Method for quickly and dynamically reconstructing ultra-long-distance air-to-air killer chain

Abstract

The invention discloses a method for quickly and dynamically reconstructing an ultra-long-distance air-to-air killing chain, belongs to the technical field of ultra-long-distance air-to-air killing chains, and realizes dynamic recombination of the killing chain by using a self-organizing iterative clustering algorithm. In the execution process of the killing chain, if a reconstruction condition is triggered, all the combat platforms in normal operation are divided into six clusters according to the execution stages of discovery, positioning, tracking, aiming, combat and evaluation, each cluster is divided into different groups according to different loads loaded by the combat platforms, and the loading loads comprise ESM, light thunder and radar; for the combat platforms loaded with the same load, dividing the combat platforms in the same area into the same cluster by taking the current position as a division condition; all the combat platforms are subjected to dimension reduction clustering in sequence. According to the method, the high-dimensional complex calculation degree generated in the killer chain reconstruction process is subjected to dimension reduction processing, the calculation amount is remarkably reduced, and the reconstruction time is shortened.

Description

Technical Field

[0001] This invention relates to the field of ultra-long-range air-to-air kill chain technology, and specifically to a method for rapid dynamic reconfiguration of ultra-long-range air-to-air kill chains. Background Technology

[0002] Modern warfare has rapidly shifted from platform-based operations to system-of-systems (SoC) operations. The effectiveness of equipment SoC operations increasingly relies on the overall combat capability of the kill chain formed by the close coordination and interconnection of combat platforms. However, current research and analysis of kill chains have given relatively little consideration to the dynamic reconfiguration process. Dynamic reconfiguration is a typical characteristic reflecting the stability of a kill chain, enabling it to adapt to rapidly changing battlefield situations and holding significant research potential. A kill chain is a complete operational path from reconnaissance to the destruction of enemy targets. The more kill chains there are, the stronger the combat capability and resilience of the combat network system. In actual combat, if a kill chain fails, it will select other nodes with the same function as backups, thus dynamically reconfiguring to form new connections and new kill chains, improving the network's resilience and continuing to perform its combat functions. Furthermore, with the further expansion of the air combat battlefield, the types and numbers of combat platforms involved in constructing ultra-long-range kill chains are increasing. The traditional method of constructing kill chains using forward traversal is no longer sufficient to meet the time response requirements of rapidly changing battlefield situations.

[0003] This application addresses the problems existing in the prior art by establishing a method for rapid dynamic reconfiguration of ultra-long-range air-to-air kill chains, aiming to solve the problem of excessive kill chain closure time caused by too many combat platforms participating in kill chain reconfiguration. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, this invention provides an event-triggered method for rapid dynamic reconfiguration of ultra-long-range air-to-air kill chains, belonging to the field of ultra-long-range air-to-air kill chain technology. It utilizes a self-organizing iterative clustering algorithm to achieve dynamic kill chain reconfiguration. During kill chain execution, if a reconfiguration condition is triggered, all normally operating combat platforms are divided into six clusters according to the execution phases of detection, location, tracking, targeting, engagement, and assessment. Within each cluster, they are further divided into different groups based on the payloads carried by the combat platforms, including ESM, optical radar, and radar. For combat platforms carrying the same payload, their current location is used as the dividing criterion, and combat platforms within the same area are grouped into the same cluster. This process is repeated to reduce the dimensionality of all combat platforms during the dimensionality reduction clustering. This invention reduces the high-dimensional complexity of computation generated during kill chain reconfiguration, significantly reducing the computational load and shortening the reconfiguration time.

[0005] A method for rapid dynamic reconstruction of an ultra-long-range air-to-air kill chain includes the following steps:

[0006] Step 1: Construct the initial kill chain based on the combat platform;

[0007] Step 2: Cluster the combat platforms using a self-organizing iterative clustering algorithm;

[0008] Step 3: Select the optimal cluster to complete the reconstruction of the kill chain.

[0009] Furthermore, in step 1, the initial kill chain construction process is as follows:

[0010] Step 1.1, Initialize the combat platform and target:

[0011] Initialize the combat platform, clarify its position and attitude information, and indicate the sensor and weapon payloads carried by the combat platform.

[0012] Initialize targets, clarify their positions, attitude information, and threat level of each target, and form a situational model of the entire battlefield;

[0013] Step 1.2: Determine the targets and tasks of the initial kill chain based on the threat level;

[0014] Step 1.3: Select the combat platform with the highest performance value for the mission;

[0015] Step 1.4: Combine the combat platforms selected in Step 1.3 according to mission requirements according to the discovery-location-tracking-aiming-engagement-evaluation phases to form an initial kill chain; and determine the specific combat platforms that will perform the missions in each phase of the initial kill chain.

[0016] Furthermore, in step 2, the clustering algorithm performs the following clustering process on the combat platforms:

[0017] Step 2.1: Classify the combat platforms;

[0018] Step 2.2: Clustering is performed on combat platforms with the same payload.

[0019] Furthermore, in step 2.1, the combat platforms are classified according to different payloads, including optoelectronic sensor platforms, radar sensor platforms, ESM platforms, engagement platforms, and guidance platforms;

[0020] The optoelectronic sensor platform is a combat platform equipped with optoelectronic sensors, including reconnaissance drones and reconnaissance aircraft;

[0021] The radar sensor platform is a combat platform equipped with radar sensors, including early warning aircraft and fighter jets;

[0022] The ESM sensor platform is a combat platform equipped with ESM sensors, including electronic reconnaissance aircraft and electronic warfare aircraft;

[0023] The combat platform is a combat platform carrying different weapons and ammunition, including fighter jets and attack aircraft;

[0024] The guidance platform is a combat platform equipped with a guidance system.

[0025] Furthermore, in step 2.2, the clustering process for the combat platforms with the same payload is as follows:

[0026] Step 2.2.1: When the initial kill chain triggers the reconstruction condition during the operation, the self-organizing iterative clustering algorithm dynamically reconstructs the initial kill chain; the reconstruction condition includes the failure of the combat platform, the change of the target threat value, and the interruption of communication between combat platforms.

[0027] The specific steps of the dynamic reconstruction are as follows:

[0028] First, check the status of the combat platform:

[0029] The status of each combat platform is monitored in real time, including its health status, payload type, and remaining resources; the remaining resources include ammunition and fuel.

[0030] Then, determine the combat platform for the payload required for dynamic reconfiguration:

[0031] Determine the stage requiring dynamic reconfiguration, and reconfigure the combat platform according to the operational missions specified in the dynamic reconfiguration stage and the types of payloads required by each combat platform in each stage; combat platforms that need to carry radar, electro-optical, or ESM sensor payloads in the detection, location, tracking, and assessment stages, and combat platforms that need to carry guidance and weapon payloads in the aiming and engagement stages.

[0032] Step 2.2.2, Initialize the self-organizing iterative clustering algorithm:

[0033] Input: Feature dataset of the payload combat platform {x i , i = 1, 2, ..., 80}; x i For each combat platform, i represents the number of combat platforms;

[0034] Pre-selected parameters:

[0035] k: Initial number of cluster centers;

[0036] k max Maximum number of cluster centers;

[0037] k min Minimum number of cluster centers;

[0038] N min Minimum number of samples allowed per cluster center;

[0039] θs Splitting threshold of intra-class variance;

[0040] θ c The merging threshold for inter-class distance;

[0041] θ n The threshold for merging the number of samples within a class;

[0042] max_iter: The maximum number of iterations allowed.

[0043] Furthermore, the self-organizing iterative clustering algorithm employs an adaptive method to calculate parameters, as follows:

[0044] An adaptive method is used to calculate the inter-cluster merging threshold θ of cluster centers. c The splitting threshold θ of the intra-class variance s For each dimension i of the sample coordinates, find the maximum value max of the coordinate dimension in the feature dataset. i and minimum value min i ; Calculate the range of each coordinate dimension i Scope:

[0045] range i =max i -min i

[0046] Calculate the average value across all coordinate dimensions.

[0047]

[0048] Where n is the number of coordinate dimensions, and the value of n ranges from 1 to 100;

[0049] Inter-cluster merging threshold θ for cluster centers c The splitting threshold θ of the intra-class variance s Take the proportion of the average range k c and k s ,Right now

[0050]

[0051] In ultra-long-range air combat scenarios, the actual combat distance between aircraft ranges from several kilometers to tens of kilometers. Considering this actual combat distance, the inter-class clustering merging threshold θ for cluster centers is... c , the ratio k c Set to 15% to 20%; the splitting threshold θ for within-class variance. s , the ratio k s Setting it to 10% to 15% improves the adaptability and robustness of the clustering algorithm.

[0052] Step 2.2.3, Clustering:

[0053] Randomly select initial cluster centers: randomly select k initial cluster centers from the feature dataset as samples;

[0054] (1) Sample allocation: Assign each sample to the nearest cluster center; for each combat platform x i Computational combat platform x i With each cluster center c j Distance d(x) i ,c j ):

[0055]

[0056] Where m represents a random coordinate dimension and d represents the maximum coordinate dimension;

[0057] Combat platform x i Assigned to the nearest cluster center

[0058]

[0059] l represents the current number of cluster centers, c l Represents the set of cluster centers;

[0060] (2) Update cluster centers: Calculate the center of each cluster, which is the mean of all samples in the cluster; for each cluster center c j According to the cluster center c j Combat platform update location:

[0061]

[0062] Where, n j It is assigned to cluster center c j The number of combat platforms, C j It is assigned to cluster center c j A collection of combat platforms;

[0063] (3) Split clustering: If the standard deviation of the cluster exceeds θ s And the number of samples is greater than 2N min Then the cluster will be split into two new clusters; for each cluster center c j Calculate the within-class variance

[0064]

[0065] If the following conditions are met, then the cluster center c is split. j :

[0066]

[0067] During splitting, select the dimension with the largest within-class variance and set the cluster centers c. j It splits into two new cluster centers c j1 and c j2

[0068] c j1 =c j +δ·e m

[0069] c j2 =c j -δ·e m

[0070] Where δ is a constant, and the range of δ is 1×10⁻⁶. -3 Up to 1×10 -5 e m It is the unit vector in the dimension with the largest intra-class variance;

[0071] (4) Merge clusters: If the distance between two cluster centers is less than θ c And the sum of the sample sizes of the two samples is greater than θ. n If the two clusters are not equal, then the two clusters will be merged into one cluster; for cluster center c p and c j Calculate the cluster center c p and c j The distance d(c) between p c j ):

[0072]

[0073] If the following conditions are met, then the cluster centers c are merged. p and c j :

[0074] d(c p c j )<θ c and n p +n j >θ n

[0075] For each cluster center c p The number of samples to be checked is n i If the sample size n i Less than the minimum number of samples N allowed in each cluster center min ,Right now

[0076] n i <Nmin

[0077] Then find the distance c from the cluster center p Recent cluster center And cluster center c p and Merge clusters while removing the centers of already merged clusters;

[0078] (5) Iteration: Repeat the steps of allocating samples, updating clusters, splitting clusters and merging clusters until the cluster centers no longer change or the maximum number of iterations max_iter is reached.

[0079] Furthermore, in step 3, the process of reconstructing the kill chain is as follows:

[0080] Step 3.1: Within the same load category, calculate the efficiency value of the cluster center using the performance indicators of the current load, such as firepower, mobility, and positional advantage; select the cluster with the highest efficiency value as the representative of the load category; for the reconstruction of the photoelectric sensor platform during the positioning phase, select the cluster with the optimal efficiency value of the cluster center during the positioning phase from all clusters of the photoelectric sensor platform as the representative.

[0081] Step 3.2: Select a specific combat platform from the optimal cluster of each payload category; calculate the effectiveness value of the performance indicators of the combat platforms in the optimal cluster, including firepower, mobility, positional advantage, and survivability; select the combat platform with the highest effectiveness value as the representative platform in the optimal cluster; for the reconstruction of the optoelectronic sensor platform in the positioning phase, select the reconnaissance UAV with the best comprehensive effectiveness value in terms of mobility, detection capability, and survivability from the optimal cluster of optoelectronic sensor platforms as the representative platform to participate in the reconstruction of the kill chain;

[0082] Step 3.3: For the required reconstruction phase and subsequent phases, use steps 3.1 and 3.2 to select the corresponding combat platform and reconstruct a new kill chain of discovery-location-tracking-aiming-engagement-evaluation; after each reconstruction of the kill chain, the effectiveness evaluation model evaluates the effectiveness value of the newly generated kill chain.

[0083] Compared with the prior art, the present invention has the following beneficial technical effects:

[0084] This invention provides a method for rapid dynamic reconstruction of ultra-long-range air-to-air kill chains. This method can reduce the high-dimensional complexity of computations generated during the kill chain reconstruction process, significantly reducing the amount of computation, shortening the reconstruction time, and solving the problem of excessive kill chain closure time caused by too many combat platforms participating in the kill chain reconstruction. Attached Figure Description

[0085] Figure 1 This is a schematic diagram of the self-organizing iterative clustering algorithm. Detailed Implementation

[0086] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0087] like Figure 1 As shown, a method for rapid dynamic reconstruction of an ultra-long-range air-to-air kill chain includes the following steps:

[0088] Step 1: Construct the initial kill chain based on the combat platform;

[0089] Step 2: Cluster the combat platforms using a self-organizing iterative clustering algorithm;

[0090] Step 3: Select the optimal cluster to complete the reconstruction of the kill chain.

[0091] Furthermore, in step 1, the initial kill chain construction process is as follows:

[0092] Step 1.1, Initialize the combat platform and target:

[0093] Initialize the combat platform, clarify the ID, location and attitude information of our combat platform, and indicate the sensor payload and weapon payload carried by each combat platform;

[0094] Initialize targets, clarify their positions, attitude information, and threat level of each target, and form a situational model of the entire battlefield;

[0095] Step 1.2: Determine the targets and tasks for initializing the kill chain based on the threat level;

[0096] Step 1.3: Select the combat platform with the highest performance value for the mission;

[0097] Step 1.4: Combine the combat platforms selected in Step 1.3 according to mission requirements in the six stages of discovery, location, tracking, aiming, engagement, and assessment to form an initial kill chain; and determine the combat platform IDs that will perform the specific tasks in each stage of the initial kill chain.

[0098] Furthermore, in step 2, the clustering algorithm performs the following clustering process on the combat platforms:

[0099] During the execution of the initial kill chain, if a reconstruction condition is triggered, all combat platforms are divided into six clusters according to the execution phases of detection, location, tracking, targeting, engagement, and assessment. Within each cluster, they are further divided into different groups based on the payloads they carry, including ESM, optical radar, and radar. For combat platforms carrying the same payload, their current location is used as the dividing condition, and combat platforms in the same area are grouped into the same cluster. This dimensionality reduction and clustering of all combat platforms reduces the computational burden of dynamic kill chain reconstruction.

[0100] When the reconfiguration condition is triggered during the execution of the initial kill chain, the initial kill chain is dynamically reconfigured.

[0101] In the dynamic reconstruction process, when a reconstruction phase is required, the combat platforms in the reconstruction phase cluster are directly selected. Then, the payloads used to perform the reconstruction phase tasks are selected based on the battlefield situation. The payloads are selected from the groups, and a central point is abstracted from each cluster in the group. The central point represents the effectiveness of the cluster. The central point with the highest effectiveness is selected by comparison. The combat platforms required for this reconstruction phase are then selected from the cluster to which the central point belongs. The effectiveness of each member combat platform in the selected cluster is then compared, and the combat platform with the optimal effectiveness value is selected as the final reconstruction platform.

[0102] Step 2.1: Classify the combat platforms;

[0103] Step 2.2: Clustering is performed on combat platforms with the same payload.

[0104] Furthermore, in step 2.1, the combat platforms are classified according to different payloads, including optoelectronic sensor platforms, radar sensor platforms, ESM platforms, engagement platforms, and guidance platforms;

[0105] The optoelectronic sensor platform is a combat platform equipped with optoelectronic sensors, including reconnaissance drones and reconnaissance aircraft;

[0106] The radar sensor platform is a combat platform equipped with radar sensors, including early warning aircraft and fighter jets;

[0107] The ESM sensor platform is a combat platform equipped with ESM sensors, including electronic reconnaissance aircraft and electronic warfare aircraft;

[0108] The combat platform is a combat platform carrying different weapons and ammunition, including fighter jets and attack aircraft;

[0109] The guidance platform is a combat platform equipped with a guidance system.

[0110] Furthermore, in step 2.2, the clustering process for the combat platforms with the same payload is as follows:

[0111] Step 2.2.1: When the initial kill chain triggers the reconstruction condition during the operation, the self-organizing iterative clustering algorithm dynamically reconstructs the initial kill chain; the reconstruction condition includes the failure of the combat platform, the change of the target threat value, and the interruption of communication between combat platforms.

[0112] The specific steps of the dynamic reconstruction are as follows:

[0113] First, check the status of the combat platform:

[0114] The status of each combat platform is monitored in real time, including its health status, payload type, and remaining resources; the remaining resources include ammunition and fuel.

[0115] Then, determine the combat platform for the payload required for dynamic reconfiguration:

[0116] Determine the stage requiring dynamic reconfiguration, and reconfigure the combat platform according to the operational missions specified in the dynamic reconfiguration stage and the types of payloads required by each combat platform in each stage; combat platforms that need to carry radar, electro-optical, or ESM sensor payloads in the detection, location, tracking, and assessment stages, and combat platforms that need to carry guidance and weapon payloads in the aiming and engagement stages.

[0117] Step 2.2.2, Initialize the self-organizing iterative clustering algorithm:

[0118] Input: Feature dataset of the payload combat platform {x i , i = 1, 2, ..., 80}; x i For each combat platform, i represents the number of combat platforms;

[0119] Pre-selected parameters:

[0120] k: Initial number of cluster centers;

[0121] k max Maximum number of cluster centers;

[0122] k min Minimum number of cluster centers;

[0123] N min Minimum number of samples allowed per cluster center;

[0124] θ s Splitting threshold of intra-class variance;

[0125] θ c The merging threshold for inter-class distance;

[0126] θ n The threshold for merging the number of samples within a class;

[0127] max_iter: The maximum number of iterations allowed.

[0128] Furthermore, the adaptive parameter calculation method in the self-organizing iterative clustering algorithm is as follows:

[0129] An adaptive method is used to calculate the inter-cluster merging threshold θ of cluster centers. c The splitting threshold θ of the intra-class variance s For each dimension i of the sample coordinates, find the maximum value max of the coordinate dimension in the feature dataset. i and minimum value min i ; Calculate the range of each coordinate dimension i Scope:

[0130] range i =max i -min i

[0131] Calculate the average value across all coordinate dimensions.

[0132]

[0133] Where n is the number of coordinate dimensions, and the value of n ranges from 1 to 100;

[0134] Inter-cluster merging threshold θ for cluster centers c The splitting threshold θ of the intra-class variance s Take the proportion of the average range k c and k s ,Right now

[0135]

[0136] In ultra-long-range air combat scenarios, the actual combat distance between aircraft ranges from several kilometers to tens of kilometers. Considering this actual combat distance, the inter-class clustering merging threshold θ for cluster centers is... c , the ratio k c Set to 15% to 20%; for the splitting threshold θ of the intra-class variance. s , the ratio k s Setting it to 10% to 15% improves the adaptability and robustness of the clustering algorithm;

[0137] Step 2.2.4, Clustering:

[0138] Randomly select initial cluster centers: randomly select k samples from the feature dataset as initial cluster centers;

[0139] (1) Sample allocation: Assign each sample to the nearest cluster center; for each combat platform x i Computational combat platform xi With each cluster center c j Distance d(x) i ,c j ):

[0140]

[0141] Where m represents a random coordinate dimension and d represents the maximum coordinate dimension;

[0142] Combat platform x i Assigned to the nearest cluster center c j min

[0143]

[0144] l represents the current number of cluster centers, c l Represents the set of cluster centers;

[0145] (2) Update cluster centers: Calculate the center of each cluster, which is the mean of all samples in the cluster; for each cluster center c j According to the cluster center c j Combat platform update location:

[0146]

[0147] Where, n j It is assigned to cluster center c j The number of combat platforms, C j It is assigned to cluster center c j A collection of combat platforms;

[0148] (3) Split clustering: If the standard deviation of the cluster exceeds θ s And the number of samples is greater than 2N min Then the cluster will be split into two new clusters; for each cluster center c j Calculate the within-class variance

[0149]

[0150] If the following conditions are met, then the cluster center c is split. j :

[0151]

[0152] During splitting, select the dimension with the largest within-class variance and set the cluster centers c. j It splits into two new cluster centers c j1 and c j2 :

[0153] c j1 =c j +δ·e m

[0154] c j2 =c j -δ·e m

[0155] Where δ is a constant, and the range of δ is 1×10⁻⁶. -3 Up to 1×10 -5 e m It is the unit vector in the dimension with the largest intra-class variance;

[0156] (4) Merge clusters: If the distance between two cluster centers is less than θ c And the sum of the sample sizes of the two samples is greater than θ. n Then, these two clusters will be merged into one cluster; for cluster center c p and c j Calculate the cluster center c p and c j The distance d(c) between p c j ):

[0157]

[0158] If the following conditions are met, then the cluster centers c are merged. p and c j :

[0159] d(c p c j )<θ c and n p +n j >θ n

[0160] For each cluster center c p The number of samples to be checked is n i If the sample size n i Less than the minimum number of samples N allowed in each cluster center min ,Right now

[0161] n i <N min

[0162] Then find the distance c from the cluster center p Recent cluster center And cluster center c p and Merge clusters while removing the centers of already merged clusters;

[0163] (5) Iteration: Repeat the steps of assigning, updating, splitting and merging until the cluster centers no longer change or the maximum number of iterations max_iter is reached.

[0164] Furthermore, in step 3, the process of reconstructing the kill chain is as follows:

[0165] Step 3.1: Within the same load category, calculate the efficiency value of the cluster center using the performance indicators of the current load, such as firepower, mobility, and positional advantage; select the cluster with the highest efficiency value as the representative of the load category; for the reconstruction of the photoelectric sensor platform during the positioning phase, select the cluster with the optimal efficiency value of the cluster center during the positioning phase from all clusters of the photoelectric sensor platform as the representative.

[0166] Step 3.2: Select a specific combat platform from the optimal cluster of each payload category; calculate the effectiveness value of the performance indicators of the combat platforms in the optimal cluster, including firepower, mobility, positional advantage, and survivability; select the combat platform with the highest effectiveness value as the representative platform in the optimal cluster; for the reconstruction of the optoelectronic sensor platform in the positioning phase, select the reconnaissance UAV with the best comprehensive effectiveness value in terms of mobility, detection capability, and survivability from the optimal cluster of optoelectronic sensor platforms as the representative platform to participate in the reconstruction of the kill chain;

[0167] Step 3.3: For the required reconstruction phase and subsequent phases, use steps 3.1 and 3.2 to select the corresponding combat platform and reconstruct a new kill chain of discovery-location-tracking-aiming-engagement-evaluation, i.e., the new kill chain; after each reconstruction of the kill chain, evaluate the effectiveness value of the new kill chain through the effectiveness evaluation model; ensure that the new kill chain meets the operational requirements, and the effectiveness value of the new kill chain cannot be lower than the minimum effectiveness value requirement on the battlefield.

[0168] Finally, it should be noted that, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples. The above descriptions are merely embodiments of the embodiments in this specification and are not intended to limit the embodiments of this specification. For those skilled in the art, the embodiments in this specification can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the embodiments in this specification should be included within the scope of the claims of the embodiments in this specification.

Claims

1. A method for rapid dynamic reconstructing of an ultra-long-range air-to-air kill chain, characterized in that, The method for rapid dynamic reconstruction of the kill chain includes the following steps: Step 1: Construct the initial kill chain based on the combat platform; Step 2: Cluster the combat platforms using a self-organizing iterative clustering algorithm; Step 3: Select the optimal cluster to complete the reconstruction of the kill chain.

2. The method for rapid dynamic reconfiguration of an ultra-long-range air-to-air kill chain according to claim 1, characterized in that, In step 1, the initial kill chain construction process is as follows: Step 1.1, Initialize the combat platform and target: Initialize the combat platform, clarify its position and attitude information, and indicate the sensor and weapon payloads carried by the combat platform. Initialize targets, clarify their positions, attitude information, and threat level of each target, and form a situational model of the entire battlefield; Step 1.2: Determine the targets and tasks of the initial kill chain based on the threat level; Step 1.3: Select the combat platform with the highest performance value for the mission; Step 1.4: Combine the combat platforms selected in Step 1.3 according to mission requirements according to the discovery-location-tracking-aiming-engagement-evaluation phases to form an initial kill chain; and determine the specific combat platforms that will perform the missions in each phase of the initial kill chain.

3. The method for rapid dynamic reconfiguration of an ultra-long-range air-to-air kill chain according to claim 1, characterized in that, In step 2, the clustering algorithm performs the following clustering process on the combat platforms: Step 2.1: Classify the combat platforms; Step 2.2: Clustering is performed on combat platforms with the same payload.

4. The method for rapid dynamic reconstructing of an ultra-long-range air-to-air kill chain according to claim 3, characterized in that, In step 2.1, the combat platforms are classified according to different payloads, including optoelectronic sensor platforms, radar sensor platforms, ESM platforms, engagement platforms, and guidance platforms; The optoelectronic sensor platform is a combat platform equipped with optoelectronic sensors, including reconnaissance drones and reconnaissance aircraft; The radar sensor platform is a combat platform equipped with radar sensors, including early warning aircraft and fighter jets; The ESM sensor platform is a combat platform equipped with ESM sensors, including electronic reconnaissance aircraft and electronic warfare aircraft; The combat platform is a combat platform carrying different weapons and ammunition, including fighter jets and attack aircraft; The guidance platform is a combat platform equipped with a guidance system.

5. The method for rapid dynamic reconstructing of an ultra-long-range air-to-air kill chain according to claim 3, characterized in that, In step 2.2, the clustering process for the combat platforms with the same payload is as follows: Step 2.2.1: When the initial kill chain triggers the reconstruction condition during the operation, the self-organizing iterative clustering algorithm dynamically reconstructs the initial kill chain; the reconstruction condition includes the failure of the combat platform, the change of the target threat value, and the interruption of communication between combat platforms. The specific steps of the dynamic reconstruction are as follows: First, check the status of the combat platform: The status of each combat platform is monitored in real time, including its health status, payload type, and remaining resources; the remaining resources include ammunition and fuel. Then, determine the combat platform for the payload required for dynamic reconfiguration: Determine the stage requiring dynamic reconfiguration, and reconfigure the combat platform according to the operational missions specified in the dynamic reconfiguration stage and the types of payloads required by each combat platform in each stage; combat platforms that need to carry radar, electro-optical, or ESM sensor payloads in the detection, location, tracking, and assessment stages, and combat platforms that need to carry guidance and weapon payloads in the aiming and engagement stages. Step 2.2.2, Initialize the self-organizing iterative clustering algorithm: Input: Feature dataset of the payload combat platform {x i , i = 1, 2, ..., 80}; x i For each combat platform, i represents the number of combat platforms; Pre-selected parameters: k: Initial number of cluster centers; k max Maximum number of cluster centers; k min Minimum number of cluster centers; N min Minimum number of samples allowed per cluster center; θ s Splitting threshold of intra-class variance; θ c The merging threshold for inter-class distance; θ n The threshold for merging the number of samples within a class; max_iter: The maximum number of iterations allowed. Step 2.2.3, Clustering: Randomly select initial cluster centers: randomly select k initial cluster centers from the feature dataset as samples; (1) Sample allocation: Assign each sample to the nearest cluster center; for each combat platform x i Computational combat platform x i With each cluster center c j Distance d(x) i ,c j ): Where m represents a random coordinate dimension and d represents the maximum coordinate dimension; Combat platform x i Assigned to the nearest cluster center l represents the current number of cluster centers, c l Represents the set of cluster centers; (2) Update cluster centers: Calculate the center of each cluster, which is the mean of all samples in the cluster; for each cluster center c j According to the cluster center c j Combat platform update location: Where, n j It is assigned to cluster center c j The number of combat platforms, C j It is assigned to cluster center c j A collection of combat platforms; (3) Split clustering: If the standard deviation of the cluster exceeds θ s And the number of samples is greater than 2N min Then the cluster will be split into two new clusters; for each cluster center c j Calculate the within-class variance If the following conditions are met, then the cluster center c is split. j : During splitting, select the dimension with the largest within-class variance and set the cluster centers c. j It splits into two new cluster centers c j1 and c j2 : c j1 =c j +d·e m c j2 =c j -d·e m Where δ is a constant, and the range of δ is 1×10⁻⁶. -3 Up to 1×10 -5 e m It is the unit vector in the dimension with the largest intra-class variance; (4) Merge clusters: If the distance between two cluster centers is less than θ c And the sum of the sample sizes of the two samples is greater than θ. n If the two clusters are not equal, then the two clusters will be merged into one cluster; for cluster center c p and c j Calculate the cluster center c p and c j The distance d(c) between p c j ): If the following conditions are met, then the cluster centers c are merged. p and c j : d(c p ,c j )<θ c and n p +n j >θ n For each cluster center c p The number of samples to be checked is n i If the sample size n i Less than the minimum number of samples N allowed in each cluster center min ,Right now n i <N min Then find the distance c from the cluster center p Recent cluster center And cluster center c p and Merge clusters while removing the centers of already merged clusters; (5) Iteration: Repeat the steps of allocating samples, updating clusters, splitting clusters and merging clusters until the cluster centers no longer change or the maximum number of iterations max_iter is reached.

6. The method for rapid dynamic reconfiguration of an ultra-long-range air-to-air kill chain according to claim 5, characterized in that, The adaptive parameter calculation method in the self-organizing iterative clustering algorithm is as follows: An adaptive method is used to calculate the inter-cluster merging threshold θ of cluster centers. c The splitting threshold θ of the intra-class variance s For each dimension i of the sample coordinates, find the maximum value max of the coordinate dimension in the feature dataset. i and minimum value min i ; Calculate the range of each coordinate dimension i Scope: tidy i <max i -min i Calculate the average value across all coordinate dimensions. Where n is the number of coordinate dimensions, and the value of n ranges from 1 to 100; Inter-cluster merging threshold θ for cluster centers c The splitting threshold θ of the intra-class variance s Take the proportion of the average range k c and k s ,Right now In ultra-long-range air combat scenarios, the actual combat distance between aircraft ranges from several kilometers to tens of kilometers. Considering this actual combat distance, the inter-class clustering merging threshold θ for cluster centers is... c , the ratio k c Set to 15% to 20%; the splitting threshold θ for within-class variance. s , the ratio k s Setting it to 10% to 15% improves the adaptability and robustness of the clustering algorithm.

7. The method for rapid dynamic reconfiguration of an ultra-long-range air-to-air kill chain according to claim 1, characterized in that, In step 3, the process of reconstructing the kill chain is as follows: Step 3.1: Within the same load category, calculate the efficiency value of the cluster center using the performance indicators of the current load, such as firepower, mobility, and positional advantage; select the cluster with the highest efficiency value as the representative of the load category; for the reconstruction of the photoelectric sensor platform during the positioning phase, select the cluster with the optimal efficiency value of the cluster center during the positioning phase from all clusters of the photoelectric sensor platform as the representative. Step 3.2: Select a specific combat platform from the optimal cluster of each payload category; calculate the effectiveness value of the performance indicators of the combat platforms in the optimal cluster, including firepower, mobility, positional advantage and survivability; select the combat platform with the highest effectiveness value as the representative platform in the optimal cluster. For the reconstruction of the optoelectronic sensor platform in the positioning phase, the reconnaissance UAV with the best comprehensive performance value such as mobility, detection capability and survivability is selected from the optimal cluster of optoelectronic sensor platforms as the representative platform to participate in the reconstruction of the kill chain. Step 3.3: For the required reconstruction phase and subsequent phases, use steps 3.1 and 3.2 to select the corresponding combat platform and reconstruct a new kill chain of discovery-location-tracking-aiming-engagement-evaluation; after each reconstruction of the kill chain, the effectiveness evaluation model evaluates the effectiveness value of the newly generated kill chain.

Images