A decentralized clustering method for cluster-oriented multi-objective allocation

Abstract

The present application relates to the field of unmanned cluster, specifically relates to a kind of cluster multi-objective allocation-oriented decentralization clustering method, comprising the following steps: determining multi-operation target allocation model, the multi-operation target allocation model includes: multiple platforms, multiple target nodes, each platform includes a platform starting point node;Each of the platform includes first constraint attribute;Each of the target node includes second constraint attribute;Determine the target constraint condition of multi-operation target allocation model;Multi-objective allocation is executed, including: obtaining the benefit function of multiple platforms to multiple target nodes;Perform cluster autonomous clustering calculation;The cluster in the result of the cluster autonomous clustering calculation is sequentially arranged;The present application can guarantee certain cluster allocation benefit under the condition of considering resource constraint, while greatly reducing the time complexity of clustering algorithm.

Description

Technical Field

[0001] This invention relates to the field of unmanned swarm technology, and more specifically to a decentralized clustering method for multi-objective allocation in swarms. Background Technology

[0002] This invention is applied to the field of military operations, specifically the technical field of target allocation for multiple platforms (such as aircraft, ground vehicles, ships, etc.) in actual combat missions.

[0003] In modern military operations, multi-platform collaborative operations have become essential for improving operational efficiency and flexibility. However, in a multi-platform, multi-target battlefield environment, target allocation remains a complex and challenging problem. In actual combat missions, enemy targets often appear with varying values. Considering the constraints of our platform's range and ammunition capacity, how to allocate limited resources to multiple targets to maximize the benefits of cluster operations becomes a crucial issue.

[0004] Clustering algorithms are one of the most efficient techniques for handling multi-target allocation problems in combat scenarios. By dividing specific features and clustering similar targets, clustering algorithms effectively improve resource allocation efficiency, enabling each platform to perform tasks more effectively. Among existing clustering algorithms, the K-means clustering algorithm is the most relevant to this invention. The K-means algorithm is simple and easy to understand, computationally fast, and suitable for large-scale datasets; it exhibits good scalability when processing large-scale data; and its clustering results are intuitive, with the center of each cluster representing the cluster, facilitating interpretation and understanding. However, the K-means clustering algorithm requires initializing the number of target allocation clusters and is sensitive to initial target attribute values. It struggles to effectively cluster outliers and cannot establish a sequential order for cluster members. In the context of current multi-platform collaborative operations, the K-means clustering algorithm struggles to address various issues such as the significant differences in target value, rapidly changing battlefield situations and easily altered target information, constraints on friendly platform resources, and the need for sequential target allocation.

[0005] Exhaustive methods, exemplified by the Hungarian algorithm, are also an effective means of solving multi-target allocation problems in combat contexts. The Hungarian algorithm considers both gains and losses, making it suitable for maximizing the benefits of target allocation under constraints of specific combat resources. However, from an algorithmic perspective, the Hungarian algorithm has a very high time complexity (O(n...). 3 It is difficult to deal with the target allocation problem in the context of large-scale combat. Summary of the Invention

[0006] In view of the above problems, this invention provides a decentralized clustering method for multi-target allocation in clusters, which solves the problems of existing clustering algorithms being sensitive to outliers, having difficulty considering resource constraints, being unable to consider the sequential requirements of actual combat target allocation, and having high time costs for exhaustive algorithms in target allocation problems in large-scale combat scenarios.

[0007] In a first aspect, the present invention provides a decentralized clustering method for multi-objective allocation in clusters, comprising:

[0008] Step S1: Determine a multi-combat target allocation model, which includes: multiple platforms for combat and multiple target nodes for attack during combat. Each platform includes a platform origin node. Each platform includes a first constraint attribute. Each target node includes a second constraint attribute. The first constraint attribute includes: maximum payload and maximum range. The second constraint attribute includes: target value, lower bound of required ammunition quantity, and upper bound of required ammunition quantity.

[0009] Determine the target constraints for the multi-operational target allocation model;

[0010] Step S2, performing multi-objective allocation, includes: obtaining the benefit functions of multiple platforms for multiple target nodes; performing cluster autonomous clustering calculation; and sequentially arranging the clusters in the results of the cluster autonomous clustering calculation.

[0011] Preferably, the target constraints include: the total amount of ammunition allocated to each target node is within the upper and lower bounds of the required ammunition amount for each target node; the sum of the ammunition allocated to each target node by each platform does not exceed the maximum payload of each platform; the path length of each platform through the target nodes does not exceed the maximum range of each platform; the path of each platform through the target nodes is continuous; each platform only allocates ammunition to the target nodes it passes through; each platform is located at the platform starting node of each platform; there cannot be internal loops in the sequence of target nodes allocated to each platform; multi-target allocation is performed based on the value of the multiple target nodes and the amount of ammunition allocated to the target nodes by the multiple platforms.

[0012] Preferably, in step S2, obtaining the benefit function of multiple platforms for multiple target nodes includes: obtaining the benefit function of multiple platforms for multiple target nodes based on the value weight and distance weight of the platform relative to the target node, the attack value and distance of the platform relative to the target node, and the number of platforms that have been assigned to the target node.

[0013] Preferably, the benefit function for multiple platforms to multiple target nodes is obtained based on the platform's value weight and distance weight relative to the target node, the platform's attack value and distance relative to the target node, and the number of platforms already assigned to the target node. Specifically, this includes:

[0014] The benefit function Ω of platform i with respect to objective k ik The calculation method is as follows:

[0015]

[0016] in These are the value weight and distance weight of platform i relative to target k, respectively, r ik d ik V represents the attack value and distance of platform i relative to target k, respectively. k |V represents the set of platforms that have been assigned to target k. k ∪{i}| represents the number of platform members in the union of the set of platforms already assigned to target k and the currently assigned platform i.

[0017] Preferably, the method for obtaining the attack value of the platform relative to the target node is as follows:

[0018] When the remaining maximum amount of ammunition that the target node can withstand is greater than the remaining amount of ammunition on the platform, the attack value is the target value of the target node multiplied by the remaining amount of ammunition on the platform.

[0019] When the remaining maximum amount of ammunition that the target node can withstand does not exceed the remaining amount of ammunition on the platform, the attack value is the target value of the target node multiplied by the remaining maximum amount of ammunition that the target node can withstand.

[0020] The attack value is zero when the lower bound of the target node's required ammunition quantity is greater than the platform's remaining ammunition quantity.

[0021] Preferably, in step S2, the execution of cluster autonomous clustering calculation includes:

[0022] Obtain the initial state parameters of the target node and platform;

[0023] Traverse each target node. During the traversal, traverse each platform. If the relationship between the platform and the target node satisfies the target constraint conditions of the multi-objective allocation model in step S1, then the target node with the largest benefit function of the multiple platforms to the multiple target nodes is allocated to the corresponding platform.

[0024] Update the initial state parameters of the target node and the platform; the initial state parameters of the target node include the remaining maximum amount of ammunition to be hit; the initial state parameters of the platform include the remaining amount of ammunition and the range.

[0025] Preferably, in step S2, sequentially arranging the clusters in the result of the cluster autonomous clustering calculation includes: sequentially arranging the target nodes in the clusters in the calculation result using an exhaustive method.

[0026] Compared with the prior art, the present invention has at least the following beneficial effects:

[0027] This invention addresses the problem of multi-platform target allocation in combat scenarios by employing a target allocation method for multi-target sequential attacks, primarily based on a decentralized clustering algorithm. The first clustering result is used for a second sequential target sorting, resolving the issue of traditional clustering algorithms failing to form effective sequences among cluster members. Furthermore, by considering the number of cluster members and battlefield resource constraints when setting the benefit function, this invention addresses, to some extent, the inability of traditional clustering algorithms to handle anomalous value members. The method of this invention significantly reduces the time complexity of traditional exhaustive search algorithms while maintaining a certain level of global benefit. Attached Figure Description

[0028] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0029] Figure 1 This is a flowchart of the decentralized clustering method for multi-objective allocation in a cluster disclosed in this invention;

[0030] Figure 2 This is a schematic diagram illustrating the principle of platform cluster collaboration in performing attack tasks against multiple targets as disclosed in this invention.

[0031] Figure 3 This is a schematic diagram of the logical flow of the autonomous clustering process disclosed in this invention;

[0032] Figure 4 This is a schematic diagram of the multi-platform target attack path disclosed in this invention. Detailed Implementation

[0033] To better understand the above-described objectives, features, and advantages of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other. Furthermore, the present invention can be implemented in other ways different from those described herein; therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0034] This invention considers how a platform can allocate ammunition resources to multiple targets during collaborative attack missions to maximize attack effectiveness. It also needs to consider constraints such as target value, required ammunition quantity limits, and geographical location, and impose limitations on payload and range based on the platform's characteristics. A schematic diagram illustrating the principle of platform cluster collaboration in executing attack missions against multiple targets is shown below. Figure 2 As shown.

[0035] To illustrate the effectiveness of the method proposed in this invention, the following detailed description of the technical solution is provided through a specific embodiment. Please refer to... Figure 1 This invention provides a decentralized clustering method for multi-objective allocation in clusters, comprising the following steps:

[0036] Step S1: Determine a multi-combat target allocation model, which includes: multiple platforms for combat, multiple target nodes for attack during combat, each platform including a platform starting node; each platform including a first constraint attribute; and each target node including a second constraint attribute.

[0037] Determine the target constraints for the multi-combat target allocation model.

[0038] This invention assumes that the set of platforms participating in the combat mission is U = {1,2,...,m}, and the set of nodes is V = {0,1,...,n}, where node 0 represents the platform's starting point, from which the platform takes off and returns to the node after completing the mission, and T = {1,2,...,n} is the target set.

[0039] For platform i∈U, define the association variable: Q i Maximum payload; D i For the maximum range, define the correlation variable w for target l∈T. l For the goal of value; L l U is the lower bound of the amount of ammunition required to destroy target l; l Let w be the upper bound of the amount of ammunition dropped on target l; for target k∈T, define the correlation variable: w k Value for target k; L k U is the lower bound of the amount of ammunition required to destroy target k; k This is the upper bound of the amount of ammunition deployed against target k.

[0040] Define decision variables:

[0041]

[0042]

[0043]

[0044] z ki Allocate the amount of ammunition to target k for platform i.

[0045] In some embodiments, the first constraint attribute includes: maximum payload and maximum range; the second constraint attribute includes: target value, lower bound of required ammunition quantity, and upper bound of required ammunition quantity.

[0046] Cluster collaborative tasks and resource allocation can be modeled as follows:

[0047]

[0048] The above formula represents the sum of the value of multiple target nodes and the amount of ammunition allocated to target nodes by multiple platforms.

[0049] And satisfy the following constraints:

[0050]

[0051] The above formula indicates that the total amount of ammunition allocated to target k must be within the upper and lower bounds of its required ammunition amount, where T a A set of objectives for allocating corresponding resources;

[0052]

[0053] The above formula indicates that the sum of the amount of ammunition allocated by platform i to each target does not exceed its ammunition load limit;

[0054]

[0055] The above formula indicates that the path length traversed by platform i does not exceed its flight distance, where d kl Let l be the distance between targets k and l;

[0056]

[0057] The above formula indicates that the path should be continuous;

[0058]

[0059] The above formula indicates that ammunition can only be allocated to targets that have passed through it, where M is a sufficiently large constant;

[0060]

[0061] The above formula indicates that all platforms are at node 0;

[0062] a ki -a li +nx kli ≤n-1k≠l

[0063] The above formula indicates that the target sequence assigned to platform i must not contain internal loops.

[0064] x kli ∈{0,1}y ki ∈{0,1}

[0065] z ki >0aki ≥0

[0066] The above formula represents the decision variable x kli y ki a ki and z ki The range of values ​​for .

[0067] In some embodiments, the target constraints include: the total amount of ammunition allocated to each target node is within the upper and lower bounds of the required ammunition amount for each target node; the sum of the ammunition allocated to each target node by each platform does not exceed the maximum payload of each platform; the path length of each platform through the target nodes does not exceed the maximum range of each platform; the path of each platform through the target nodes is continuous; each platform only allocates ammunition to the target nodes it passes through; each platform is located at the platform starting node of each platform; there cannot be internal loops in the sequence of target nodes allocated to each platform; multi-target allocation is performed based on the value of the multiple target nodes and the amount of ammunition allocated to the target nodes by the multiple platforms.

[0068] Through the above steps, this invention establishes an allocation pattern for distributing ammunition resources across multiple targets to maximize attack effectiveness. Simultaneously, constraints such as target value, required ammunition quantity limits, and geographical location must be considered, and limitations on payload and range must be imposed based on platform characteristics for subsequent multi-target allocation.

[0069] Step S2, perform multi-objective assignment, including:

[0070] Obtain the benefit functions of multiple platforms for multiple target nodes;

[0071] Perform cluster autonomous clustering calculations;

[0072] The clusters in the results of the autonomous clustering calculation of the cluster are sequentially arranged.

[0073] The implementation process of the multi-platform sequential attack target allocation method dominated by decentralized clustering algorithm mainly includes two aspects: setting the benefit function of the target corresponding to the platform and constructing the logical framework for the entire clustering process.

[0074] Regarding the setting of the target-platform benefit function, in this invention, the direct basis for how a target is selected from multiple platforms for allocation is defined as the benefit function of the target-platform. The starting point for the benefit function design of the target allocation method dominated by decentralized clustering algorithm mainly considers two factors: the value obtained by the platform attacking the target and the flight distance required by the platform to attack the target. At the same time, it avoids the situation where the same target is repeatedly attacked by multiple platforms as much as possible, that is, it ensures that multiple targets are evenly allocated to multiple platforms as much as possible.

[0075] The benefit function Ω of platform i with respect to objective k ik The definition is as follows:

[0076]

[0077] in Let r be the weights of the value and distance of platform i relative to target k, respectively, both of which are constants. ik d ik V represents the attack value and distance of platform i relative to target k, respectively. k Represents the number of platform clusters currently assigned to target k; {i} represents platform i, |V k ∪{i}| represents the number of platform members in the union of the platform clusters already assigned to target k and the currently assigned platform i.

[0078] For example, if target k has already been assigned to two platforms, when determining whether platform i should be assigned to target k, |V k The value of ∪{i}| at this point is 3. There are two reasons for taking the union. First, during the initial allocation, when the set is empty, if 1 is not added, the benefit function will tend to infinity. Second, the benefit function is set to determine whether target k can be successfully allocated to platform i. For platform i, the allocation set of target 1 already contains 2 platforms. If platform i is added again, the number of target sets should be 3. Therefore, when calculating the benefit function considering the number of platforms in the target allocation cluster, it should be divided by 3, which is the number after the new platform is added.

[0079] As can be seen, the attack value of platform i relative to target k decreases as the number of targets already attacked in the platform cluster increases. This platform setup makes it less likely to target a target cluster with a large number of platforms, thus avoiding the problem of attacking too many targets on the same platform.

[0080] For the benefit function of the platform corresponding to the target, r ik In terms of settings, there are three forms, as follows:

[0081]

[0082] This invention will r ikDefined as the highest possible benefit that platform i can obtain when it can successfully attack target k, meaning that if platform i chooses to attack target k, it will definitely obtain the highest value in this attack without saving ammunition. The initial intention of this design is: 1) To reduce the time complexity of the algorithm. Learning from the shortcomings of the traditional Hungarian exhaustive search algorithm, simplifying the value obtained by platform i when it can successfully attack target k to obtaining the highest benefit can greatly reduce the algorithm's time complexity. However, this will cause the final allocation result to fall into a local maximum rather than a global maximum benefit. The relationship between the local maximum and global maximum benefit of the final algorithm will be theoretically derived in the next section. 2) To save on the consumption of constraint resources. In this invention, under the operational context, among the platform's ammunition load limit, fuel range limit, and the target's maximum and minimum amount of ammunition to be hit, the benefit function Ω... ik As the definition suggests, fuel range limitations do not directly affect the effectiveness of the platform and the cluster. However, considering actual combat requirements, designing this attack mode for the platform can effectively save fuel consumption of the platform cluster.

[0083] By setting the benefit function of the target to correspond to the platform in this way, the platform will tend to select targets with high value, few cluster members, and close proximity, thereby achieving the ideal allocation result.

[0084] In some embodiments, obtaining the benefit function of multiple platforms to multiple target nodes includes: obtaining the benefit function of multiple platforms to multiple target nodes based on the value weight and distance weight of the platform relative to the target node, the attack value and distance of the platform relative to the target node, and the number of platforms that have been assigned to the target node.

[0085] In some embodiments, the attack value of the platform relative to the target node is obtained in the following way:

[0086] When the remaining maximum amount of ammunition that the target node can withstand is greater than the remaining amount of ammunition on the platform, the attack value is the target value of the target node multiplied by the remaining amount of ammunition on the platform.

[0087] When the remaining maximum amount of ammunition that the target node can withstand does not exceed the remaining amount of ammunition on the platform, the attack value is the target value of the target node multiplied by the remaining maximum amount of ammunition that the target node can withstand.

[0088] The attack value is zero when the lower bound of the target node's required ammunition quantity is greater than the platform's remaining ammunition quantity.

[0089] Regarding the logical framework for the clustering process, this invention mainly divides the construction of the clustering process into two parts: autonomous clustering and sequential arrangement of targets within the clusters. In the autonomous clustering part, this invention performs preliminary clustering on the target allocation problem of multi-platform attacks against multiple targets in a combat context, based on the aforementioned benefit function. After the clustering process is completed, the clusters obtained are sequentially arranged using an exhaustive method to obtain the shortest path sequence for target attacks within each cluster.

[0090] The target allocation method dominated by the decentralized clustering algorithm in this invention includes a first step of cluster autonomous clustering and a second step of sequential arrangement of targets in the cluster.

[0091] Cluster autonomous clustering is an unordered allocation of multiple objectives based on a defined benefit function in this invention, resulting in a cluster of targets attacked by each platform. For example... Figure 3 As shown in the figure, the logical flow of the autonomous clustering process is illustrated in the table below.

[0092]

[0093] In some embodiments, performing cluster autonomous clustering computation includes:

[0094] Obtain the initial state parameters of the target node and platform;

[0095] For each target node, all platforms are traversed. If the relationship between the platform and the target node satisfies the target constraint conditions of the multi-combat target allocation model in step S1, the target node with the maximum benefit function is allocated to the corresponding platform, and the initial state parameters of the target node and the platform are updated.

[0096] The initial state parameters of the target node include the remaining maximum amount of ammunition to be hit; the initial state parameters of the platform include the remaining amount of ammunition and range.

[0097] In some embodiments, the first target node and platform selected when traversing each target node and each platform are randomly selected.

[0098] As can be seen from the pseudocode flow above, the cluster autonomous clustering algorithm in this invention does not traverse all the sequences to which targets are assigned. Instead, it performs clustering iterations according to the number order, taking the target sequence order during initialization as an example. This approach significantly reduces the time complexity of the algorithm, but it also leads to a loss in the global benefits of the cluster. Therefore, it is necessary to study the relationship between the optimal target allocation set obtained in this algorithm and the global benefits obtained from the theoretically optimal target allocation set.

[0099] For an allocation set ∏, the global benefit J of the system is defined as the benefit Ω of all individual members i in the allocation set ∏.iII(i) The sum is:

[0100]

[0101] Among them Π (i) L represents the target cluster index value finally selected by platform i according to the above algorithm, and L represents the platform cluster.

[0102] In the cluster autonomous clustering algorithm of this invention, the criterion for selecting the target cluster is set based on the benefit function of the platform itself and the target cluster. This means that the platform selection only considers its own preferences and does not take into account other platforms and the overall benefit of the entire platform cluster. Consequently, the result of the entire cluster allocation may fall into a local optimum rather than a globally optimal solution. Theoretically, it can be proven that the system global benefit J0 under the suboptimal solution obtained by this algorithm is different from the system global benefit J of the allocation set in the globally optimal solution. * The ratio ρ between them has a lower bound, that is:

[0103]

[0104] For any Nash stable allocation set, its suboptimal solution has a lower bound. ρ satisfy:

[0105]

[0106] in:

[0107]

[0108] Where q is an arbitrary natural number.

[0109] For a Nash stable allocation set Π, by definition we know that:

[0110]

[0111] Where L represents the platform cluster.

[0112] Where k * Represents the allocation set Π in the global optimal solution cluster. * The target cluster index value of platform i, please note the following. This represents allocating platform i to target cluster k under the current suboptimal solution allocation strategy. * The benefits under the current situation, and represent the global optimal solution strategy (i.e., allocation set Π). * (China) Platform i targets cluster k * The benefits are expressed as Furthermore, it is possible to Represented as:

[0113]

[0114] Therefore, we get:

[0115]

[0116] The term on the left side of the inequality above represents the global system benefit J0 under the current suboptimal allocation set Π, and the first term on the right side represents the global optimal solution allocation set Π. * The overall benefits of the lower system J * The second term on the right-hand side of the inequality has an upper bound:

[0117]

[0118] in Indicates that for target k * In terms of its global optimal allocation set Π * The set of all platforms, where T represents the target cluster. Although The exact value of is unknown, but its upper bound can be easily derived:

[0119] The basic definition of a utility function is known to be:

[0120]

[0121] For all q∈{1, 2, ..., |L|}, i∈L, we know that an upper bound can be expressed as:

[0122]

[0123] What you will get afterwards is:

[0124]

[0125] Equivalent to:

[0126]

[0127] This theoretically proves that the suboptimal local optimal solution achievable by the autonomous clustering program incurs a certain loss compared to the global optimal solution, and that this loss has an upper bound related to the real-time allocation set. For multi-target clustering and allocation problems, finding the optimal solution within a reasonable time may not always be feasible, especially in target allocation problems involving multi-platform attacks and multiple targets in a combat context, where the battlefield is large-scale and the situation changes rapidly, requiring the algorithm to ensure fast solution processing.

[0128] The decentralized clustering algorithm-driven target allocation method in this invention obtains the target set clusters attacked by each platform after completing the first step of autonomous clustering. The second step involves sequentially arranging the targets within each cluster. This means that without changing the members of each target set cluster, only the order of the target members is changed to minimize the attack path of the entire target sequence. Since the sorting problem now involves target sets clusters, the number of members is significantly smaller than the number of targets in the original problem. Furthermore, due to the design of the clustering benefit function, the number of targets in each cluster tends to be consistent. Therefore, an exhaustive method is used to sequentially arrange the targets within the clusters. Clearly, the time complexity of the algorithm is reduced to O(n^2). i -1)!), where n i This represents the number of targets in the target cluster assigned to the platform.

[0129] Sequential arrangement refers to a way of arranging elements in a set, where the order of elements is determined according to a certain rule or pattern. For example, a sequence may require elements to be arranged in ascending or descending order according to a certain attribute of the elements. Specifically, in this invention, the elements can be arranged in ascending or descending order according to attributes such as the target number, value, lower bound of the required ammunition quantity, and upper bound of the required ammunition quantity. This invention does not limit the specific method of the above sequential arrangement.

[0130] In some embodiments, sequentially arranging the clusters in the result of the cluster autonomous clustering calculation includes: sequentially arranging the target nodes in the clusters in the calculation result using an exhaustive method.

[0131] The following is another embodiment of the present invention that provides a detailed explanation of the above technical solution, including the following steps:

[0132] 1) Initialize multi-platform, multi-target information

[0133] The target allocation problem is simulated by attacking 12 targets from 4 platforms. The initial information of each platform and target is shown in the table below:

[0134] Table 1 Simulation target information for multi-objective assignment

[0135]

[0136] Multi-objective allocation simulation platform information

[0137]

[0138] The parameters of the benefit function in the simulation are set as shown in the table below:

[0139] Table 2 Simulation Algorithm Parameters

[0140]

[0141] Simulation benefit function parameters and The selection criteria should be determined by the order of magnitude of the target value and distance in actual combat. In this simulation experiment, the target value was selected as a value between 0 and 1, while the distance was concentrated in two digits. Therefore, [the following criteria were used]. yes 100 times.

[0142] 2) Output and analyze simulation results

[0143] The decentralized autonomous clustering process, after the first step of cluster autonomous clustering, yields the target set cluster information for each platform as shown in the table below:

[0144] Table 3 Simulation results of cluster autonomous clustering

[0145]

[0146] After obtaining the target set cluster information after the cluster's autonomous clustering, the second step is to sequentially arrange the targets in the cluster, and finally obtain the sequential attack sequence in each target set cluster as shown in the table below.

[0147] Table 4 Simulation results of target attack sequences in target allocation clusters

[0148]

[0149] Where 0 represents the same starting point across all platforms, returning after completing the attack mission; the final multi-platform target attack path is as follows: Figure 4 As shown.

[0150] Results of the target assignment method dominated by decentralized clustering algorithms:

[0151] Table 5

[0152]

[0153] As can be seen from the hypothetical simulation allocation results, the method proposed in this invention achieves full allocation of platform ammunition under constraints of range, platform payload ammunition quantity, and target hit ammunition quantity, and minimizes range and fuel consumption without changing the attack effectiveness, ultimately achieving a relatively ideal allocation effect.

[0154] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

[0155] Furthermore, although the actions or steps are described in a specific order, this should be understood as requiring that such actions or steps be performed in the specific order shown or in sequential order, or requiring that all illustrated actions or steps be performed to achieve the desired result. In certain environments, multitasking and parallel processing may be advantageous. Similarly, although several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations.

Claims

1. A decentralized clustering method for multi-objective allocation in clusters, comprising the following steps: Step S1: Determine a multi-combat target allocation model, which includes: multiple platforms for combat and multiple target nodes for attack during combat. Each platform includes a platform origin node. Each platform includes a first constraint attribute. Each target node includes a second constraint attribute. The first constraint attribute includes: maximum payload and maximum range. The second constraint attribute includes: target value, lower bound of required ammunition quantity, and upper bound of required ammunition quantity. Determine the target constraints for the multi-operational target allocation model; Step S2, perform multi-objective assignment, including: Obtain the benefit functions of multiple platforms for multiple target nodes; Perform cluster autonomous clustering calculations; The clusters in the results of the autonomous clustering calculation of the cluster are sequentially arranged. In step S2, obtaining the benefit function of multiple platforms for multiple target nodes includes: obtaining the benefit function of multiple platforms for multiple target nodes based on the value weight and distance weight of the platform relative to the target node, the attack value and distance of the platform relative to the target node, and the number of platforms already assigned to the target node, specifically including: platform For the goal Benefit function The calculation method is as follows: in , Platforms relative target Value weight and distance weight , Representing the platform Relative to the target The attack value and range, The target has now been assigned. A collection of platforms The number of platform members in the union of the set of platforms already assigned to target k and the currently assigned platform i; The method by which the platform obtains the attack value relative to the target node is as follows: When the target node has more remaining maximum ammunition than the platform has more remaining ammunition, the attack value is the target value of the target node multiplied by the platform's remaining ammunition. When the remaining maximum amount of ammunition that the target node can withstand does not exceed the remaining amount of ammunition on the platform, the attack value is the target value of the target node multiplied by the remaining maximum amount of ammunition that the target node can withstand. The attack value is zero when the lower bound of the target node's required ammunition quantity is greater than the platform's remaining ammunition quantity.

2. The decentralized clustering method for multi-objective allocation in clusters as described in claim 1, characterized in that, In step S1, the target constraints include: the total amount of ammunition allocated to each target node is within the upper and lower bounds of the required ammunition amount for each target node; the sum of the ammunition allocated to each target node by each platform does not exceed the maximum payload of each platform; the path length of each platform through the target nodes does not exceed the maximum range of each platform; the path of each platform through the target nodes is continuous; each platform only allocates ammunition to the target nodes it passes through; each platform is located at the platform starting node of each platform; there cannot be internal loops in the sequence of target nodes allocated to each platform; multi-target allocation is performed based on the value of the multiple target nodes and the amount of ammunition allocated to the target nodes by the multiple platforms.

3. The decentralized clustering method for multi-objective allocation in clusters as described in claim 2, characterized in that, In step S2, the execution of cluster autonomous clustering calculation includes: Obtain the initial state parameters of the target node and platform; For each target node, all platforms are traversed. If the relationship between the platform and the target node satisfies the target constraint conditions of the multi-combat target allocation model in step S1, the target node with the maximum benefit function is allocated to the corresponding platform, and the initial state parameters of the target node and the platform are updated. The initial state parameters of the target node include the remaining maximum amount of ammunition to be hit; the initial state parameters of the platform include the remaining amount of ammunition and range.

4. The decentralized clustering method for multi-objective allocation in clusters as described in claim 3, characterized in that, In step S2, sequentially arranging the clusters in the results of the autonomous clustering calculation includes: sequentially arranging the target nodes in the clusters in the calculation results using an exhaustive method.

Images