recovery strategy mainly includes a random recovery strategy, a maximum degree recovery strategy, and a maximum betweenness recovery strategy.

[0146] The random recovery strategy means that at any moment, a failed node set is constructed according to the failure information of all nodes stored in the task network model. The failed node set includes multiple failed nodes, and random numbers with an exponential distribution are generated for all the failed nodes correspondingly; if the random number corresponding to the failed node is less than the preset simulation parameter, it is considered that the failed node has completed recovery at this moment.

[0147] The maximum degree recovery strategy means that the degrees generated by all the failed nodes in the failed node set are sorted in descending order of nodes, and the failed node corresponding to the largest degree is selected for recovery; if the failed node is successfully recovered, other failed nodes are recovered according to this recovery step; if the failed node is not successfully recovered, the recovery process at this moment ends.

[0148] The maximum betweenness recovery strategy means that the betweennesses generated by all the failed nodes in the failed node set are sorted in descending order of nodes, and the failed node corresponding to the largest betweenness is selected for recovery; if the failed node is successfully recovered, other failed nodes are recovered according to this recovery step; if the failed node is not successfully recovered, the recovery process at this moment ends.

[0149] In step S103 of this embodiment, as Figure 4As shown, performance indicators for the unmanned cluster are formulated based on node failure and recovery strategies, and resilience indicators for the unmanned cluster are formulated based on the dynamic changes of these performance indicators. Step S103 in this embodiment may include the following sub-steps:

[0150] Sub-step S1031: Select appropriate node failure strategies and node recovery strategies, and design relevant simulation experiments.

[0151] Sub-step S1032: In each simulation experiment cycle, calculate the initial global efficiency of the task network based on the initial state of all nodes in the task network model.

[0152] Sub-step S1033: Calculate the global efficiency of the task network based on the node failure strategy and node recovery strategy at each moment during the simulation experiment period.

[0153] Furthermore, the expression for the global efficiency of the task network is:

[0154]

[0155] Among them, NE on (t) represents the global efficiency of the task network, and N represents the number of effective closed loops for all target nodes in the task network model; N r The number of target nodes is represented by t, and t represents time t within the task cycle. n The parameter representing the index of the target node. This represents the length of the shortest effective closed loop that passes through the target node.

[0156] In an unmanned swarm task network model, an efficient closed-loop path represents the entire cycle from task initiation to goal completion. Global task network efficiency is a measure of the unmanned swarm task network model's ability to achieve independent closed-loop operations for each target node at a macroscopic level. It reflects the overall effectiveness of the unmanned swarm task network model and the efficiency of overall collaborative tasks.

[0157] Sub-step S1034: Sum the global efficiency of the task network at all times during the simulation experiment period, divide the sum of the efficiency by the initial global efficiency of the task network, and then multiply by the time to obtain the unmanned swarm resilience index.

[0158] Using the global efficiency of the task network as a performance metric for unmanned clusters, the resilience of the task network model is considered in light of node failure and recovery strategies. Figure 4 As can be seen, the horizontal axis represents time, and the vertical axis represents performance. P(t) represents the global efficiency of the task network model at time t. Starting from time t0, nodes in the task network model gradually begin to fail according to the node failure strategy. dThe performance of the time-based task network model begins to decline. The unmanned cluster autonomously selects the recovery model and begins node recovery. m At time t, the performance of the task network model drops to its lowest point, after which its performance gradually recovers and reaches its maximum at time t. f It will eventually reach a stable state.

[0159] Furthermore, the expression for the resilience index of unmanned swarms is:

[0160]

[0161] Where R represents the unmanned swarm resilience index, t0 represents the initial time, and t f P(t) represents the time when the unmanned cluster's performance returns to stability after the performance improvement, P(t) represents the performance of the unmanned cluster at time t within the task cycle, and P(t0) represents the initial performance of the unmanned cluster.

[0162] In step S104, the unmanned swarm resilience index is used to assess the resilience of the unmanned swarm.

[0163] To verify the effectiveness of the unmanned swarm resilience assessment method based on task network efficiency proposed in this application, the following simulation experiments were conducted.

[0164] This simulation experiment establishes a task network of an unmanned cluster containing 105 nodes, including 20 target nodes, 45 perception nodes, 10 decision nodes, and 30 execution nodes, for simulation verification.

[0165] like Figure 5 As shown, Ar-Rr represents the random failure-random recovery strategy, Ar-Rd represents the random failure-maximum recovery strategy, and Ar-Rb represents the random failure-maximum betweenness recovery strategy.

[0166] Depend on Figure 5 As can be seen, in the case of random failure, the random failure-maximum recovery strategy results in the least reduction in the global efficiency of the task network, faster recovery, and the highest global efficiency of the task network when it finally recovers to a new stable phase.

[0167] When using the random failure-maximum betweenness recovery strategy, the degree of reduction in global efficiency of the task network is the same as that when using the random failure-maximum betweenness recovery strategy. However, after a period of rapid recovery, the recovery speed of the random failure-maximum betweenness recovery strategy decreases, and the global efficiency of the task network is the lowest when it finally recovers to a new stable phase.

[0168] When adopting the random failure-random recovery strategy, the global efficiency of the task network decreases the most, but the recovery rate is the highest. Finally, the global efficiency of the task network when it recovers to a new stable phase ranks second.

[0169] The three strategies were calculated based on the expression for the unmanned swarm resilience index, yielding the following results: R Ar-Rr =0.886, R Ar-Rd =0.933 and R Ar-Rb =0.912, which indicates that the unmanned cluster has the lowest performance degradation and the best recovery after adopting the random failure-maximum recovery strategy under random failure scenarios, and the unmanned cluster has the highest resilience value.

[0170] like Figure 6 As shown, Ad-Rr represents the maximum failure-random recovery strategy, Ad-Rd represents the maximum failure-maximum recovery strategy, and Ad-Rb represents the maximum failure-maximum betweenness recovery strategy.

[0171] Depend on Figure 6 As can be seen, under the maximum failure scenario, the task network global efficiency is reduced the least when the maximum failure-random recovery strategy is adopted, and the task network global efficiency ranks second when it finally recovers to a new stable phase.

[0172] When the maximum degree failure-maximum betweenness recovery strategy is adopted, the global efficiency of the task network decreases more than when the maximum degree failure-random recovery strategy is adopted, and the global efficiency of the task network is the lowest when it finally recovers to a new stable phase.

[0173] When the maximum failure-maximum recovery strategy is adopted, the degree of reduction in global efficiency of the task network is the same as that when the maximum failure-maximum betweenness recovery strategy is adopted. The global efficiency of the task network is the highest when it finally recovers to a new stable phase.

[0174] The three modes were calculated based on the expression for the unmanned swarm resilience index, yielding: R Ad-Rr =0.934, R Ad-Rd =0.933 and R Ad-Rb =0.932, indicating that in the maximum failure scenario, the failed nodes are all nodes with relatively high degree in the task network model. Under the same parameters, the maximum failure-random recovery strategy results in a larger number of nodes and the least performance degradation. Considering that the resilience indices calculated by the three strategies are very similar, and that the unmanned cluster has the highest performance recovery rate when using the maximum failure-maximum recovery strategy, it is expected that the maximum failure-maximum recovery strategy will outperform the maximum failure-random recovery strategy over a wider time frame.

[0175] Simulation experiments show that employing the random failure-maximum recovery strategy and the maximum failure-maximum recovery strategy not only effectively restores the performance of the unmanned swarm but also significantly improves the resilience of the task network model, thus demonstrating the unmanned swarm resilience assessment proposed in this application.

[0176] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0177] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. In addition, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.

[0178] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the scope of the technology disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application.

[0179] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

Claims

1. A method for evaluating the resilience of unmanned swarms based on task network efficiency, characterized in that, Includes the following steps: A task network model for an unmanned cluster is constructed. The task network model includes a topological network structure consisting of a detection layer, a decision layer, an execution layer, and a task layer in sequence. The topological network structure includes different types of nodes, all of which are connected to each other and have multiple association patterns. Based on the described network topology, design node failure strategies and node recovery strategies; Based on the node failure strategy and the node recovery strategy, formulate unmanned cluster performance indicators, and based on the dynamic changes of the unmanned cluster performance indicators, formulate unmanned cluster resilience indicators. The unmanned swarm resilience index is used to assess the resilience of the unmanned swarm.

2. The method for evaluating the resilience of unmanned swarms based on task network efficiency according to claim 1, characterized in that, The types of nodes include: the detection layer contains multiple sensing nodes, the decision layer contains multiple decision nodes, the execution layer contains multiple execution nodes, and the task layer contains multiple target nodes; The set of all the aforementioned sensing nodes is represented as: S = {S1, S2, ..., S...} i ,…,S j ,…,S I }, S I Indicates the Sth I There are 1 sensing nodes, where I represents the number of sensing nodes; The set of all decision nodes is represented as: D = {D1, D2, ..., D} i ,…,D j ,…,D J }, D J Indicates the Dth J There are 1 decision nodes, where J represents the number of decision nodes; The set of all execution nodes is represented as: W = {W1, W2, ..., W...} i ,…,W j ,…,W N }, W N Indicates the Wth N There are N execution nodes, where N represents the number of execution nodes; The set of all the target nodes is represented as: T = {T1, T2, ..., T} i ,…,T j ,…,T M }, T M Indicates the Tth M There are 1 target nodes, where M represents the number of target nodes.

3. The method for evaluating the resilience of unmanned swarms based on task network efficiency according to claim 2, characterized in that, The various association patterns mentioned include: T i →S i , indicating the Sth i The sensing node detected the Tth... i The target node, and from the Tth... i Relevant information was obtained at the target nodes; S i →S j , indicating the Sth i The sensing node will obtain the relevant information and combine it with the Sth... j The sensing nodes exchange information; S i →D i , indicating the Sth i The first sensing node will transmit the acquired relevant information to the Dth node. i The aforementioned decision nodes; D i →D j , indicating the Dth i The decision node mentioned above starts from the Dth node. j The decision node receives the task instruction and implements the decision node D. i and the decision node D j Decision-making coordination between them; D i →S i , indicating the Dth i The decision node transmits the task instruction to the Sth node. i The Sth sensing node makes the Sth i Each of the aforementioned sensing nodes adjusts its detection task; D i →W i , indicating the Dth i The decision node transmits the task instruction to the Wth node. i The execution node mentioned above makes the Wth node... i Each execution node performs a specific task; W i →T i , indicating the Wth i The execution node affects the Tth node according to the task instruction. i The target nodes.

4. The method for evaluating the resilience of unmanned swarms based on task network efficiency according to claim 3, characterized in that, The probe layer uses G S This indicates that the decision-making layer uses G. D This indicates that the execution layer uses G. W This indicates that the task layer uses G. T express; All the association patterns between the detection layer and the decision layer are represented by E. SD E represents all the association patterns between the decision-making layer and the execution layer. DW This indicates that all the association patterns between the execution layer and the task layer are represented by E. WT This indicates that all the association patterns between the task layer and the probe layer are represented by E. TS express; The expression for the task network model is: Among them, G AN G represents the task network model. TS This represents the detection and sensing network pointing from the task layer to the detection layer; G SS G represents an information-sharing network between probe layers. SD G represents the information transmission network from the detection layer to the decision-making layer; DS This represents the information feedback network from the decision-making level to the detection level; G DD G represents an information fusion network among decision-makers; DW This represents the command network from the decision-making level to the execution level; G WT This represents the task allocation network that points from the execution layer to the task layer.

5. The method for evaluating the resilience of unmanned swarms based on task network efficiency according to claim 1, characterized in that, The node failure strategies include random failure strategy, maximum degree failure strategy, and maximum betweenness failure strategy; the maximum degree failure strategy and the maximum betweenness failure strategy are collectively referred to as intentional failure strategy; The random failure strategy is used to simulate the node failure of the task network model due to its own internal faults and external environmental interference. The deliberate failure strategy is used to characterize the overall performance changes of the task network model when it is subjected to various deliberate failure factors; The node recovery strategies include a random recovery strategy, a maximum degree recovery strategy, and a maximum betweenness recovery strategy; The random recovery strategy means that at any given time, a set of failed nodes is constructed based on the failure information of all nodes stored in the task network model. The set of failed nodes includes multiple failed nodes, and an exponentially distributed random number is generated for each of the failed nodes. If the random number corresponding to a failed node is less than a preset simulation parameter, then the failed node is considered to have completed recovery at that time. The maximum degree recovery strategy means sorting the degrees of all the failed nodes in the set of failed nodes in descending order, and selecting the failed node with the maximum degree for recovery. If the failed node is successfully recovered, the recovery steps are repeated for the other failed nodes. If the failed node is not successfully recovered, the recovery process ends at that moment. The maximum betweenness recovery strategy means sorting all the betweennesses generated by the failed nodes in the set of failed nodes in descending order, and selecting the failed node with the maximum betweenness for recovery. If the failed node is successfully recovered, the recovery steps are repeated for the other failed nodes. If the failed node is not successfully recovered, the recovery process ends at that moment.

6. The method for evaluating the resilience of unmanned swarms based on task network efficiency according to claim 5, characterized in that, The steps for constructing the random failure strategy include: The failure status of all the nodes within a certain time period is probabilistically characterized using the exponential distribution sampling method. The expression for the probability representation is: Among them, F(t; λ) represents the probability that a node operates normally from the initial moment to the failure moment, t represents the t-th moment within the task cycle, and λ represents the failure rate of nodes in the task network model; Within a certain period of time, all the nodes operating normally generate first random numbers within [0, 1], and the first random numbers are represented by random1; Compare all the first random numbers with F(t; λ) respectively. If random1 < F(t; λ), then the node is the failure node, and extract the relevant information of all the associated patterns of each failure node from the task network model respectively, to be used after the node recovers; The steps for constructing the maximum degree failure strategy include: Use the exponential distribution sampling method to probabilistically characterize the failure situations of all the nodes within a certain period of time; Within a certain period of time, all the nodes operating normally generate second random numbers within [0, 1], and the second random numbers are represented by random2; Compare all the second random numbers with F(t; λ) respectively. If random2 < F(t; λ), then the node is the failure node, and extract the relevant information of all the associated patterns of each failure node from the task network model respectively, to be used after the node recovers; Select the node corresponding to the maximum degree among the remaining nodes, and iteratively execute the above two steps until the generated second random number is greater than or equal to F(t; λ), at which time the node failure process ends; Among them, the degree represents the number of connections between a certain node in the task network model and all other nodes; The steps for constructing the maximum betweenness failure strategy include: Use the exponential distribution sampling method to probabilistically characterize the failure situations of all the nodes within a certain period of time; Within a certain period of time, all the nodes operating normally generate third random numbers within [0, 1], and the third random numbers are represented by random3; Compare all the third random numbers with F(t; λ) respectively. If random3 < F(t; λ), then the node is a failure node, and extract the relevant information of all the associated patterns of each failure node from the task network model respectively, to be used after the node recovers; Select the node corresponding to the maximum betweenness among the remaining nodes, and iteratively execute the above two steps until the generated third random number is greater than or equal to F(t; λ), at which time the node failure process at this moment ends; Among them, the betweenness represents the proportion of the number of shortest paths passing through the node in the task network model to the total number of shortest paths; 7. The method for evaluating the resilience of unmanned swarms based on task network efficiency according to claim 6, characterized in that, The steps for designing the node failure strategy and the node recovery strategy according to the topological network structure include: Determine the initial states of all the nodes in the task network model, initialize the network parameters of the task network model, and set the seed of the random number generator according to the design strategy; At any given time, for all nodes of the task network model, node failure sequences and node recovery sequences are generated for all nodes according to the probability representation of the design strategy. Select all the failed nodes and all the recovered nodes, update the status of all the failed nodes and all the recovered nodes according to the probability characterization, and record the relevant information of all the failed nodes and all the recovered nodes.

8. The method for evaluating the resilience of unmanned swarms based on task network efficiency according to claim 7, characterized in that, The steps of formulating an unmanned cluster performance function based on the node failure strategy and the node recovery strategy, and formulating an unmanned cluster resilience index based on the dynamic changes of the unmanned cluster performance function, include: Select appropriate node failure strategies and node recovery strategies, and design relevant simulation experiments; Within each simulation cycle, the initial global efficiency of the task network is calculated based on the initial state of all nodes in the task network model. Calculate the global efficiency of the task network based on the node failure strategy and the node recovery strategy at each moment during the simulation experiment period; The global efficiency of the task network at all times within the simulation experiment period is summed, and the sum of the efficiency is divided by the initial global efficiency of the task network, and then multiplied by time to obtain the unmanned swarm resilience index.

9. The method for evaluating the resilience of unmanned swarms based on task network efficiency according to claim 8, characterized in that, The expression for the global efficiency of the task network is: Among them, NE on (t) represents the global efficiency of the task network, and N represents the number of effective closed loops for all target nodes in the task network model; N r The number of target nodes is represented by t, and t represents time t within the task cycle. n The parameter representing the index of the target node. This represents the length of the shortest effective closed loop that passes through the target node.

10. The method for evaluating the resilience of unmanned swarms based on task network efficiency according to claim 9, characterized in that, The expression for the unmanned swarm resilience index is: Where R represents the unmanned swarm resilience index, t0 represents the initial time, and t f P(t) represents the time when the unmanned cluster's performance returns to stability after the performance improvement, P(t) represents the performance of the unmanned cluster at time t within the task cycle, and P(t0) represents the initial performance of the unmanned cluster.

Images