Task-oriented heterogeneous uav swarm resilience evaluation and adaptive reconfiguration method and system

Abstract

This invention discloses a mission-oriented method and system for assessing the resilience of heterogeneous UAV swarms and for adaptive reconfiguration. The method divides surviving nodes into reconnaissance and perception nodes, command and decision-making nodes, and strike execution nodes. A mission communication coupling network is constructed on the same node set, sharing nodes at the physical, communication, and mission layers. The activation probability of directed links is calculated based on node distance and role-phase logical matching, and effective edges at the mission layer are generated under the conditions of physical reachability, communication reliability, and functional effectiveness. The performance of the physical, communication, and mission layers is calculated separately, and the overall mission performance is aggregated. Under hard-kill and soft-kill perturbations, a mission resilience index is obtained through threshold gating and time-discounted integration. Furthermore, adaptive reconfiguration is performed based on local communication utility and node role constraints, outputting the resilience assessment results and the mission chain recovery scheme. This invention avoids overestimating mission persistence capabilities in soft-kill scenarios solely based on topological connectivity.

Description

Technical Field

[0001] This invention relates to the fields of drone swarm network resilience assessment, task chain modeling, and adaptive reconfiguration technology, and particularly to a task-oriented method and system for heterogeneous drone swarm resilience assessment and adaptive reconfiguration. Background Technology

[0002] Heterogeneous drone swarms are typically composed of information acquisition nodes, coordination and decision-making nodes, and task execution nodes working together. Their task completion capability depends not only on physical connectivity, but also on whether the communication links meet the information flow requirements between roles and whether the task chain is closed.

[0003] Existing resilience assessment methods often employ structural metrics such as topological connectivity, average path length, or node survival rate, which struggle to distinguish between node damage caused by physical disruption and communication logic failure caused by functional inhibition. When the physical formation remains connected but communication inhibition breaks the OODA task chain, traditional structural metrics tend to overestimate the actual task maintenance capability of the colony.

[0004] Therefore, a resilience assessment method is needed that can simultaneously characterize physical reachability, task-driven communication activation, and task chain effectiveness, and provide an adaptive reconfiguration scheme that matches the task role after perturbation. Summary of the Invention

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] Firstly, a task-oriented method for assessing the resilience of heterogeneous UAV swarms and adaptive reconfiguration includes the following steps:

[0007] S1. Obtain the node role, node location, and communication radius of each drone node in the heterogeneous drone swarm; divide the set of surviving task nodes into three role sets: information collection nodes, coordination and decision-making nodes, and task execution nodes, and obtain the current task stage and disturbance information;

[0008] S2. Construct a task communication coupling network, including a physical layer, a communication layer, and a task layer, on the same set of surviving task nodes;

[0009] The physical layer describes the physical reachability between nodes; the communication layer describes the directed random communication activation between nodes; and the task layer describes the effective collaborative relationship between nodes that can form a closed-loop task chain, provided that the physical layer is reachable, the communication layer is activated, and the role transfer satisfies the OODA information flow rules.

[0010] S3. Based on the task-communication coupled network, calculate the physical layer performance, communication layer performance and task layer performance respectively, and weighted aggregate to obtain the comprehensive task performance; perform disturbance modeling and calculate the task resilience index based on the comprehensive task performance to evaluate the task's ability to maintain and recover after disturbance.

[0011] S4. Identify the broken closed-loop task chain and its corresponding failed nodes or failed communication edges based on the set of effective edges in the task layer after the disturbance. Calculate the local communication utility of candidate recovery nodes and perform enhanced reconstruction in combination with node role constraints. Update node positions and network status, and output the task resilience index and enhanced reconstruction scheme after the disturbance.

[0012] Preferably, S2 includes:

[0013] The physical reachability terms of the physical layer are determined based on the Euclidean distance and communication radius between each node; the task logical matching terms are determined based on the node role code and the role-stage weight matrix corresponding to the current task stage; the physical reachability terms and the task logical matching terms are multiplied to obtain the directed communication activation probability between each pair of nodes; when the directed communication activation probability reaches the task communication threshold and the role transfer between nodes satisfies the task observation, judgment and decision information flow rules, the corresponding directed edges are retained in the task layer to form an effective task relationship set; when physical damage occurs, the availability of the corresponding node in the three layers is changed; when functional inhibition occurs, the directed communication activation probability and the effective edges of the task layer are changed.

[0014] Preferably, in S2, the physical reachability terms of the physical layer are determined based on the Euclidean distance and communication radius between each node, including:

[0015] When the Euclidean distance between two nodes is less than the product of the distance attenuation segmentation coefficient and the communication radius, the physical reachability is fully reachable; when the Euclidean distance between two nodes is greater than or equal to the product of the distance attenuation segmentation coefficient and the communication radius but less than or equal to the communication radius, the physical reachability decreases linearly with respect to distance; when the Euclidean distance between two nodes is greater than the communication radius, the physical reachability is unreachable.

[0016] Preferably, in S2, the task logic matching items are determined based on the node role code and the role-stage weight matrix corresponding to the current task stage, including:

[0017] Each node's role is represented as a three-dimensional vector, with each vector component corresponding to one of the three roles: information gathering, coordination and decision-making, and task execution. A role-stage weight matrix is ​​determined based on the current task stage: when in the information aggregation stage, the weights of the convergence links between information gathering nodes and coordination and decision-making nodes, between information gathering nodes, and between coordination and decision-making nodes are increased; when in the task execution stage, the weights of the links between coordination and decision-making nodes and task execution node feedback links are increased. The transpose of the sending node's role encoding vector, the role-stage weight matrix, and the receiving node's role encoding vector are multiplied sequentially to obtain the task logic matching term.

[0018] Preferably, S3 calculates the physical layer performance, communication layer performance, and task layer performance separately, including:

[0019] The physical layer performance is obtained by fusing node survival rate, average degree, clustering coefficient and global efficiency. The average degree, clustering coefficient and global efficiency are respectively calculated by weighting the ratio of their current value to their initial value, and then multiplied by the node survival rate.

[0020] The communication layer performance is obtained as follows: For each surviving task node, calculate the communication entropy of its outgoing communication probability normalized distribution, take the complement of the ratio of the communication entropy to the upper limit of the communication entropy, multiply it by the communication strength mapping value, and then calculate the mean for all surviving task nodes; where the communication strength mapping value is represented by a negative exponential function of the sum of communication probabilities;

[0021] The task layer performance is the ratio of the total task efficiency of all valid closed-loop task chains at the current moment to the total task efficiency of all valid closed-loop task chains at the initial moment; where the task efficiency of each execution node is the maximum value of the reciprocal of the link length among all valid closed-loop task chains it participates in.

[0022] Preferably, in S3, the disturbance includes physical damage disturbance and functional suppression disturbance;

[0023] When physical damage occurs, delete the damaged node and its associated edges;

[0024] When functional suppression occurs, the directed communication activation probability associated with the suppressed node is reduced or set to zero. This causes the suppressed node to remain alive in the physical layer, but its directed communication activation probability in the communication layer decreases, thereby breaking the corresponding effective edge in the task layer.

[0025] As a preferred option, in S3, the task resilience index is calculated as follows:

[0026] Taking the time of disturbance occurrence as the starting point of integration and the time of evaluation termination or the time of recovery of observation as the end point of integration, the non-negative part of the difference between the comprehensive task performance and the minimum task feasibility threshold is taken as the effective performance contribution of the integrand.

[0027] The effective performance contribution term is weighted by exponential decay using a time discount factor;

[0028] The ratio of the integral value of the weighted effective performance contribution item over the integral interval to the corresponding integral value under the ideal state is used as the task resilience index.

[0029] Preferably, in S4, the local communication utility of the candidate recovery nodes is calculated and enhanced reconstruction is performed in conjunction with node role constraints, including:

[0030] For each candidate recovery node, calculate the communication entropy of its outgoing communication probability normalized distribution, take the complement of the ratio of the communication entropy to the upper limit of the communication entropy, and multiply it with the communication strength mapping value. Based on the multiplication result and the directed communication activation probability between the candidate recovery node and its neighboring nodes, obtain the local communication utility.

[0031] For nodes that still have communication neighbors, select the node with the highest local communication utility as the target for restoring cooperation;

[0032] For isolated nodes, select the target area or target node that can restore the closed-loop task chain based on its role;

[0033] Update node positions based on the selected reconstruction target, recalculate network state and overall task performance, and output the perturbed task resilience index and the set of executable task chains after reconstruction.

[0034] Secondly, a task-oriented heterogeneous UAV swarm resilience assessment and adaptive reconfiguration system is provided to implement the task-oriented heterogeneous UAV swarm resilience assessment and adaptive reconfiguration method described in the first aspect, comprising:

[0035] The data acquisition module is used to acquire the node role, node location, and communication radius of each drone node in the heterogeneous drone swarm; to divide the set of surviving task nodes into three role sets: information collection nodes, coordination and decision-making nodes, and task execution nodes; and to acquire the current task stage and disturbance information.

[0036] The Coupled Network Construction Module is used to construct a task communication coupled network, including a physical layer, a communication layer, and a task layer, on the same set of surviving task nodes. The physical layer describes the physical reachability between nodes; the communication layer describes the directed random communication activation between nodes; and the task layer describes the effective collaborative relationship between nodes that can form a closed-loop task chain, provided that the physical layer is reachable, the communication layer is activated, and the role transfer satisfies the OODA information flow rules.

[0037] The resilience assessment module is used to calculate the physical layer performance, communication layer performance, and task layer performance based on the task communication coupled network, and then weight and aggregate them to obtain the comprehensive task performance; it performs perturbation modeling and calculates the task resilience index based on the comprehensive task performance to evaluate the task's ability to maintain and recover after perturbation.

[0038] The enhanced reconstruction module is used to identify broken closed-loop task chains and their corresponding failed nodes or failed communication edges based on the set of valid edges in the task layer after the disturbance; calculate the local communication utility of candidate recovery nodes and perform enhanced reconstruction in combination with node role constraints; update node positions and network status; and output the task resilience index after the disturbance and the enhanced reconstruction scheme.

[0039] Compared with the prior art, the beneficial effects of the present invention are reflected in:

[0040] This invention constructs a shared node coupling model of physical layer, communication layer and task layer on the same set of surviving nodes. It characterizes and evaluates physical reachability, communication activation probability and task chain closure effectiveness separately and in a coordinated manner. It can identify functional suppression scenarios where the physical topology is still connected but the communication link is suppressed and the closed-loop task chain has been broken, thus avoiding overestimating the resilience of the bee colony by relying solely on topological connectivity.

[0041] This invention adopts a role-stage driven communication probability model, which introduces node role weights and task stages into the communication layer activation process, so that the information aggregation stage and the task execution stage correspond to different information flow priorities, which can reflect the collaborative contributions of information collection nodes, coordination decision nodes and task execution nodes at different stages.

[0042] This invention takes the effective closed-loop task chain as the core evaluation object, and unifies the local communication utility of nodes, effective edges of the task layer, single node performance of the task layer, and comprehensive task performance into the same resilience evaluation framework, which facilitates the location of broken task chains, weak role nodes, and collaborative relationships that need to be prioritized for recovery.

[0043] This invention calculates task resilience by using threshold gating and time-discounted integrals, which can reflect both the constraint of the minimum task feasibility threshold on task success and the time value of rapid recovery after disturbance, thus obtaining a more accurate quantitative result of resilience that is consistent with the task process.

[0044] The adaptive reconfiguration strategy of this invention aims at task chain recovery and can output the performance degradation curve after disturbance, the set of executable task chains, the node position adjustment target, and the enhanced reconfiguration scheme. It provides quantitative basis and operable support for the redundant configuration, self-healing access, and role replacement design of heterogeneous UAV swarms. Attached Figure Description

[0045] Figure 1This is a four-step flowchart of the communication coupling network resilience assessment and enhancement method based on closed-loop task chain of the present invention. It is used to illustrate the processing order between the acquisition of task nodes, communication nodes and cross-layer access relationships, construction of communication coupling network and effective transmission matrix, statistics of effective number of closed-loop task chains and calculation of resilience retention rate, and identification of vulnerable nodes and output of enhancement and reconstruction scheme.

[0046] Figure 2 The graph shows the system performance retention rate under random attacks for different communication topologies.

[0047] Figure 3 A comparison chart of overall resilience against random attacks, degree-first attacks, and betweenness-first attacks under a BA communication topology;

[0048] Figure 4 A comparison chart of the overall resilience of different target node types under priority attacks;

[0049] Figure 5 Radar chart showing the closed-loop task chain retention rate under different attack strategies when the node removal rate is 15.0%;

[0050] Figure 6 A vulnerability heatmap of the target node type and closed-loop task chain when the node removal rate is 15.0%. Detailed Implementation

[0051] To make the technical means, inventive features, objectives, and effects of the invention readily understandable, the invention is further described below with reference to specific illustrations. However, the invention is not limited to the embodiments described below.

[0052] It should be understood that the structures, proportions, sizes, etc., shown in the accompanying drawings of this specification are only used to complement the content disclosed in the specification for those skilled in the art to understand and read, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0053] Example 1:

[0054] like Figure 1 As shown, the present invention provides a task-oriented heterogeneous UAV swarm resilience assessment and adaptive reconfiguration method, comprising the following steps:

[0055] Step 1: Obtain the task nodes, communication nodes, cross-layer access relationships, node roles, node locations, and communication radius of the heterogeneous drone swarm. Task phases, closed-loop task chain patterns, and disturbance information;

[0056] Based on the obtained node roles, the surviving task nodes are divided into a set of information collection nodes. Set of Coordination Decision Nodes and task execution node set It also preserves the cross-layer access relationship between task nodes and communication nodes; the node role division is as follows:

[0057]

[0058] in, Denotes the set of surviving task nodes at time t. This represents the set of information collection nodes that are alive at time t. Denotes the set of coordination decision nodes that are alive at time t. Let represent the set of task execution nodes that are alive at time t, and let “∪” represent the union operation of sets. The node role classification formula is used to divide the surviving UAV nodes into three roles: information collection, coordination and decision-making, and task execution.

[0059] The task phase indicator function is as follows:

[0060]

[0061] in, This represents the task phase indicator function at time t. This indicates the moment when the information aggregation phase transitions to the task execution phase. Corresponding to the information aggregation stage, This corresponds to the task execution phase. The task phase indicator function is used to determine whether the current collaborative task is in the information aggregation phase or the task execution phase: When in the information aggregation phase, the information acquisition node collects target, environmental, and situational information and aggregates this information to the coordination and decision-making node. The communication requirements focus on information fusion between perception and decision-making, perception and perception, and within the decision-making process. When in the task execution phase, the coordination and decision-making node issues target allocation, attack, or interception commands to the task execution node. After execution, the task execution node can provide feedback on the effects to the information acquisition node or the coordination and decision-making node. The communication requirements focus on the transition from decision-making to execution and feedback on execution results.

[0062] Step 2: Construct a task communication coupled network (a three-layer coupled network with shared nodes), and calculate the end-to-end communication reachability, logical compatibility matrix, and effective transmission matrix;

[0063] The physical layer describes geometric reachability determined by spatial distance and communication radius; the communication layer describes directed random communication activation determined by distance decay and role-stage logical matching; the task layer is not an independently pre-set third set of node networks, but is generated under the conditions that the physical layer is reachable, the communication layer is activated, and the role transfer satisfies the OODA information flow rules, and is used to describe effective collaborative relationships that can form a closed-loop task chain.

[0064] Compared with existing schemes that evaluate the resilience of UAV swarms by using the dependencies between the communication layer, the structural layer, and the task layer, the shared node three-layer coupled network of the present invention unifies the state of the same UAV node in the physical layer, the communication layer, and the task layer into the same set of surviving nodes. Physical damage directly changes the availability of the node in the three layers, while functional suppression changes the communication activation probability and the effective edges in the task layer. Therefore, it can distinguish the failure mode of "the node is still there but the task chain is broken".

[0065] The aforementioned shared node modeling also reduces the additional setting of inter-layer dependency edges, so that the effective edges of the task layer are triggered by three types of conditions: physical reachability, communication reliability, and role logic. The resulting technical effect is that it can maintain the interpretable correspondence between structural connectivity, communication activation, and task closure, and can directly serve the location of broken task chains after disturbance and the selection of adaptive reconstruction targets.

[0066] In step 2, the activation probability of the directed link is calculated based on the distance between nodes and the role-phase logical matching. And generate effective edges for the task layer based on three conditions: physical reachability, reliable communication, and functional effectiveness;

[0067] Building the physical layer When the Euclidean distance between two nodes is... Not greater than the communication radius Then, undirected edges are established in the physical layer:

[0068]

[0069] in, This represents the physical reachability term between node i and node j at time t. This represents the Euclidean distance between the two at time t. Indicates the maximum communication radius. This is the distance attenuation piecewise coefficient; when the distance is less than The link is considered fully reachable when the distance is within a certain range. to The decay rate decreases linearly with distance between points, and decreases when the distance exceeds a certain threshold. The link is unreachable.

[0070] The physical reachability term definition is used to incorporate distance attenuation and maximum communication radius into link reliability.

[0071] Building a communication layer At that time, the directed communication activation probability from node i to node j is calculated to simultaneously characterize physical reachability and task logical matching:

[0072]

[0073]

[0074] in, express Time Node To the node The directed communication activation probability, Represents physically reachable terms. Indicates the task logic matching item; and Representing nodes respectively and nodes The role encoding vectors have vector components corresponding to three types of roles: information gathering, coordination and decision-making, and task execution. express Transpose of; Indicates the task phase A defined role-stage weight matrix is ​​used to increase the weight of the convergence links within the perception-to-decision, perception-to-perception, and decision-making stages during the information aggregation stage, and to increase the weight of the links from decision to execution and the feedback of execution results during the task execution stage. The directed link activation probability is obtained by multiplying the above task logical matching terms by the physical reachability terms, thus ensuring that the communication layer edges reflect both spatial distance and task role matching.

[0075] The task layer is constructed in the following order: physical layer constraints, communication layer activation, and task layer generation. First, the physical reachability between nodes is determined. Then, the directed communication activation probability is calculated. Finally, directed edges in the task layer are retained only when the directed communication activation probability reaches the task communication threshold and the role transition satisfies the OODA information flow rules. Thus, the task layer is a set of valid task relationships after the constraints of the other two layers are satisfied.

[0076] Step 3: Count the effective number of closed-loop task chains in the damage evolution scenario, and calculate the physical layer performance for each chain. Communication layer performance and task layer performance And aggregated to obtain comprehensive task performance ;

[0077] Physical layer performance Reflecting node survival and physical topology quality, it is used to fuse node survival rate, average degree, clustering, and global efficiency, and is defined as follows:

[0078]

[0079] in, express Physical layer performance at any time Indicates node survival rate. and They represent The average degree of time relative to the initial time, and They represent Clustering coefficients at time step and the initial time step. and They represent Global efficiency at time step and initial time step These are the weighting coefficients for average degree, clustering coefficient, and global efficiency, respectively.

[0080] Communication layer performance Reflects communication strength and orderliness:

[0081]

[0082]

[0083]

[0084] in, This is the definition of communication entropy, used to reflect the degree of dispersion of communication objects among nodes. This is a normalized communication probability formula used to form the outbound communication distribution of nodes. express Real-time communication layer performance, Represents the normalization coefficient of the communication layer. express The number of nodes that are alive at any given time. Represents a node To the node The directed communication activation probability, The exponential term represents the sensitivity coefficient of the communication probability mapping. Indicates the upper limit of communication entropy. It represents the natural logarithm.

[0085] Task layer performance It is used to measure the degree to which the current effective task chain is preserved relative to the initial state, and is defined as follows:

[0086]

[0087] in, This represents the task layer performance at time t. express The sum of the total task efficiency of all effective closed-loop task chains at any given moment. This represents the sum of the task performance of all valid closed-loop task chains at the initial moment. Task performance is calculated as follows:

[0088]

[0089] in, Indicates the execution node exist The performance of a single node in the task layer at any given time. express The execution node is always present. A set of effective closed-loop task chains, express Any valid closed-loop task chain in the process, Represents a task chain exist The link length or hop count at any given time; the maximum value indicates that the shorter, more efficient closed-loop task chain is prioritized to characterize the task contribution of the execution node. The single-node performance at the task layer is used to evaluate the shortest effective closed-loop task chain length that execution node i can participate in; if the set of effective task chains to which execution node i belongs is empty, then the single-node performance at the task layer for that execution node is 0. This formula is used to calculate task layer performance: first, the single-node performance of each execution node is calculated; then, the single-node performance of all execution nodes is summed to obtain the total task layer performance; finally, it is substituted into the task layer performance formula to calculate the degree to which the current effective task chain is preserved relative to the initial state. Therefore, this formula is used both to quantify task layer performance and to locate broken task chains and execution nodes that need to be prioritized for recovery.

[0090] The above three performance categories are aggregated into a comprehensive task performance based on their weights. :

[0091]

[0092] in, express Overall task performance at all times , , These represent the aggregated weights of physical layer performance, communication layer performance, and task layer performance, respectively. These three weights are used to adjust the contribution of different layer performances to the overall evaluation based on task requirements.

[0093] When modeling disturbances, physical damage and functional suppression disturbances are modeled separately;

[0094] When physical damage occurs, delete the corresponding node and its associated edges; when functional suppression occurs, reduce or zero the number of edges associated with the suppressed node. This demonstrates the situation where the communication logic fails but the physical node still exists.

[0095] After completing the perturbation update, the task resilience index is calculated using threshold gating and time-discounted integral. It is used to evaluate the task retention and recovery capability after perturbation, and is defined as follows:

[0096]

[0097] Where R represents the mission resilience index. Indicates the time when the disturbance occurred. Indicates the time when the assessment is terminated or resumed. This represents the minimum feasible threshold for the task. This represents the overall task performance under ideal, unperturbed conditions. This represents the time discount factor. This means that only effective performance contributions above the minimum feasible threshold are counted.

[0098] Step 4: Identify vulnerable nodes and perform enhanced reconstruction based on local communication utility and node role constraints: First, identify the broken closed-loop task chain and its corresponding failed nodes or failed communication edges after the disturbance. Then, select the reconstruction target that can restore the perception-decision-execution closed loop from the candidate neighbors or candidate recovery nodes, update the UAV node position and network status, and output the resilience assessment results, vulnerable nodes and enhanced reconstruction scheme.

[0099] During the recovery process, this invention defines local communication utility and selects reconstruction targets in conjunction with node role constraints. The so-called "identification of broken task chains after perturbation" refers to finding task chains that could originally be closed but are currently unable to complete the closed loop of perception, decision-making, execution, and feedback based on the set of effective edges in the task layer after perturbation, and identifying the key nodes or links that caused the break; then, priority is given to restoring candidate nodes that can fill the broken link, rather than simply increasing the density of general topological edges.

[0100] After the disturbance, calculate the local communication utility of candidate neighbors or candidate recovery nodes. This is used to select task-related recovery neighbors or target nodes for adaptive reconstruction, and is calculated as follows:

[0101]

[0102] in, Indicates candidate nodes exist The local communication efficiency at any given moment Represents a node Its adjacent nodes The activation probability of directed communication between them Represents a node Communication entropy, Indicates the upper limit of communication entropy. This is the sensitivity coefficient for the communication probability mapping; this formula is used to prioritize candidate nodes with reliable communication and relatively ordered distribution of communication objects during reconstruction.

[0103] Adaptive reconstruction combined with node role constraints includes: for nodes that still have communicating neighbors, prioritizing local communication utility. The largest node is selected as the target for recovery and collaboration; for isolated nodes, the target area or target node that can restore the perception-decision-execution task chain is selected based on its role.

[0104] Update node position based on selected target. Recalculate network state and Output the task resilience index after perturbation And the set of executable task chains after reconstruction.

[0105] The experimental verification and comparison results are as follows:

[0106] In terms of experimental setup, controlled variable simulations were conducted based on a task-communication coupled network and eight types of closed-loop task chains. The communication layer adopted BA scale-free networks, ER random networks, and WS small-world networks, respectively. The perturbation methods included random attacks, degree-priority attacks, betweenness-priority attacks, and targeted attacks for different node types. The target node types included communication nodes, information acquisition nodes, coordination and decision-making nodes, task execution nodes, and all task nodes. At the same time, the scale ratio of communication nodes to task nodes was changed to verify the impact of changes in the scale of the communication infrastructure on the closed-loop task maintenance capability.

[0107] like Figure 2 As shown, under random attacks, the system performance of BA, ER, and WS communication topologies all decreases as the proportion of communication nodes removed increases, but the rate of decrease is different. Among them, the BA topology has a certain redundancy absorption capacity under random failures, while the ER and WS topologies are more likely to fail prematurely due to local path interruption. This indicates that the present invention can distinguish the impact of different communication infrastructures on the closed-loop task maintenance capability.

[0108] like Figures 3 to 6 As shown, in the BA communication topology, degree-first and betweenness-first attacks cause a sharp drop in system performance earlier than random attacks, reflecting the "communication infrastructure veto effect" of highly centralized communication nodes. When the attack target changes to different task node types, the coordination decision node has a significant truncation effect on the closed-loop task chain, while the information collection node and task execution node exhibit certain functional redundancy and substitutability. Radar charts and heatmaps further show that, under the same node removal ratio, different closed-loop task chains have significant differences in sensitivity to target node types and attack strategies.

[0109] To address the aforementioned vulnerabilities, multi-homing redundancy, dynamic self-healing, and role substitution strategies can be introduced to generate enhanced reconfiguration schemes. Among these, targeted multi-homing redundancy for coordination decision nodes can improve the communication access redundancy of key decision nodes, dynamic self-healing can alleviate the cascading isolation of task nodes caused by communication access loss, and role substitution can form a simplified closed-loop task chain during severe disturbance phases, thereby improving the lower bound of residual task performance.

[0110] Compared with methods that only use node survival rate, connected component size, average path length, or physical layer reachability as resilience indicators, the comparison results of this invention can directly reflect the impact of communication node failure, decision node truncation, and closed-loop task chain breakage on task completion capability. Therefore, in terms of technical effect, this invention can not only provide a structural judgment on whether the system is connected, but also provide a task-oriented judgment on whether the task chain is executable, which role nodes need to be prioritized for enhancement, and which reconstruction strategy is more effective.

[0111] Example 2:

[0112] A mission-oriented heterogeneous UAV swarm resilience assessment and adaptive reconfiguration system includes:

[0113] The data acquisition module is used to acquire the node role, node location, and communication radius of each drone node in the heterogeneous drone swarm; to divide the set of surviving task nodes into three role sets: information collection nodes, coordination and decision-making nodes, and task execution nodes; and to acquire the current task stage and disturbance information.

[0114] The Coupled Network Construction Module is used to construct a task communication coupled network, including a physical layer, a communication layer, and a task layer, on the same set of surviving task nodes. The physical layer describes the physical reachability between nodes; the communication layer describes the directed random communication activation between nodes; and the task layer describes the effective collaborative relationship between nodes that can form a closed-loop task chain, provided that the physical layer is reachable, the communication layer is activated, and the role transfer satisfies the OODA information flow rules.

[0115] The resilience assessment module is used to calculate the physical layer performance, communication layer performance, and task layer performance based on the task communication coupled network, and then weight and aggregate them to obtain the comprehensive task performance; it performs perturbation modeling and calculates the task resilience index based on the comprehensive task performance to evaluate the task's ability to maintain and recover after perturbation.

[0116] The enhanced reconstruction module is used to identify broken closed-loop task chains and their corresponding failed nodes or failed communication edges based on the set of valid edges in the task layer after the disturbance; calculate the local communication utility of candidate recovery nodes and perform enhanced reconstruction in combination with node role constraints; update node positions and network status; and output the task resilience index after the disturbance and the enhanced reconstruction scheme.

Claims

1. A task-oriented method for assessing the resilience of heterogeneous UAV swarms and for adaptive reconfiguration, characterized in that, Includes the following steps: S1. Obtain the node role, node location, and communication radius of each drone node in the heterogeneous drone swarm; divide the set of surviving task nodes into three role sets: information collection nodes, coordination and decision-making nodes, and task execution nodes, and obtain the current task stage and disturbance information; S2. Construct a task communication coupling network, including a physical layer, a communication layer, and a task layer, on the same set of surviving task nodes; The physical layer describes the physical reachability between nodes; the communication layer describes the directed random communication activation between nodes; and the task layer describes the effective collaborative relationship between nodes that can form a closed-loop task chain, provided that the physical layer is reachable, the communication layer is activated, and the role transfer satisfies the OODA information flow rules. S3. Based on the task-communication coupled network, calculate the physical layer performance, communication layer performance and task layer performance respectively, and weighted aggregate to obtain the comprehensive task performance; perform disturbance modeling and calculate the task resilience index based on the comprehensive task performance to evaluate the task's ability to maintain and recover after disturbance. S4. Identify the broken closed-loop task chain and its corresponding failed nodes or failed communication edges based on the set of effective edges in the task layer after the disturbance. Calculate the local communication utility of candidate recovery nodes and perform enhanced reconstruction in combination with node role constraints. Update node positions and network status, and output the task resilience index and enhanced reconstruction scheme after the disturbance.

2. The method for assessing and adaptively reconstructing the resilience of heterogeneous UAV swarms according to claim 1, characterized in that, S2 include: The physical reachability terms of the physical layer are determined based on the Euclidean distance and communication radius between each node; the task logical matching terms are determined based on the node role code and the role-stage weight matrix corresponding to the current task stage; the physical reachability terms and the task logical matching terms are multiplied to obtain the directed communication activation probability between each pair of nodes; when the directed communication activation probability reaches the task communication threshold and the role transfer between nodes satisfies the task observation, judgment and decision information flow rules, the corresponding directed edges are retained in the task layer to form an effective task relationship set; when physical damage occurs, the availability of the corresponding node in the three layers is changed; when functional inhibition occurs, the directed communication activation probability and the effective edges of the task layer are changed.

3. The method for assessing and adaptively reconstructing the resilience of heterogeneous UAV swarms according to claim 2, characterized in that, In S2, the physical reachability terms of the physical layer are determined based on the Euclidean distance and communication radius between nodes, including: When the Euclidean distance between two nodes is less than the product of the distance attenuation segmentation coefficient and the communication radius, the physical reachability is fully reachable; when the Euclidean distance between two nodes is greater than or equal to the product of the distance attenuation segmentation coefficient and the communication radius but less than or equal to the communication radius, the physical reachability decreases linearly with respect to distance; when the Euclidean distance between two nodes is greater than the communication radius, the physical reachability is unreachable.

4. The method for assessing and adaptively reconstructing the resilience of heterogeneous UAV swarms according to claim 2, characterized in that, In S2, the task logic matching items are determined based on the node role code and the role-stage weight matrix corresponding to the current task stage, including: Each node's role is represented as a three-dimensional vector, with each vector component corresponding to one of the three roles: information gathering, coordination and decision-making, and task execution. A role-stage weight matrix is ​​determined based on the current task stage: when in the information aggregation stage, the weights of the convergence links between information gathering nodes and coordination and decision-making nodes, between information gathering nodes, and between coordination and decision-making nodes are increased; when in the task execution stage, the weights of the links between coordination and decision-making nodes and task execution node feedback links are increased. The transpose of the sending node's role encoding vector, the role-stage weight matrix, and the receiving node's role encoding vector are multiplied sequentially to obtain the task logic matching term.

5. The method for assessing and adaptively reconstructing the resilience of heterogeneous UAV swarms according to claim 1, characterized in that, S3 calculates the physical layer performance, communication layer performance, and task layer performance separately, including: The physical layer performance is obtained by fusing node survival rate, average degree, clustering coefficient and global efficiency. The average degree, clustering coefficient and global efficiency are respectively calculated by weighting the ratio of their current value to their initial value, and then multiplied by the node survival rate. The communication layer performance is obtained as follows: For each surviving task node, calculate the communication entropy of its outgoing communication probability normalized distribution, take the complement of the ratio of the communication entropy to the upper limit of the communication entropy, multiply it by the communication strength mapping value, and then calculate the mean for all surviving task nodes; where the communication strength mapping value is represented by a negative exponential function of the sum of communication probabilities; The task layer performance is the ratio of the total task efficiency of all valid closed-loop task chains at the current moment to the total task efficiency of all valid closed-loop task chains at the initial moment; where the task efficiency of each execution node is the maximum value of the reciprocal of the link length among all valid closed-loop task chains it participates in.

6. The method for assessing and adaptively reconstructing the resilience of heterogeneous UAV swarms according to claim 1, characterized in that, In S3, disturbances include physical damage disturbances and functional suppression disturbances; When physical damage occurs, delete the damaged node and its associated edges; When functional suppression occurs, the directed communication activation probability associated with the suppressed node is reduced or set to zero. This causes the suppressed node to remain alive in the physical layer, but its directed communication activation probability in the communication layer decreases, thereby breaking the corresponding effective edge in the task layer.

7. The method for assessing and adaptively reconstructing the resilience of heterogeneous UAV swarms according to claim 1, characterized in that, In S3, the task resilience index is calculated as follows: Taking the time of disturbance occurrence as the starting point of integration and the time of evaluation termination or the time of recovery of observation as the end point of integration, the non-negative part of the difference between the comprehensive task performance and the minimum task feasibility threshold is taken as the effective performance contribution of the integrand. The effective performance contribution term is weighted by exponential decay using a time discount factor; The ratio of the integral value of the weighted effective performance contribution item over the integral interval to the corresponding integral value under the ideal state is used as the task resilience index.

8. The method for assessing and adaptively reconstructing the resilience of heterogeneous UAV swarms according to claim 1, characterized in that, In S4, the local communication utility of candidate recovery nodes is calculated and enhanced reconstruction is performed in conjunction with node role constraints, including: For each candidate recovery node, calculate the communication entropy of its outgoing communication probability normalized distribution, take the complement of the ratio of the communication entropy to the upper limit of the communication entropy, and multiply it with the communication strength mapping value. Based on the multiplication result and the directed communication activation probability between the candidate recovery node and its neighboring nodes, obtain the local communication utility. For nodes that still have communication neighbors, select the node with the highest local communication utility as the target for restoring cooperation; For isolated nodes, select the target area or target node that can restore the closed-loop task chain based on its role; Update node positions based on the selected reconstruction target, recalculate network state and overall task performance, and output the perturbed task resilience index and the set of executable task chains after reconstruction.

9. A task-oriented heterogeneous UAV swarm resilience assessment and adaptive reconfiguration system, used to implement the task-oriented heterogeneous UAV swarm resilience assessment and adaptive reconfiguration method as described in any one of claims 1-8, characterized in that, include: The data acquisition module is used to acquire the node role, node location, and communication radius of each drone node in a heterogeneous drone swarm. The set of surviving task nodes is divided into three roles: information collection nodes, coordination and decision-making nodes, and task execution nodes, and the current task stage and disturbance information are obtained. The Coupled Network Construction Module is used to construct a task communication coupled network, including a physical layer, a communication layer, and a task layer, on the same set of surviving task nodes. The physical layer describes the physical reachability between nodes; the communication layer describes the directed random communication activation between nodes; and the task layer describes the effective collaborative relationship between nodes that can form a closed-loop task chain, provided that the physical layer is reachable, the communication layer is activated, and the role transfer satisfies the OODA information flow rules. The resilience assessment module is used to calculate the physical layer performance, communication layer performance, and task layer performance based on the task communication coupled network, and then weight and aggregate them to obtain the comprehensive task performance; it performs perturbation modeling and calculates the task resilience index based on the comprehensive task performance to evaluate the task's ability to maintain and recover after perturbation. The enhanced reconstruction module is used to identify broken closed-loop task chains and their corresponding failed nodes or failed communication edges based on the set of valid edges in the task layer after the disturbance; calculate the local communication utility of candidate recovery nodes and perform enhanced reconstruction in combination with node role constraints; update node positions and network status; and output the task resilience index after the disturbance and the enhanced reconstruction scheme.

Images