A local rapid reconstruction method for a killing net based on a complementary net strategy

Abstract

The present application belongs to the field of military command control and combat simulation, and relates to a local rapid reconstruction method for a killing network based on a complementary network strategy. The method acquires current state information of the killing network and detects failed nodes or edges, constructs a local repair area centered on the failed nodes, establishes a multi-objective optimization model in the repair area aiming to maximize task recovery benefits, minimize resource use costs and minimize task completion delays, solves to obtain a Pareto optimal solution set by using a TM-NSGA-III algorithm guided by a conversion matrix, and finally screens and executes a reconstruction scheme according to a battlefield situation. The present application reduces the search space by localized modeling, combines the conversion matrix to inherit the effective information of the original network, realizes rapid and high-quality reconstruction of the killing network, and improves the resilience and self-healing ability of the killing network in a dynamic battlefield environment.

Description

Technical Field

[0001] This invention belongs to the field of military command and control and combat simulation technology, and relates to a method for rapid recovery and reconstruction of a kill network after being subjected to a local attack, especially a method for rapid local reconstruction of a kill network based on a network replenishment strategy. Background Technology

[0002] As a new operational concept, the military kill network forms a resilient and adaptive operational system by dynamically interconnecting heterogeneous nodes for reconnaissance, command, and strike, in order to cope with highly dynamic and highly contested battlefield environments. However, during combat, critical nodes and links of the kill network are susceptible to local failure due to enemy attacks, which can lead to the interruption of critical mission links, affecting operational continuity and system effectiveness.

[0003] Existing methods for reconstructing damaged kill networks mostly employ a "global reconstruction" paradigm, which involves re-optimizing and generating a new network structure within the global solution space. While these methods can theoretically achieve global optimality, they are typically computationally complex and time-consuming, making them unsuitable for real-time battlefield decision-making. Furthermore, global reconstruction often disrupts the existing collaborative relationships and structural advantages of the original network, leading to unnecessary reconfigurations of operational organization and command and control processes, thus reducing recovery efficiency. For these reasons, there is an urgent need for a localized, rapid reconstruction method capable of quickly restoring the kill network's operational capabilities after damage, while preserving as much of the original network's healthy structural and mission information as possible. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for local rapid reconstruction of the kill network based on a network replenishment strategy.

[0005] To achieve the objectives of this invention, the following technical solutions are adopted.

[0006] A method for rapid local reconstruction of a kill network based on a network replenishment strategy includes the following steps:

[0007] S1: Obtain the current status information of the kill net and detect whether there are any failed nodes or failed edges in the network;

[0008] S2: Upon detecting a failed node or edge, a local repair zone centered on the failed node is constructed based on a network patching strategy, and the reconstruction decision variables are confined within this local repair zone; the localization modeling concept corresponding to this step can be combined with... Figure 1 Understanding the diagram in (b); its differences from the traditional global reconstruction paradigm can be combined with... Figure 1 Compare and contrast figures (a) and (b) in the text to understand the meaning.

[0009] S3: Establish a multi-objective optimization model within the local repair area. The multi-objective optimization model aims to maximize the task recovery benefits, minimize the resource usage costs, and minimize the task completion delay.

[0010] S4: The TM-NSGA-III algorithm guided by the transformation matrix is ​​used to solve the multi-objective optimization model and obtain the Pareto optimal solution set;

[0011] S5: Evaluate the Pareto optimal solution set, determine the reconstruction plan to be executed based on the battlefield situation, perform network topology update, and complete the local reconstruction of the kill network.

[0012] Furthermore, the specific process of constructing the local repair zone includes the following steps:

[0013] S21. Determine the neighborhood range based on the hop count constraint criterion, with the failed node as the center; where: hop count constraint criterion: define the hop neighborhood of each damaged node based on the original network topology.

[0014] S22. Within the neighborhood, candidate nodes with alternative capabilities are selected based on the functional relevance criterion; wherein: functional relevance criterion: for damaged nodes of a specific functional type, available nodes with similar functions are included in the repair area;

[0015] S23. Select nodes that are physically reachable based on resource reachability criteria; where: resource reachability criteria: consider the physical location and connection constraints between nodes to ensure the spatial feasibility of the repair solution;

[0016] The set of nodes and their connecting edges that satisfy the above conditions is defined as the local repair region.

[0017] Furthermore, the multi-objective optimization model includes the following objective function:

[0018] S31. Maximize task recovery benefits:

[0019] ;

[0020] in, Indicates task priority weight. This indicates the task recovery status variable. Represents the set of failed tasks;

[0021] S32. Minimize resource usage cost:

[0022] ;

[0023] in, Indicates the cost of enabling a node. This indicates the cost of establishing and maintaining the link. Represents the set of nodes within the local repair region. This represents the set of edges within the local repair region. This indicates the decision variable for node selection. Indicates the decision variables for link establishment;

[0024] S33. Minimize task completion delay:

[0025] ;

[0026] in, Indicates link transmission delay. Indicates task In the link The allocation decision variables.

[0027] Furthermore, the method for constructing the transformation matrix-guided TM-NSGA-III algorithm includes the following steps:

[0028] S41. Construct a functional substitution transition matrix between failed nodes and candidate nodes to characterize the confidence level of candidate nodes in functional substitution of failed nodes.

[0029] S42. Generate an initial population based on the functional substitution transformation matrix, prioritizing node combinations with high substitution confidence as initial solutions; where: the initial population generation process based on the transformation matrix is ​​as follows: Figure 2 As shown;

[0030] S43. Introduce local repair zone constraints in genetic operations so that candidate solutions only contain nodes and edges within the local repair zone;

[0031] S44. Based on the functional substitution transformation matrix, perform targeted mutation. In the mutation operation, prioritize replacing nodes with high substitution confidence in the functional substitution transformation matrix to guide the search direction.

[0032] Furthermore, the specific method for generating the initial population based on the transformation matrix.

[0033] S51. Extract path seeds for each interrupted task flow. Extract its position in the original network from historical data. Optimal historical path ;

[0034] S52, Map the core path and traverse it. For each node, if node If it is invalid, then query the transformation matrix for the first... Okay, choose alternative confidence level The highest candidate node Replace the faulty node; if the node is healthy, retain it. Through this process, a core repair path within the repair area is generated for each interrupted task flow, and it is encoded as an initial solution.

[0035] S53. Inject population diversity by performing controlled random perturbation operations on all core repair path individuals obtained above, with a low probability. Randomly replace a node or edge in the path (replacement options are limited to feasible resources within the repair zone) to generate a batch of perturbation individuals to enhance the population's exploration capabilities;

[0036] S54. Assemble the initial population by merging all unmutated core repair path individuals with mutated perturbation individuals to form the initial population for the multi-objective optimization algorithm. .

[0037] Furthermore, the method for determining the refactoring scheme to be executed includes the following steps:

[0038] S61. Calculate the comprehensive score of each candidate solution in the Pareto optimal solution set using the comprehensive scoring method. The scoring dimensions include task recovery rate, resource cost ratio and task completion delay compliance rate.

[0039] S62. Adjust the weight of each scoring dimension dynamically based on the current battlefield situation and the emphasis on timeliness or economy.

[0040] S63. Select the candidate solution with the best score as the final execution plan, and perform network topology update to complete the reconstruction.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. By forcibly localizing the repair area, the reconstruction search space is significantly reduced, avoiding large-scale calculations for global reconstruction, improving solution efficiency and response speed, and meeting the needs of real-time battlefield decision-making.

[0043] 2. By inheriting the topology and mission information of the original network health part through the transformation matrix, the disruption to existing cooperative relationships is reduced, unnecessary resetting of combat organization and command and control processes is reduced, and recovery stability is improved.

[0044] 3. By achieving a comprehensive balance between mission recovery benefits, resource costs, and completion delay through multi-objective joint optimization, multiple Pareto optimal solutions are generated to adapt to different battlefield situation requirements and enhance the resilience and self-healing capability of the kill net in dynamic battlefield environments.

[0045] 4. The algorithm guides the initial population generation and directional mutation through the transformation matrix, further improving the quality of the solution set and the convergence speed, and ensuring the scientific validity and effectiveness of the reconstruction scheme. Attached Figure Description

[0046] Figure 1 The diagrams show a comparison between global reconstruction and local reconstruction based on the network supplementation strategy; where: (a) is a schematic diagram of global reconstruction; (b) is a schematic diagram of local reconstruction based on the network supplementation strategy.

[0047] Figure 2 The initial population generation graph is based on the transformation matrix;

[0048] Figure 3 Provide a network topology diagram for the damaged kill net equipment;

[0049] Figure 4 A comparison chart of HV iteration curves for different algorithms;

[0050] Figure 5 Comparison of HV iteration curves under different algorithm configurations;

[0051] Figure 6 Comparison of IGD iteration curves under different algorithm configurations;

[0052] Figure 7 For restructuring scheme A: minimum average decision delay;

[0053] Figure 8 For Restructuring Option B: Lowest Average Restructuring Cost. Detailed Implementation

[0054] The present invention will be further described in conjunction with the accompanying drawings and embodiments.

[0055] As an embodiment 1 of the present invention, such as Figure 1 As shown, a method for rapid local reconstruction of a kill network based on a network replenishment strategy includes the following steps:

[0056] S1: Obtain the current status information of the kill net and detect whether there are any failed nodes or failed edges in the network;

[0057] S2: Upon detecting a failed node or edge, a local repair zone centered on the failed node is constructed based on a network patching strategy, and the reconstruction decision variables are confined within this local repair zone; the localization modeling concept corresponding to this step can be combined with... Figure 1 Understanding the diagram in (b); its differences from the traditional global reconstruction paradigm can be combined with... Figure 1 Compare and contrast figures (a) and (b) in the text to understand the meaning.

[0058] S3: Establish a multi-objective optimization model within the local repair area. The multi-objective optimization model aims to maximize the task recovery benefits, minimize the resource usage costs, and minimize the task completion delay.

[0059] S4: The TM-NSGA-III algorithm guided by the transformation matrix is ​​used to solve the multi-objective optimization model and obtain the Pareto optimal solution set;

[0060] S5: Evaluate the Pareto optimal solution set, determine the reconstruction plan to be executed based on the battlefield situation, perform network topology update, and complete the local reconstruction of the kill network.

[0061] As an embodiment 2 of the present invention, the specific process of constructing the local repair area includes the following steps:

[0062] S21. Determine the neighborhood range based on the hop count constraint criterion, with the failed node as the center; where: hop count constraint criterion: define the hop neighborhood of each damaged node based on the original network topology.

[0063] S22. Within the neighborhood, candidate nodes with alternative capabilities are selected based on the functional relevance criterion; wherein: functional relevance criterion: for damaged nodes of a specific functional type, available nodes with similar functions are included in the repair area;

[0064] S23. Select nodes that are physically reachable based on resource reachability criteria; where: resource reachability criteria: consider the physical location and connection constraints between nodes to ensure the spatial feasibility of the repair solution;

[0065] The set of nodes and their connecting edges that satisfy the above conditions is defined as the local repair region.

[0066] As an embodiment 3 of the present invention, the multi-objective optimization model includes the following objective function:

[0067] S31. Maximize task recovery benefits:

[0068] ;

[0069] in, Indicates task priority weight. This indicates the task recovery status variable. Represents the set of failed tasks;

[0070] S32. Minimize resource usage cost:

[0071] ;

[0072] in, Indicates the cost of enabling a node. This indicates the cost of establishing and maintaining the link. Represents the set of nodes within the local repair region. This represents the set of edges within the local repair region. This indicates the decision variable for node selection. Indicates the decision variables for link establishment;

[0073] S33. Minimize task completion delay:

[0074] ;

[0075] in, Indicates link transmission delay. Indicates task In the link The allocation decision variables.

[0076] As an embodiment 4 of the present invention, the method for constructing the transformation matrix-guided TM-NSGA-III algorithm includes the following steps:

[0077] S41. Construct a functional substitution transition matrix between failed nodes and candidate nodes to characterize the confidence level of candidate nodes in functional substitution of failed nodes.

[0078] S42. Generate an initial population based on the functional substitution transformation matrix, and prioritize the node combination with high substitution confidence as the initial solution;

[0079] S43. Introduce local repair zone constraints in genetic operations so that candidate solutions only contain nodes and edges within the local repair zone;

[0080] S44. Based on the functional substitution transformation matrix, perform targeted mutation. In the mutation operation, prioritize replacing nodes with high substitution confidence in the functional substitution transformation matrix to guide the search direction.

[0081] Embodiment 5 of the present invention, as follows Figure 2 As shown, the specific method for generating the initial population based on the transformation matrix includes the following steps:

[0082] S51. Extract path seeds for each interrupted task flow. Extract its position in the original network from historical data. Optimal historical path ;

[0083] S52, Map the core path and traverse it. For each node, if node If it is invalid, then query the transformation matrix for the first... Okay, choose alternative confidence level The highest candidate node Replace the faulty node; if the node is healthy, retain it. Through this process, a core repair path within the repair area is generated for each interrupted task flow, and it is encoded as an initial solution.

[0084] S53. Inject population diversity by performing controlled random perturbation operations on all core repair path individuals obtained above, with a low probability. Randomly replace a node or edge in the path (replacement options are limited to feasible resources within the repair zone) to generate a batch of perturbation individuals to enhance the population's exploration capabilities;

[0085] S54. Assemble the initial population by merging all unmutated core repair path individuals with mutated perturbation individuals to form the initial population for the multi-objective optimization algorithm. This strategy ensures that the initial population, while inheriting the excellent structure of the original network, possesses sufficient diversity to explore potentially better regions.

[0086] Embodiment 6 of the present invention provides a method for determining a reconstruction scheme to be executed, comprising the following steps:

[0087] S61. Calculate the comprehensive score of each candidate solution in the Pareto optimal solution set using the comprehensive scoring method. The scoring dimensions include task recovery rate, resource cost ratio and task completion delay compliance rate.

[0088] S62. Adjust the weight of each scoring dimension dynamically based on the current battlefield situation and the emphasis on timeliness or economy.

[0089] S63. Select the candidate solution with the best score as the final execution plan, and perform network topology update to complete the reconstruction.

[0090] As an embodiment 7 of the present invention, a method for rapid local reconstruction of a kill net based on a net replenishment strategy is verified in conjunction with a typical operational scenario of "joint fire strike in island and reef areas".

[0091] 1. Experimental scenario setup:

[0092] The initial kill network consists of 32 equipment entity nodes, including reconnaissance, decision-making, strike, and assessment types. A simulated blue team attack causes the red team's decision-making node C2 and strike node F1 to fail, triggering related link interruptions and forming a damaged network topology, such as... Figure 3 As shown.

[0093] 2. Implementation process:

[0094] S1: Status Acquisition and Repair Zone Construction: Monitor network status and determine the set of damaged nodes. and damaged edge set Based on the hop count constraint (R=2) and functional relevance (similarity threshold)... ) and resource accessibility (maximum collaboration distance) Three criteria for generating the repair area The repair zone contains healthy nodes and their connections that can be used to replace the functions of C2 and F1, such as decision nodes C1, C3, C4-C6 and strike nodes F2, M1-M6, SH1-SH2, P1-P2, etc.

[0095] S2: Model Building and Parameter Setting: In the Repair Area The invention establishes a multi-objective optimization model as described in the invention, and sets node activation cost and multi-task capability (see Table 1), link cost and latency parameters (formulas (1) and (2)), task priority weight (formula (3)) and maximum allowable task completion time according to the operational scenario.

[0096] Link cost calculation formula:

[0097] ;

[0098] in, Fixed costs for establishing and maintaining the link; The unit distance cost coefficient is taken as... , ; For nodes and The geographical distance between them.

[0099] Link delay calculation formula:

[0100] ;

[0101] In the formula: This represents the fixed latency caused by access, queuing, and protocol processing; The propagation delay coefficient per unit distance is taken as... , Link latency The total delay of a task is calculated by summing the delays of each link along its path, which is the single-hop transmission delay.

[0102] Task priority weight calculation formula:

[0103] ;

[0104] In the formula: For target threat level, and ; and These are the minimum and maximum values ​​for mission priority weights, set according to operational requirements.

[0105] Table 1 Node Costs and Multitasking Capabilities

[0106] node cost Multitasking capability node cost Multitasking capability C1 10 3 C2 10 3 C3 10 3 C4 10 3 C5 10 3 C6 10 3 A1 9 2 S1 9 2 F1 8 2 F2 8 2 SH1 8 2 SH2 8 2 M1 8 2 M2 8 2 M3 8 2 M4 8 2 M5 8 2 M6 8 2 P1 8 2 P2 8 2 U1 6 2 U2 6 2 U3 6 2 U4 6 2 R1 6 2 R2 6 2 R3 6 2 R4 6 2 SC1 5 2 SC2 5 2 O1 5 2 O2 5 2

[0107] S3: Algorithm Solution: The TM-INSGA-III algorithm is used to solve the model. Algorithm parameters are set as shown in Table 2: population size 100, number of iterations 300, crossover probability 0.7, mutation probability 0.1, etc. After the algorithm runs, it outputs a Pareto optimal solution set, which includes multiple feasible reconstruction schemes that make different trade-offs between task recovery benefits, resource costs, and completion delays.

[0108] Table 2 Algorithm Hyperparameter Settings

[0109] parameter numerical values parameter numerical values Population size 100 Number of iterations 300 Cross operator 0.7 Mutation Operator 0.1 Parallelism 3 Regularization parameters 0.05 Number of reference points 40 Convergence Criteria <![CDATA[10 -5 ]]> Diffusion factor distribution index 30 Disturbance factor distribution index 20

[0110] S4: Solution Analysis and Selection: Evaluate the obtained Pareto optimal solution set and select the execution plan based on the requirements of task timeliness, resource consumption, etc.

[0111] 3. Experimental verification and result analysis:

[0112] To verify the proposed mesh replacement strategy and the rapid reconstruction capability and solution set quality advantage of TM-INSGA-III after the kill net is damaged, three comparative methods were set up for experiments: ① Global reconstruction-NSGA-III (denoted as G-NSGA-III), which re-optimizes within the global solution space; ② Mesh replacement strategy + standard NSGA-III (denoted as P-NSGA-III), which uses a local reconstruction model and repair zone constraints, but without transformation matrix guidance; ③ The TM-INSGA-III proposed in this invention. All three methods used the unified algorithm parameters shown in Table 2, ran independently 20 times, and the mean and standard deviation of hypervolume HV and inversion generation distance IGD were statistically analyzed.

[0113] Table 3 Comparison of solution set quality indicators for different algorithms

[0114] algorithm HV (mean ± standard deviation) IGD (mean ± standard deviation) G-NSGA-Ⅲ 0.612±0.018 0.093±0.007 P-NSGA-Ⅲ 0.734±0.012 0.076±0.005 TM-INSGA-III 0.8402±0.009 0.058±0.004

[0115] Experimental results show that the method proposed in this invention has significant advantages over the traditional global reconstruction method (G-NSGA-III) and the standard algorithm (P-NSGA-III) which only uses the repair region constraint, in terms of solution set quality (higher HV value and lower IGD value), convergence speed (faster rise of HV iteration curve) and scheme efficiency (able to recover all failed functions with lower average decision delay and reconstruction cost).

[0116] Ablation experiments further validated the effectiveness of the transformation matrix-guided initialization and directed mutation strategies. The baseline algorithm was set as P-NSGA-III, and four algorithm configurations were formed by progressively adding key modules to it:

[0117] A1 is P-NSGA-III, which uses a patching strategy to constrain the solution space, but the initial population is generated completely randomly, and the mutation is a standard discrete mutation.

[0118] A2 is the initial population guided by the P-NSGA-III+ transformation matrix. Based on A1, the transformation matrix is ​​used to generate some high-quality individuals to form the initial population.

[0119] A3 is a P-NSGA-III+ targeted mutation strategy. Based on A1, it only replaces the standard mutation with a targeted mutation, while keeping the rest the same.

[0120] A4 is TM-INSGA-III, which simultaneously enables the transformation matrix to guide the initial population and the directional mutation strategy.

[0121] A comparison of convergence of different algorithms on the HV metric, such as Figure 4 As shown; ablation experimental results under different algorithm configurations, such as Figure 5 and Figure 6 As shown in the figure; the experimental results show that A4 has better convergence speed and solution set quality than the other three configurations, proving the synergistic effect of transformation matrix-guided initialization and directional mutation strategy.

[0122] 4. Typical Restructuring Schemes

[0123] 1) Scheme A (Minimum Time Delay Scheme): To quickly restore the task loop, decision nodes C1 and C6 are selected to jointly replace the failed C2, and strike nodes M3 and SH1 are selected to jointly replace the failed F1. This scheme has a low decision latency (ATD) and can quickly generate a usable solution, such as... Figure 7 As shown.

[0124] 2) Option B (Lowest Cost Option): To reduce reconfiguration resource consumption, only decision node C1 is selected to replace C2, and only attack node M3 is selected to replace F1. This option has a lower average reconfiguration cost (ARC), such as... Figure 8 As shown.

[0125] Commanders can make a final decision among multiple Pareto solutions, such as Option A and Option B, based on the current battlefield urgency and resource availability.

[0126] Experimental results show that the proposed method has significant advantages over the traditional global reconstruction method (G-NSGA-III) and the standard algorithm using only repair region constraints (P-NSGA-III) in terms of solution set quality (higher HV value, lower IGD value), convergence speed (faster HV iteration curve rise), and scheme efficiency (able to recover all failed functions with lower average decision delay and reconstruction cost). Ablation experiments further demonstrate the effectiveness and synergistic effect of transformation matrix-guided initialization and directional mutation strategies in improving algorithm performance.

[0127] In summary, this invention effectively achieves rapid and high-quality local reconstruction of the kill net through forced localization of the local repair area, multi-target integrated optimization, and TM-INSGA-III intelligent search, providing a feasible technical approach to improve the resilience and self-healing capability of the kill net in dynamic battlefield environments.

[0128] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for rapid local reconstruction of a kill network based on a network replenishment strategy, characterized in that: Includes the following steps: S1: Obtain the current status information of the kill net and detect whether there are any failed nodes or failed edges in the network; S2: When a failed node or failed edge is detected, a local repair zone centered on the failed node is constructed based on the network patching strategy, and the reconstruction decision variables are limited to the local repair zone. S3: Establish a multi-objective optimization model within the local repair area. The multi-objective optimization model aims to maximize the task recovery benefits, minimize the resource usage costs, and minimize the task completion delay. S4: The TM-NSGA-III algorithm guided by the transformation matrix is ​​used to solve the multi-objective optimization model and obtain the Pareto optimal solution set; S5: Evaluate the Pareto optimal solution set, determine the reconstruction plan to be executed based on the battlefield situation, perform network topology update, and complete the local reconstruction of the kill network.

2. The method for local fast reconstruction of a kill network based on a network replenishment strategy according to claim 1, characterized in that: The specific process of constructing a local repair area includes the following steps: S21. Determine the neighborhood range based on the hop count constraint, with the failed node as the center. S22. Within the neighborhood, candidate nodes with substitution capabilities are selected based on the functional relevance criterion. S23. Filter nodes that are physically reachable based on resource reachability criteria; The set of nodes and their connecting edges that satisfy the above conditions is defined as the local repair region.

3. The method for local fast reconstruction of a kill network based on a network replenishment strategy according to claim 1, characterized in that: The objective function of the multi-objective optimization model includes: (1) Maximize the benefits of task recovery: ； in, Indicates task priority weight. This indicates the task recovery status variable. Represents the set of failed tasks; (2) Minimize resource usage costs: ； in, Indicates the cost of enabling a node. This indicates the cost of establishing and maintaining the link. Represents the set of nodes within the local repair region. This represents the set of edges within the local repair region. This indicates the decision variable for node selection. Indicates the decision variables for link establishment; (3) Minimize task completion delay: ； in, Indicates link transmission delay. Indicates task In the link The allocation decision variables.

4. The method for rapid local reconstruction of a kill network based on a network replenishment strategy according to claim 1, characterized in that: The method for constructing the transformation matrix-guided TM-NSGA-III algorithm includes the following steps: S41. Construct a functional substitution transition matrix between failed nodes and candidate nodes to characterize the confidence level of candidate nodes in functional substitution of failed nodes. S42. Generate an initial population based on the functional substitution transformation matrix, and prioritize the node combination with high substitution confidence as the initial solution; S43. Introduce local repair zone constraints in genetic operations so that candidate solutions only contain nodes and edges within the local repair zone; S44. Perform directional mutation based on the functional substitution transformation matrix. In the mutation operation, prioritize replacing nodes with high substitution confidence in the functional substitution transformation matrix to guide the search direction.

5. The method for local fast reconstruction of a kill network based on a network replenishment strategy according to claim 4, characterized in that: The method for generating the initial population includes the following steps: S51. Extract path seeds for each interrupted task flow. Extract its position in the original network from historical data. Optimal historical path ; S52, Map the core path and traverse it. For each node, if node If it is invalid, then query the transformation matrix for the first... Okay, choose alternative confidence level The highest candidate node Replace the node; if the node is healthy, retain it; through this process, generate a core repair path within the repair area for each interrupted task flow and encode it as an initial solution individual; S53. Inject population diversity by performing controlled random perturbation operations on all obtained core repair path individuals with a low probability. Randomly replace a node or edge in the path to generate a batch of perturbed individuals, thereby enhancing the population's exploration capabilities; where: the replacement options are limited to feasible resources within the repair zone; S54. Assemble the initial population by merging all unmutated core repair path individuals with mutated perturbation individuals to form the initial population for the multi-objective optimization algorithm. .

6. The method for local fast reconstruction of a kill network based on a network replenishment strategy according to claim 1, characterized in that: The method for determining the refactoring scheme to be executed includes the following steps: S61. Calculate the comprehensive score of each candidate solution in the Pareto optimal solution set using the comprehensive scoring method. The scoring dimensions include task recovery rate, resource cost ratio and task completion delay compliance rate. S62. Adjust the weight of each scoring dimension dynamically based on the current battlefield situation and the emphasis on timeliness or economy. S63. Select the candidate solution with the best score as the final execution plan, and perform network topology update to complete the reconstruction.

Images