Optimization Method for Weapon and Equipment Deployment Points Driven by Rapid Closure of the Kill Chain

Abstract

An optimization method for the deployment points of weapon and equipment driven by the rapid closure of the kill chain, which relates to the field of equipment technology. Aiming at the problem of slow speed in generating the deployment point information of weapon and equipment in the existing technology, this application first constructs a kill chain model for specific tasks based on the correlation relationships among various pieces of equipment involved in the system confrontation process; then, according to the node information of the weapon and equipment to be optimized, the node positions are optimized through the proposed new large-scale multi-objective evolution method; finally, the kill chain formed by the optimized weapon and equipment has a fast closure speed, high connectivity, and strong network connectivity importance. For the combat system in offensive and defensive confrontation, the rapid closure of the kill chain needs to simultaneously possess a fast closure speed, high connectivity, and strong network connectivity importance. The application of this application can quickly generate the deployment point information of weapon and equipment with a rapid kill chain closure.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment, and specifically to an optimization method for the deployment points of weapon equipment driven by the rapid closure of the kill chain. Background Art

[0002] In practical applications, there are a large number of optimization problems, including multiple conflicting objectives and decision variables. These problems are called multi-objective optimization problems (MOPs). In order to achieve the best trade-off between these objectives, decision-makers expect to obtain a set of evenly distributed optimal solutions, rather than a single optimal solution that performs well on one MOP. Compared with mathematical programming methods, multi-objective evolutionary algorithms (MOEAs), as a population-based meta-heuristic method, stand out due to their excellent mechanisms. The purpose of MOEA is to find a non-dominated solution set in the objective space that is as close as possible to the Pareto front (PF). Although various methods have been proposed, their performance in dealing with large-scale MOPs (LMOPs) is mediocre. In fact, LMOPs have widely existed in various fields, such as software engineering, data mining, scheduling, and economy. Although there is no standard definition, an MOP can usually be designated as an LMOP because it contains more than 100 decision variables. However, compared with general MOPs, LMOPs have more stringent requirements for the search performance of their corresponding processing algorithms. In the prior art, when dealing with the problem of weapon equipment point deployment, there is a defect that the speed of generating weapon equipment deployment point information is slow. Summary of the Invention

[0003] The purpose of the present invention is: aiming at the problem of slow speed of generating weapon equipment deployment point information in the prior art, to propose an optimization method for the deployment points of weapon equipment driven by the rapid closure of the kill chain.

[0004] The technical solution adopted by the present invention to solve the above technical problems is:

[0005] An optimization method for the deployment points of weapon equipment driven by the rapid closure of the kill chain includes the following steps:

[0006] Step 1: Obtain the association relationships between equipment included in the combat system confrontation process, and construct a kill chain model based on this;

[0007] Step 2: Use the maximum connected time of the kill chain, the closure of the kill chain, and the importance of the kill chain as optimization functions in the multi-objective optimization problem to optimize the kill chain model;

[0008] Step 3: Use the optimized kill chain model to optimize the positions of all nodes in the deployment area during the combat system confrontation process.

[0009] Further, the specific steps of Step 1 are:

[0010] Classify the weapons and equipment in the combat system according to categories, and the categories include reconnaissance weapon and equipment node D, decision-making weapon and equipment node A, strike weapon and equipment node B, and target weapon and equipment node C;

[0011] The main combat mission of the reconnaissance weapon and equipment node D is to detect, reconnoiter, warn, and monitor enemy targets, ensure battlefield target information, and transmit the information to other equipment nodes in the system;

[0012] The main combat mission of the decision-making weapon and equipment node A is to process and analyze the input battlefield information, make action command decisions, transmit the decision information to the strike node, and conduct command and control;

[0013] The main combat mission of the strike weapon and equipment node B is the weapon and equipment with strike capabilities;

[0014] The target weapon and equipment node C is the target equipment that the strike node needs to strike;

[0015] The combat attributes of the reconnaissance weapon and equipment node D, decision-making weapon and equipment node A, strike weapon and equipment node B, and target weapon and equipment node C are as follows:

[0016]

[0017] Among them, attribute 1 represents the deployment area of the node, attribute 2 represents the function ability of the node in the corresponding area, attribute 3 represents that the node is subject to distance constraints, attribute 4 represents that the node is subject to connectivity constraints, that is, 1 in the first column means it can be connected, and 1 in the second column means it can connect to others, attribute 5 represents the importance of the node, and attribute 6 represents the residence time of the node;

[0018] Nodes can only be connected in the order of D→A→B→C→D, and the following types of edges in the combat system network are obtained:

[0019] Type Meaning C→D Combat tasks such as reconnaissance and surveillance of target nodes D→A Information transmission activity from reconnaissance node to decision-making node A→B Activity of transmitting decision-making instructions from decision-making node to influencing node B→C Precision strike activity on target nodes

[0020] Use attribute 3 and attribute 4 to obtain a directed connectivity matrix, and then obtain a kill chain model.

[0021] Furthermore, the connectivity is expressed as:

[0022]

[0023] Where d I→O represents the Euclidean distance between the I node and the O node, I DIt represents the attribute 3 corresponding to the I node. The types of the I node and the O node are one of the reconnaissance weapon equipment node D, the decision-making weapon equipment node A, the strike weapon equipment node B, and the target weapon equipment node C.

[0024] Furthermore, the multi-objective optimization problem in the second step is expressed as:

[0025] F(X,Y,Z) = [f 1 (X,Y,Z), f 2 (X,Y,Z), f 3 (X,Y,Z)] T

[0026] Among them, (x,y,z) represents the position of the node, and f 1 represents the maximum connected time of the kill chain, and f 2 represents the closeness of the kill chain, and f 3 represents the importance of the kill chain.

[0027] Furthermore, the maximum connected time f 1 of the kill chain is expressed as:

[0028] f 1 = min{v 1 , v 2 ,..., v N}

[0029]

[0030] Among them, v 1 , v 2 ,..., v N represent the kill chain closeness speeds corresponding to N weapon equipments respectively. Dis 1 represents the length of the complete kill chain where the first weapon equipment is located, t 1 represents the closing time corresponding to the kill chain where this weapon equipment is located, d SD represents the distance between the S-class weapon equipment and the D-class weapon equipment in the kill chain where this weapon equipment is located, d DB represents the distance between the D-class weapon equipment and the B-class weapon equipment in the kill chain where this weapon equipment is located, d BC represents the distance between the B-class weapon equipment and the C-class weapon equipment in the kill chain where this weapon equipment is located, V SD represents the communication speed between the S-class weapon equipment and the D-class weapon equipment in the kill chain where this weapon equipment is located, V DB represents the communication speed between the D-class weapon equipment and the B-class weapon equipment in the kill chain where this weapon equipment is located, V BCIndicates the communication speed between Class B weapon equipment and Class C weapon equipment in the kill chain where this weapon equipment is located, t S Indicates the corresponding residence time of Class S weapon equipment in the kill chain where this weapon equipment is located, t D Indicates the corresponding residence time of Class D weapon equipment in the kill chain where this weapon equipment is located, t B Indicates the corresponding residence time of Class B weapon equipment in the kill chain where this weapon equipment is located, t C Indicates the corresponding residence time of Class C weapon equipment in the kill chain where this weapon equipment is located.

[0031] Furthermore, the closeness f of the kill chain 2 Is expressed as:

[0032]

[0033] Where k i Indicates the number of edges possessed by the i-th Class C weapon equipment that can be connected by the DABC chain, and nC represents the number of Class C nodes.

[0034] Furthermore, the importance f of the kill chain 3 Is expressed as:

[0035]

[0036] Where N all Indicates the quantity in other kill chains, C on Indicates node connectivity, N represents the number of kill chains, j represents the j-th weapon equipment, and N represents the total number of weapon equipment.

[0037] Furthermore, the probability of the success of the kill chain operation activities in the kill chain model is expressed as:

[0038] w t = h(a u x , a v y ), h(·) ∈ (0, 1)

[0039] Where a u x And a v y Are respectively the sets of performance indicators of nodes v u x And v u y , and h(·) represents the mapping relationship between node performance indicators and weights.

[0040] Furthermore, the specific steps of Step 3 are as follows:

[0041] Obtain the number of weapon and equipment to be optimized and the location information of each node within the deployment area during the combat system confrontation process, obtain the domain control parameters, and set the maximum number of evaluations. Then, input the number of weapon and equipment to be optimized, the location information of each node, the domain control parameters, and the maximum number of evaluations into the optimized kill chain model to obtain the optimized equipment deployment location.

[0042] The specific steps executed by the optimized kill chain model are as follows:

[0043] Step 3-1: Initialize the number of weapon and equipment to be optimized and the location information of each node in the search space as the initial state.

[0044] Step 3-2: Set the external document to be consistent with the initial state.

[0045] Step 3-3: When the number of model evaluations is less than the maximum number of evaluations, execute Steps 3-4 to 3-10.

[0046] Step 3-4: Generate elite individuals according to the external document and the neighborhood control parameters.

[0047] Step 3-5: Construct a mating pool according to the elite individuals and the initial state.

[0048] Step 3-6: For each piece of equipment in the initial state, execute Steps 3-7 and 3-8.

[0049] Step 3-7: Generate offspring according to the mating pool and each piece of equipment.

[0050] Step 3-8: Incorporate the generated offspring into the external document and remove the individual with the worst contribution degree.

[0051] Step 3-9: End after traversing all the equipment in the initial state.

[0052] Step 3-10: Use the external document to replace the initial state.

[0053] Step 3-11: End, and obtain the optimized equipment deployment location.

[0054] Furthermore, the contribution degree is determined through the following steps:

[0055] When an equipment individual is in the last layer of non-dominated sorting within the current iteration; or when the non-dominated sorting is only one layer and the hypervolume value formed by an equipment individual is the smallest, that is, the contribution degree of this equipment individual is the worst.

[0056] The beneficial effects of the present invention are:

[0057] In this application, first, for the correlation relationships among various equipment involved in the system confrontation process, a kill chain model oriented to specific tasks is constructed; then, according to the node information of the weapon equipment to be optimized, the node positions are optimized through the proposed new large-scale multi-objective evolution method; finally, the kill chain formed by the optimized weapon equipment has a fast closing speed, high connectivity, and strong network connectivity importance.

[0058] For the combat system in offensive and defensive confrontation, the fast closing of the kill chain requires simultaneously having a fast closing speed, high connectivity, and strong network connectivity importance. The application of this application can quickly generate the deployment point information of weapon equipment with a fast kill chain closing. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic diagram of the kill chain model of this application;

[0060] Figure 2 It is an example diagram for optimizing the kill chain closing speed;

[0061] Figure 3 It is a schematic diagram of the pseudo code of the optimization algorithm of this application;

[0062] Figure 4 It is a schematic diagram of the existing algorithm 2;

[0063] Figure 5 It is a schematic diagram of the existing algorithm 3;

[0064] Figure 6 It is a schematic diagram of the existing algorithm 4;

[0065] Figure 7 It is a schematic diagram of the existing algorithm 5. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] It should be particularly noted that, without conflict, the various embodiments disclosed in this application can be combined with each other.

[0067] DETAILED DESCRIPTION OF THE EMBODIMENT 1: Refer to Figure 1 This embodiment is specifically described. The method for optimizing the deployment point of weapon equipment driven by the fast closing of the kill chain according to this embodiment is characterized by including the following steps:

[0068] Step 1: Obtain the correlation relationships among the equipment involved in the combat system confrontation process, and construct a kill chain model based on this;

[0069] Step 2: Use the maximum connected time of the kill chain, the closability of the kill chain, and the importance of the kill chain as the optimization functions in the multi-objective optimization problem to optimize the kill chain model;

[0070] Step 3: Use the optimized kill chain model to optimize the positions of all nodes within the deployment area during the combat system confrontation process.

[0071] Specific Embodiment 2: This embodiment is a further elaboration of Specific Embodiment 1. The difference between this embodiment and Specific Embodiment 1 is that the specific steps of Step 1 are as follows:

[0072] Node Model:

[0073] Classify the weapons and equipment in the combat system according to their categories. The categories include reconnaissance weapon and equipment nodes D, decision-making weapon and equipment nodes A, strike weapon and equipment nodes B, and target weapon and equipment nodes C.

[0074] The main combat mission of the reconnaissance weapon and equipment node D is to detect, scout, warn, and monitor enemy targets, ensure battlefield target information, and transmit the information to other equipment nodes in the system.

[0075] The main combat mission of the decision-making weapon and equipment node A is to process and analyze the input battlefield information, make action command decisions, transmit the decision information to the strike nodes, and conduct command and control.

[0076] The main combat mission of the strike weapon and equipment node B is the weapons and equipment with strike capabilities.

[0077] The target weapon and equipment node C is the target equipment that the strike node needs to strike.

[0078] Note: The corresponding positions of the three types of weapons and equipment D, A, and B need to be optimized. The corresponding position of the C-type weapon and equipment is the enemy position, which is known.

[0079] The combat attributes possessed by the reconnaissance weapon and equipment node D, decision-making weapon and equipment node A, strike weapon and equipment node B, and target weapon and equipment node C are as follows:

[0080]

[0081] Among them, Attribute 1 represents the deployment area of the node, Attribute 2 represents the function ability of the node in the corresponding area, Attribute 3 represents the distance constraint on the node, Attribute 4 represents the connectivity constraint on the node, that is, 1 in the first column means it can be connected, and 1 in the second column means it can connect to others, Attribute 5 represents the importance of the node, and Attribute 6 represents the residence time of the node.

[0082] The set of nodes can be represented by V = V D ∪V A ∪V B ∪V C to represent, V D ={v1 D , v 2 D ,..., v nD D}, V A = {v 1 A , v 2 A ,..., v nA A}, V B = {v 1 B , v 2 B ,..., v nB B}, and V C = {v 1 C , v 2 C ,..., v nC C}, respectively represent the sets of reconnaissance nodes, decision nodes, strike nodes, and target nodes. nD, nA, nB, and nC represent the numbers of the four types of nodes respectively. In addition, the differences in the priorities of strike tasks can be reflected by weights. Let w T = (w 1 T , w 2 T ,..., w N T ) represent the importance weight vector of N target nodes. The element w T in w 1 T represents the weight of the l-th target node, and its value is between 0 and 1 and satisfies ∑w T = 1.

[0083] where N = nD, nA, nB, and nC.

[0084] Edge model:

[0085] Model the edges. The directed edges in the combat system network represent the functional interaction relationships between equipment nodes. According to the type combinations of nodes, there are theoretically 16 different types of directed edges. After considering the actual requirements, only 7 types of directed edges are analyzed.

[0086] Nodes can only be connected in the order of D → A → B → C → D. Thus, the types of edges in the combat system network are as follows:

[0087] Type Meaning C→D Combat tasks such as reconnaissance and surveillance of target nodes D→A Information transmission activity from reconnaissance node to decision-making node A→B Activity of transmitting decision-making instructions from decision-making node to influencing node B→C Precision strike activity on target nodes

[0088] The directed connectivity matrix is obtained by using Attribute 3 and Attribute 4, and then the kill chain model is obtained.

[0089] Example of the directed connectivity matrix for a certain task C→D→A→B

[0090] <![CDATA[D 1 > <![CDATA[D 2 > <![CDATA[A 1 > <![CDATA[A 2 > <![CDATA[B 1 > <![CDATA[B 2 > <![CDATA[B 3 > <![CDATA[B 4 > <![CDATA[C n > <![CDATA[D 1 > 0 0 1 1 0 0 0 0 1 <![CDATA[D 2 > 0 0 1 1 0 0 0 0 1 <![CDATA[A 1 > 1 1 1 0 1 1 0 1 0 <![CDATA[A 2 > 0 1 0 1 1 0 1 1 0 <![CDATA[B 1 > 0 0 1 1 1 0 1 0 0 <![CDATA[B 2 > 0 0 1 0 0 1 0 0 0 <![CDATA[B 3 > 0 0 0 1 1 0 1 0 0 <![CDATA[B 4 > 0 0 1 1 0 0 0 1 0 <![CDATA[C n > 1 1 0 0 0 0 0 0 1

[0091] The nodes can only be connected in the order of D→A→B→C→D, and other orders are not allowed.

[0092] According to the above operations, an edge model of weaponry and equipment can be formed to constitute the kill chain model. As Figure 1 shown.

[0093] Specific Embodiment 3: This embodiment is a further description of Specific Embodiment 2. The difference between this embodiment and Specific Embodiment 2 is that the connectivity is expressed as:

[0094]

[0095] where d I→O represents the Euclidean distance between the I node and the O node, and I D represents Attribute 3 corresponding to the I node. The types of the I node and the O node are one of the reconnaissance weaponry and equipment node D, the decision-making weaponry and equipment node A, the strike weaponry and equipment node B, and the target weaponry and equipment node C.

[0096] After forming the directed connectivity matrix, it should be noted that the connectivity of a certain node is jointly determined by the corresponding Attribute 3, Attribute 4, and the connectivity matrix of the node. That is to say, whether each weaponry and equipment can have an edge with other equipment not only needs to consider the connectivity matrix but also requires the distance between the equipment to be less than the constraint distance.

[0097] For example:

[0098] Connectivity of node C:

[0099]

[0100] where d D→C represents the Euclidean distance between a certain D node and a certain C node, and D D represents Attribute 3 corresponding to the D node.

[0101] Specific Embodiment 4: This embodiment is a further description of Specific Embodiment 3. The difference between this embodiment and Specific Embodiment 3 is that the multi-objective optimization problem in Step 2 is expressed as:

[0102] F(X,Y,Z)=[f 1 (X,Y,Z),f 2 (X,Y,Z),f 3(X,Y,Z) T [ [

[0103] [ Among them, (x,y,z) represents the position of the node, and f 1 [ represents the maximum connected time of the kill chain, and f 2 [ represents the closeness of the kill chain, and f 3 [ represents the importance of the kill chain. [

[0104] [ The problem of optimizing the closing speed of the formed kill chain is considered as a large-scale multi-objective optimization problem for solution. The large-scale multi-objective optimization problem is one of the branches in the field of evolutionary optimization, which mainly faces multi-objective optimization problems where the objectives are mutually coupled and the number of decision variables is large. Among them, the optimization variable is the position (x,y,z) of each node, and the optimization problem is considered as a three-objective minimization optimization problem F(X,Y,Z) = [f 1 [ (X,Y,Z), f 2 [ (X,Y,Z), f 3 [ (X,Y,Z)] T [ . [

[0105] [ Specific Embodiment 5: This embodiment is a further description of Specific Embodiment 4. The difference between this embodiment and Specific Embodiment 4 is that the maximum connected time f of the kill chain 1 [ is expressed as: [

[0106] [ f 1 [ = min{v 1 [ , v 2 [ ,..., v N [} [

[0107] [ [ [

[0108] [ Among them, v 1 [ , v 2 [ ,..., v N [ represent the closing speed of the kill chain corresponding to N weapons and equipment respectively. Dis 1 [ represents the length of the complete kill chain where the first weapon and equipment is located (if the weapon and equipment exists in multiple kill chains, take the average value), t 1 [ represents the closing time corresponding to the kill chain where the weapon and equipment is located, d SD [ represents the distance between the S-type weapon and equipment and the D-type weapon and equipment in the kill chain where the weapon and equipment is located, d DB [ represents the distance between the D-type weapon and equipment and the B-type weapon and equipment in the kill chain where the weapon and equipment is located, d BC [ represents the distance between the B-type weapon and equipment and the C-type weapon and equipment in the kill chain where the weapon and equipment is located, V SD [ represents the communication speed between the S-type weapon and equipment and the D-type weapon and equipment in the kill chain where the weapon and equipment is located, VDB Denotes the communication speed V between weapon equipment of type D and weapon equipment of type B in the kill chain where the weapon equipment is located. BC Denotes the communication speed t between weapon equipment of type B and weapon equipment of type C in the kill chain where the weapon equipment is located. S Denotes the residence time t corresponding to weapon equipment of type S in the kill chain where the weapon equipment is located. D Denotes the residence time t corresponding to weapon equipment of type D in the kill chain where the weapon equipment is located. B Denotes the residence time t corresponding to weapon equipment of type B in the kill chain where the weapon equipment is located. C Denotes the residence time corresponding to weapon equipment of type C in the kill chain where the weapon equipment is located.

[0109] Kill chain closing speed:

[0110] For each of the N formed kill chains (N represents the number of complete kill chains), calculate its connection time (t 1 , t 2 , …, t N ) and the corresponding kill chain distance (DBD 1 , DBD 2 , …, DBD N ). It should be noted that when calculating the connection time, the sum of the residence times of the corresponding kill chain passing through the nodes needs to be added. As Figure 2 shown, its maximum kill chain connection time f 1 is expressed as:

[0111]

[0112] f 1 = min{v 1 , v 2 ,..., v N}.

[0113] Specific implementation method six: This implementation method is a further description of specific implementation method five. The difference between this implementation method and specific implementation method five is that the kill chain closeness f 2 is expressed as:

[0114]

[0115] where k i denotes the number of edges possessed by the i-th weapon equipment of type C that can be connected by the DABC chain, and nC denotes the number of type C nodes.

[0116] For the formed kill chain, its connectivity is positively correlated with the number of complete kill chains formed by Class C weapons and equipment in the model. Considering that the multi-objective optimization problem is a minimization problem, the negation process is used in the kill chain connectivity objective function model:

[0117] Specific Embodiment 7: This embodiment is a further description of Specific Embodiment 6. The difference between this embodiment and Specific Embodiment 6 is that the importance f of the kill chain 3 is expressed as:

[0118]

[0119]

[0120] where N all represents the quantity in other kill chains, C on represents the node connectivity, N represents the number of kill chains, j represents the jth weapon and equipment, and N represents the total number of weapons and equipment.

[0121] For the formed kill chain model, first calculate the quantity N of T-type nodes covered simultaneously in other kill chains in a certain complete kill chain all ; then calculate the T-type node connectivity C on , which is obtained by dividing the number of complete kill chains by the number of T-type nodes, as follows

[0122]

[0123] Finally, calculate the network connectivity importance. Considering that the multi-objective optimization problem is a minimization problem, the negation process is used in the kill chain network connectivity importance objective function model:

[0124]

[0125] Specific Embodiment 8: This embodiment is a further description of Specific Embodiment 7. The difference between this embodiment and Specific Embodiment 7 is that the probability of the success of the kill chain operation activities in the kill chain model is expressed as:

[0126] w t =h(a u x ,a v y ), h(·) ∈ (0,1)

[0127] where a u x and a v y are respectively the nodes v u x and v u ySet of performance metrics, where h(·) represents the mapping relationship between node performance metrics and weights.

[0128] Model metrics:

[0129] After forming the kill chain model, for the formed kill chain, the following can be counted:

[0130] Number of kill chains N formed;

[0131] Length L corresponding to each kill chain;

[0132] Calculate the success probability of different kill chains: It is reflected by assigning weights to the edges. Let e t =(v u x , v u y ) represent the t-th directed edge in the network. The nodes at both ends of e t are v u x and v u y (X and Y represent the types of nodes). Then the weight w t of the edge e t is shown in the following formula.

[0133] w t =h(a u x , a v y ), h(·) ∈ (0, 1)

[0134] In the formula, a u x and a v y are respectively the sets of performance metrics of nodes v u x and v u y ; h(·) represents the mapping relationship between node performance metrics and weights, which can be obtained by means of mathematical analysis, etc. (quantitative analysis is carried out for those indicators that can be quantitatively analyzed in the corresponding tasks, such as the number of kill chains, battle damage, battle damage ratio, etc.), and the value range is between 0 and 1. The weight w t obtained in this way can be understood as the probability of the edge e t being connected, that is, the probability of the combat activity represented by e t being successful.

[0135] Specific implementation method nine: This implementation method is a further description of specific implementation method eight. The difference between this implementation method and specific implementation method eight is that the specific steps of step three are as follows:

[0136] Obtain the quantity of weaponry and equipment to be optimized and the location information of each node within the deployment area during the combat system confrontation process, obtain the domain control parameters, and set the maximum number of evaluations. Then, input the quantity of weaponry and equipment to be optimized, the location information of each node, the domain control parameters, and the maximum number of evaluations into the optimized kill chain model to obtain the optimized equipment deployment locations;

[0137] The specific steps executed by the optimized kill chain model are as follows:

[0138] Step 3-1: Initialize the quantity of weaponry and equipment to be optimized and the location information of each node in the search space as the initial state;

[0139] Step 3-2: Set up an external document to be consistent with the initial state;

[0140] Step 3-3: When the number of model evaluations is less than the maximum number of evaluations, execute Steps 3-4 to 3-10;

[0141] Step 3-4: Generate elite individuals based on the external document and the neighborhood control parameters;

[0142] Step 3-5: Construct a mating pool based on the elite individuals and the initial state;

[0143] Step 3-6: For each piece of equipment in the initial state, execute Steps 3-7 and 3-8;

[0144] Step 3-7: Generate offspring based on the mating pool and each piece of equipment;

[0145] Step 3-8: Incorporate the generated offspring into the external document and remove the individual with the worst contribution;

[0146] Step 3-9: End after traversing all the equipment in the initial state;

[0147] Step 3-10: Use the external document to replace the initial state;

[0148] Step 3-11: End, and obtain the optimized equipment deployment locations.

[0149] For the formed optimization model, use the elite strategy-driven large-scale multi-objective optimization algorithm, adopt a two-stage recombination operator with a brand-new search strategy, clearly utilize the local similarity neighborhood attribute between the early decision-making space and the objective space of the solution to be optimized for this optimization problem, and guide the individuals to generate a diverse approximation of the Pareto front to greatly improve the search efficiency. The pseudo-code of the optimization algorithm is as Figure 3 shown:

[0150] Algorithm 1( Figure 3) Input: A model with m objectives and n (the number of attribute variables) variables, the number of weapon systems P to be optimized and deployed, the parameter α for adjusting the neighborhood range, and the maximum number of function evaluations MaxFes. The proposed algorithm first adaptively divides the non-dominant individuals in the objective space into multiple neighborhoods, and respectively selects representative objectives as elite objectives. Then, it constructs a mating pool in the decision space using these objectives, as shown in line 4. As the evolution progresses, this strategy can improve the efficiency of maintaining diversity. Then, these elite individuals are used as clustering centroids to collect objectives with similar information into the corresponding mating pools in the decision space, as shown in line 5, to enhance the exploration ability of the objectives in each mating pool. This can effectively avoid the over-concentration of objectives in the vicinity of the global optimal solution region and further enhance diversity. Finally, for each objective x in the mating pool MatBngPool i , the differential evolution (DE) operator is used to generate the offspring y i , and the environmental selection operator is used to delete the worst individuals in the population, as shown in lines 7 and 8 respectively. This process will be iterated continuously until the termination condition is met.

[0151] Algorithm 2( Figure 4 ) In EGEA, the elite objectives of the current population consist of three parts, namely the extreme points, non-dominated objectives, and neighboring potential objectives. Specifically, to clarify how our elite selection strategy works, the detailed pseudocode is given in Algorithm 2. First, the Arc is divided into different levels using the non-dominated sorting method. Then, the extreme points E of the current population are determined in line 2, and the extreme hyperplane H is formed in line 3, where the extreme points are defined as

[0152]

[0153] where m is the number of objective functions. In line 4, the niche ratio r is adaptively updated to

[0154]

[0155] where t is the ratio of the number of elite objectives to the number of non-dominated objectives. Here, the initial values of r and t are set to 1 and 0 respectively. Then, assuming that the neighborhood of each elite objective is an (m - 1)-dimensional polytope of size R, its size can be determined by

[0156] R = r·H·[(1 + α) 1 / (m-1) - 1] (3)

[0157] where α > 0 is a scaling factor used to adjust the neighborhood range. It should be emphasized that no elite target can be dominated by any other target within its surrounding range. In addition, there may be an overlap between two neighborhoods, which can further improve its search performance. At line 6, the distance between each target and H is calculated, and then all the targets are sorted in descending order at line 7. Then, the process from lines 8 to 14 is iteratively repeated until Arc is empty. During this process, since the extreme point E is regarded as an elite target, it is removed from the arc. Then, the individuals with the maximum distances to the extreme hyperplane H and F1 are collected and stored in the elitist algorithm, and then the individual itself and its neighboring targets are removed from Arc; afterwards, the extreme point is added to Elitist (the set of elite targets) at line 15, and the ratio of the Elitist targets to the number of non-dominated targets is updated at line 16. Finally, Elitist is returned at line 17.

[0158] Algorithm 3( Figure 5 ): After identifying the elite targets in the objective space, a k-meanD algorithm inspired by a cheap clustering method (CM) is introduced to construct a mating pool in the decision space. Specifically, instead of adopting a search strategy directly acting on the decision space, the algorithm divides the population into multiple clusters with elite targets as the centroids to guide the search. Each cluster is regarded as a mating pool, and its neighborhood relationship is determined by the local neighborhood relationship among the elite targets in the current population. In this way, the population can be explored separately in multiple sub-regions, effectively capturing the local optimal regions and the global optimal region. The learned clustering information directly guides the consistency of the search direction with the intersection point of the non-dominated optimal set, thereby promoting the maintenance of diversity in the decision space. To clarify the basic idea of mating pool construction, its detailed process is shown in Algorithm 3, with the inputs: the elite targets Elitist and the target set n. In line 1, D equals |Elitist| representing the number of clusters. In line 2, Q 1 , …, Q s are initialized as empty sets to store the mating parents. In line 3, each elite target c j ∈ c 1 , …, selects cD as the centroid of Q 1 cluster, …, Q s . Lines 4 - 7 calculate the Euclidean distance between each target xB ∈ N and the cluster centroid, where dBD(a, b) measures the Euclidean distance between a and b to measure their differences for constructing the mating pool.

[0159] Algorithm 4( Figure 6 ): First, at line 1, the mating pool Qj of each target xB is determined, where j ∈ 1, …, D and B ∈ 1, … Then, a DE operator is applied to two randomly selected parent solutions xB 1 and xB 2 in the mating pool at line 2 to generate a new solution at line 3 As follows:

[0160]

[0161] Where F is the scaling factor, which is used to scale the influence of the parent solution selected when calculating the mutation value, CR represents the parameter that controls the influence of the parent solution in the offspring, and r = rand() represents a uniform random generator within the range of [0, 1]. In line 4, a repair mechanism is applied to correct the components outside the search boundary. Next, a trial solution y in line 5 is generated using the mutation operator i , which can be achieved through

[0162]

[0163] Where

[0164]

[0165] Where PM is the mutation probability and ηm is the mutation distribution parameter. In line 6, if necessary, the trial target y can be repaired again i . Finally, a trial target y is returned in line 7 i .

[0166] Algorithm 5: Environmental selection operator, which deletes the worst equipment information according to the non-dominated sorting rank of all equipment

[0167] According to the above design content, by taking the kill chain closure speed, kill chain closure, and network connectivity importance as the objective function f(x), under the constraint conditions of a certain area (position feasible region), a large-scale multi-objective evolutionary algorithm is used for optimal design, so that the finally obtained kill chain closure speed meets the index requirements

[0168] Specific Embodiment Ten: This embodiment is a further description of Specific Embodiment Nine. The difference between this embodiment and Specific Embodiment Nine is that the contribution degree is determined through the following steps:

[0169] When an equipment individual is in the last layer of non-dominated sorting within the current iteration; or when the non-dominated sorting is only one layer, the hypervolume value composed of a certain equipment individual is the smallest, that is, the contribution degree of this equipment individual is the worst

[0170] It should be noted that the specific embodiments are only explanations and descriptions of the technical solutions of the present invention, and the scope of the right protection cannot be limited by this. Any changes that are only partial according to the claims and specifications of the present invention should still fall within the protection scope of the present invention

Claims

1. Optimization method for weapon and equipment deployment points driven by rapid closure of the kill chain, characterized in that it includes the following steps: Step 1: Obtain the association relationships between equipment involved in the combat system confrontation process, and construct a kill chain model based on this; Step 2: Use the maximum connected time of the kill chain, the closeness of the kill chain, and the importance of the kill chain as optimization functions in the multi-objective optimization problem to optimize the kill chain model; Step 3: Use the optimized kill chain model to optimize the positions of all nodes in the deployment area during the combat system confrontation process; Among them, the multi-objective optimization problem in Step 2 is expressed as: F(X, Y, Z) = [f 1 (X, Y, Z), f 2 (X, Y, Z), f 3 (X, Y, Z)] T Among them, (X, Y, Z) represents the position of the node, and f 1 represents the maximum connected time of the kill chain, and f 2 represents the closeness of the kill chain, and f 3 represents the importance of the kill chain; The maximum connected time f1 of the kill chain is expressed as: f 1 = min{v 1 , v 2 , …, v N} Among them, v 1 , v 2 ,..., v N represent the kill chain closure speeds corresponding to N weapon and equipment respectively. Dis 1 represents the length of the complete kill chain where the first weapon and equipment is located. t 1 represents the closure time corresponding to the kill chain where the weapon and equipment is located. d SD represents the distance between the S-class weapon and equipment and the D-class weapon and equipment in the kill chain where the weapon and equipment is located. d DB represents the distance between the D-class weapon and equipment and the B-class weapon and equipment in the kill chain where the weapon and equipment is located. d BC represents the distance between the B-class weapon and equipment and the C-class weapon and equipment in the kill chain where the weapon and equipment is located. V SD represents the communication speed between the S-class weapon and equipment and the D-class weapon and equipment in the kill chain where the weapon and equipment is located. V DB represents the communication speed between the D-class weapon and equipment and the B-class weapon and equipment in the kill chain where the weapon and equipment is located. V BC represents the communication speed between the B-class weapon and equipment and the C-class weapon and equipment in the kill chain where the weapon and equipment is located. t S represents the residence time corresponding to the S-class weapon and equipment in the kill chain where the weapon and equipment is located. t D represents the residence time corresponding to the D-class weapon and equipment in the kill chain where the weapon and equipment is located. t B represents the residence time corresponding to the B-class weapon and equipment in the kill chain where the weapon and equipment is located. t C represents the residence time corresponding to the C-class weapon and equipment in the kill chain where the weapon and equipment is located.

2. The optimization method for weapon and equipment deployment points driven by rapid closure of the kill chain according to claim 1, characterized in that the specific steps of Step 1 are: Classify the weapons and equipment in the combat system according to categories, and the categories include reconnaissance weapon and equipment nodes D, decision-making weapon and equipment nodes A, strike weapon and equipment nodes B, and target weapon and equipment nodes C; The combat mission of the reconnaissance weapon and equipment node D is to detect, reconnoiter, warn, and monitor the enemy target, ensure battlefield target information, and transmit the information to other equipment nodes in the system; The combat mission of the decision-making weapon and equipment node A is to process and analyze the input battlefield information, make action command decisions, transmit the decision information to the strike weapon and equipment node, and conduct command and control; The strike weapon and equipment node B is a weapon and equipment with strike capabilities; The target weapon and equipment node C is the target equipment that the strike weapon and equipment node needs to strike; The combat attributes possessed by the reconnaissance weapon and equipment node D, decision-making weapon and equipment node A, strike weapon and equipment node B, and target weapon and equipment node C are as follows: Among them, attribute 1 represents the deployment area of this node, attribute 2 represents the function ability of this node in the corresponding area, attribute 3 represents that this node is subject to distance constraints, attribute 4 represents that this node is subject to connectivity constraints, that is, 1 in the first column means it can be connected, and 1 in the second column means it can connect to others, attribute 5 represents the importance of this node, and attribute 6 represents the residence time of the node; Nodes can only be connected in the order of D→A→B→C→D, and the following types of edges in the combat system network are obtained: Use attribute 3 and attribute 4 to obtain a directed connectivity matrix, and then obtain a kill chain model.

3. The optimization method for weapon and equipment deployment points driven by rapid closure of the kill chain according to claim 2, characterized in that Connectivity is expressed as: Among them, d I→O represents the Euclidean distance between the I node and the O node, and I D represents the attribute 3 corresponding to the I node. The types of the I node and the O node are one of the reconnaissance weapon equipment node D, the decision-making weapon equipment node A, the strike weapon equipment node B, and the target weapon equipment node C.

4. The optimization method for weapon and equipment deployment points driven by rapid closure of the kill chain according to claim 3, characterized in that The kill chain closure f 2 is expressed as: Among them, k i represents the number of edges of the i-th type-C weapon and equipment that can be connected by the DABC chain, and nC represents the number of type-C nodes.

5. The optimization method for weapon and equipment deployment points driven by rapid closure of the kill chain according to claim 4, characterized in that The importance of the kill chain f 3 It is expressed as: Among them, N all represents the quantity in other kill chains, C on represents the node connectivity, j represents the j-th weapon and equipment, and N represents the total quantity of weapon and equipment.

6. The optimization method for weapon and equipment deployment points driven by rapid closure of the kill chain according to claim 5, characterized in that The probability of the success of the kill chain combat activity in the kill chain model is expressed as: Among them, and are respectively the sets of performance indicators of nodes and The mapping relationship between the node performance indicator and the weight is represented by h(·).

7. The optimization method for weapon and equipment deployment points driven by rapid closure of the kill chain according to claim 6, characterized in that The specific steps of Step 3 are as follows: Obtain the number of weapon and equipment to be optimized in the deployment area during the combat system confrontation process and the location information of each node, obtain the domain control parameters, and set the maximum number of evaluations. Then, input the number of weapon and equipment to be optimized, the location information of each node, the domain control parameters, and the maximum number of evaluations into the optimized kill chain model to obtain the optimized equipment deployment location; The specific steps executed by the optimized kill chain model are as follows: Step 3-1: Initialize the number of weapon and equipment to be optimized and the location information of each node in the search space as the initial state; Step 3-2: Set the external document to be consistent with the initial state; Step 3-3: When the number of model evaluations is less than the maximum number of evaluations, execute Steps 3-4 to 3-10; Step 3-4: Generate elite individuals according to the external document and the neighborhood control parameters; Step 3-5: Construct a mating pool based on the elite individuals and the initial state; Step 3-6: For each piece of equipment in the initial state, execute Steps 3-7 and 3-8; Step 3-7: Generate offspring according to the mating pool and each piece of equipment; Step 3-8: Incorporate the generated offspring into the external document and remove the individual with the worst contribution; Step 3-9: End after traversing all the equipment in the initial state; Step 3-10: Use the external document to replace the initial state; Step 3-11: End to obtain the optimized equipment deployment location.

8. The method for optimizing the deployment point of weapon and equipment driven by the rapid closure of the kill chain according to claim 7, characterized in that the contribution is determined through the following steps: When an equipment individual is in the last layer of non-dominated sorting within the current iteration; or when the non-dominated sorting is only one layer, the hypervolume value formed by a certain equipment individual is the smallest, that is, the contribution of this equipment individual is the worst.

Images