Unmanned cluster battle decision-making method and device

Abstract

The invention provides an unmanned cluster battle decision-making method and device. The method comprises the following steps: acquiring current state information and own resource information of a plurality of enemy target individuals in a cluster battle; based on the current state information of the plurality of enemy target individuals, estimating the state information of the plurality of enemy target individuals in the next time period by using an interactive multi-model filtering algorithm; inputting the state information of the plurality of enemy target individuals in the next time period and the resource information of the enemy target individuals into a constructed multi-target optimization model under a multi-constraint condition; solving the multi-target optimization model under the multi-constraint condition based on a multi-target optimization algorithm to obtain output target distribution information; and generating an own strike decision in different decision spaces according to the target distribution information. By adopting the scheme, the prediction of the enemy target individual is realized only by using the information observed in the confrontation, and the attack decision is obtained by comprehensively considering various constraint conditions in the battle, so that the method is more suitable for an actual confrontation scene.

Description

Technical Field

[0001] The present application relates to the field of aircraft control technology, and in particular to an unmanned swarm combat decision-making method and device. Background Art

[0002] With the development of cluster control technology, cluster combat has become one of the important combat forms in future wars. In the game confrontation scenario with clusters as combat units, how to make decisions based on the individual dynamics in the enemy cluster is the core issue. The famous "OODA" cycle divides a game confrontation link into four steps: perception, cognition, decision-making, and execution. According to this cycle, cluster game confrontation can be simplified into two major links: perception and cognition of enemy cluster targets and decision-making and execution of one's own cluster resource scheduling. However, cluster game confrontation scenarios face difficulties such as rapidly changing combat situations and huge data dimensions. How to quickly perceive the behavior of enemy cluster targets and timely schedule one's own cluster resources is still difficult to solve.

[0003] In previous research in the field of cluster game confrontation, researchers have designed many algorithms for the behavior prediction of enemy cluster targets and the target allocation of one's own cluster resources. The target behavior prediction algorithms under unknown dynamic models include behavior prediction algorithms based on hidden Markov chains and behavior prediction algorithms based on deep learning. However, the former ignores the model information of the enemy target, and the algorithm is difficult to adapt to the complex real environment, while the latter has high algorithm complexity and poor real-time performance. The scheduling and allocation algorithms for one's own cluster resources include: allocation algorithms based on value functions, allocation algorithms based on heuristic algorithms, etc. The former is difficult to solve under complex combat constraints, and the latter has the disadvantage of low exploration efficiency.

[0004] Therefore, it is particularly important to build a set of enemy behavior prediction and friendly target allocation algorithms involving the entire process of cluster game confrontation. Summary of the invention

[0005] The present application provides an unmanned cluster combat decision-making method and device to solve the above-mentioned technical problems in the prior art.

[0006] According to a first aspect of the present application, there is provided an unmanned cluster combat decision-making method, comprising:

[0007] Obtain the current status information of multiple enemy target individuals and own resource information in cluster combat;

[0008] Based on the current state information of the multiple enemy target individuals, using an interactive multiple model filtering algorithm to estimate the state information of the multiple enemy target individuals in the next time period;

[0009] Inputting the state information of the multiple enemy target individuals in the next time period and the own resource information into the constructed multi-objective optimization model under the multi-constraint conditions;

[0010] Solving the multi-objective optimization model under the multi-constraint conditions based on a multi-objective optimization algorithm to obtain output target allocation information;

[0011] Generate friendly attack decisions based on the target allocation information in different decision spaces.

[0012] According to a second aspect of the present application, an unmanned cluster combat decision-making device is provided, comprising:

[0013] The first unit is used to obtain the current status information of multiple enemy target individuals and the own resource information in the cluster battle;

[0014] The second unit is used to estimate the state information of the multiple enemy target individuals in the next time period by using an interactive multiple model filtering algorithm based on the current state information of the multiple enemy target individuals;

[0015] The third unit is used to input the state information of the multiple enemy target individuals in the next time period and the own resource information into the multi-objective optimization model under the constructed multi-constraint conditions;

[0016] A fourth unit is used to solve the multi-objective optimization model under the multi-constraint conditions based on a multi-objective optimization algorithm to obtain output target allocation information;

[0017] The fifth unit is used to generate a friendly attack decision according to the target allocation information in different decision spaces.

[0018] According to a third aspect of the present application, an electronic device is provided, comprising: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the steps of the method described in any embodiment are implemented.

[0019] According to a fourth aspect of the present application, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the steps of the method described in any embodiment are implemented.

[0020] In summary, the unmanned swarm combat decision-making method and device provided in this application have at least the following beneficial effects:

[0021] By acquiring the current state information of multiple enemy target individuals and own resource information in cluster combat, based on the current state information of the multiple enemy target individuals, an interactive multi-model filtering algorithm is used to estimate the state information of the multiple enemy target individuals in the next time period, thereby realizing the behavior prediction of the enemy target individuals based on the information observed in the game confrontation instead of relying on the prior information before the battle; by inputting the state information of the multiple enemy target individuals in the next time period and the own resource information into the constructed multi-objective optimization model under multi-constraint conditions; solving the multi-objective optimization model under multi-constraint conditions based on the multi-objective optimization algorithm, and obtaining the output target allocation information; generating the own strike decision according to the target allocation information in different decision spaces, thereby realizing the strike decision obtained by comprehensively considering various constraints in the game confrontation, and being able to characterize the environmental characteristics of the actual game confrontation, and therefore being more suitable for the actual confrontation scenario. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below in conjunction with the accompanying drawings. It is obvious that the drawings described below are some implementation methods of the present application. For those skilled in the art, other drawings or solutions can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 This is a flow chart of an unmanned cluster battle decision-making method in one embodiment of the present application;

[0024] Figure 2 A logical schematic diagram of estimating the state information of an enemy target individual in the next time period based on an interactive multi-model filtering algorithm in one embodiment of the present application;

[0025] Figure 3 A cluster game confrontation decision rule library construction process in one embodiment of the present application;

[0026] Figure 4 A schematic diagram of the predicted trajectory of an enemy target individual behavior in a simulation test in one embodiment of the present application;

[0027] Figure 5 This is a graph of error changes during the prediction of enemy target individual behavior in a simulation test in one embodiment of the present application;

[0028] Figure 6 This is a schematic diagram of the predicted probability changes of each kinematic reference model in a simulation test in one embodiment of the present application;

[0029] Figure 7 It is a distribution diagram of enemy individuals in the target resource allocation link in the simulation test in one embodiment of the present application;

[0030] Figure 8 This is a schematic diagram of the target resource attack action area in a simulation test in one embodiment of the present application;

[0031] Fig. 9 This is a graph showing changes in the loss of the optimization algorithm in a simulation test in one embodiment of the present application;

[0032] Fig.10 This is a structural block diagram of an unmanned cluster battle decision-making device in one embodiment of the present application;

[0033] Fig.11 This is a structural block diagram of an electronic device in one embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to make the above and other features and advantages of the present application more clear, the present application is further described below in conjunction with the accompanying drawings. It should be understood that the specific embodiments given herein are for the purpose of explaining to those skilled in the art and are only exemplary and not restrictive.

[0035] In the following description, many specific details are set forth to provide a thorough understanding of the present application. However, it is apparent to those skilled in the art that specific details need not be adopted to practice the present application. In other cases, well-known steps or operations are not described in detail to avoid blurring the present application.

[0036] In the description of this specification, if the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" appear, it means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0037] Previous research on cluster game confrontation environments often only focused on decision-making tasks, often simplifying the environment or assuming that the battlefield situation remains unchanged at the moment of decision-making, without considering the complexity and rapid changeability of the battlefield environment. Therefore, previous algorithms are difficult to apply to real cluster game confrontation scenarios.

[0038] This application will comprehensively consider many constraints in real combat scenarios, comprehensively predict the behavior of enemy cluster targets and target allocation of one's own cluster resources, and build an overall solution design covering the entire process of cluster game confrontation. This application is based on the premise that the enemy model is unknown in the real environment, does not rely on any pre-war prior information, and does not simplify many constraints in the battlefield environment. First, a behavior prediction algorithm based on an interactive multi-model filtering algorithm is designed to adaptively predict the enemy's actual dynamic model based on the enemy's past dynamic information, and comprehensively predict the enemy's dynamic information in the future by weighted probability of each model; secondly, a target allocation algorithm based on a multi-objective optimization algorithm under multiple constraints is designed to obtain a target allocation solution that meets all constraints under the actual battlefield situation. This application can realize the generation of a full-process solution for enemy target behavior prediction and one's own resource target allocation without relying on pre-war prior information and without simplifying battlefield environment constraints, and has strong practicality.

[0039] Reference Figure 1 The flowchart of the unmanned cluster battle decision-making method according to an embodiment of the present application is shown, and the method includes the following steps:

[0040] S100, obtaining current status information of multiple enemy target individuals and own resource information in a cluster battle;

[0041] S110, based on the current state information of the multiple enemy target individuals, using an interactive multiple model filtering algorithm to estimate the state information of the multiple enemy target individuals in the next time period;

[0042] S120, inputting the state information of the multiple enemy target individuals in the next time period and the own resource information into the constructed multi-objective optimization model under multi-constraint conditions;

[0043] S130, solving the multi-objective optimization model under the multi-constraint conditions based on a multi-objective optimization algorithm to obtain output target allocation information;

[0044] S140, generating a friendly attack decision according to the target allocation information in different decision spaces.

[0045] This application can use the interactive multi-model filtering algorithm to estimate the state information of the multiple enemy target individuals in the next time period based on the current state information of the multiple enemy target individuals, and input the state information of the multiple enemy target individuals in the next time period and the own resource information into the multi-objective optimization model under the constructed multi-constraint conditions, so as to obtain a target allocation scheme that matches the multi-constraint environment of the actual battlefield situation. This application can realize the generation of the whole process scheme of enemy target behavior prediction and own resource target allocation without relying on pre-war prior information and simplifying battlefield environment constraints.

[0046] To facilitate understanding of this solution, the above steps are further described in detail below.

[0047] The current status information of the multiple enemy target individuals in step S100 includes: position, speed and acceleration. The own resource information includes: the use of own cluster resources.

[0048] In the actual cluster game confrontation environment, the enemy's kinematic system model cannot be known in advance in most scenarios. Considering this situation, the embodiment of the present application adopts an interactive multi-model filtering algorithm to estimate the enemy target motion state.

[0049] The interactive multi-model filtering algorithm is based on Bayesian network theory and the first-order Markov process hypothesis, and uses multiple Kalman filters for parallel processing. The algorithm is mainly based on the principle of adaptive transfer between preset states of Bayesian theory: at any time, it is assumed that the transfer between models obeys the Markov chain, and the Kalman filter results of each group of kinematic models are estimated in parallel. The weight coefficient and model transfer probability are determined according to the degree of deviation of the estimation of each model, and finally the results of each model are weighted and calculated to obtain the optimal estimated state at the current moment. This application is expanded based on the results of each model (kinematic reference model), and the weighted results are used as the trajectory prediction scheme for the enemy target individual in the future period of time, which can realize the trajectory prediction of the enemy target individual without prior knowledge.

[0050] In one embodiment, the logic diagram of estimating the state information of the multiple enemy target individuals in the next time period by using the interactive multiple model filtering algorithm based on the current state information of the multiple enemy target individuals in step S110 is as follows: Figure 2 As shown in, including:

[0051] First, for each enemy target individual, the current state information (corresponding to Figure 2 The motion situation of the enemy cluster target in the target) is input into the corresponding multiple kinematic reference models preset for the enemy target individuals (corresponding to Figure 2 That is, the embodiment of the present application first presets multiple kinematic reference models for the enemy target individual. Assume that the state vector of the enemy target individual in a certain dimension satisfies X i =[p,v,a] T , which are used to describe the position p, velocity v and acceleration a in this direction respectively. Assume that the enemy target kinematic system model satisfies X i (t+1)=F i X i (t), where F iis the state transfer matrix. ΔT is defined as the sampling interval, ω is the angular velocity, and the kinematic reference model set for the enemy target individual without prior knowledge in this application includes: uniform motion model, uniform acceleration motion model and constant rate turning motion model, that is, Figure 2 The preset kinematic model set shown in includes a uniform motion model, a uniform acceleration motion model, and a constant rate turning motion model. The state transfer matrix corresponding to each model is shown below:

[0052] The state transfer matrix F of the uniform motion model i CV for:

[0053]

[0054] The state transfer matrix F of the uniform acceleration motion model i CA for:

[0055]

[0056] State transfer matrix F of constant rate turning motion model i CT for:

[0057]

[0058] The acquired current state information of the enemy target individual is input into the corresponding kinematic reference model.

[0059] Secondly, the predicted motion state vectors respectively output by the multiple kinematic reference models are obtained.

[0060] That is, a filtering operation is performed on any single model in the preset kinematic model set, including the preset filtering algorithm initialization and data normalization, and each kinematic reference model outputs a corresponding predicted motion state vector.

[0061] Afterwards, the weighted coefficients of the predicted motion state vectors output by the multiple kinematic reference models are determined, thereby obtaining a mixed state estimate and a mixed covariance estimate of the enemy target individual as state information of the enemy target individual in the next time period.

[0062] For each set of preset kinematic reference models, the Kalman filter algorithm can be used to estimate the state, and each model interacts with each other to solve the weighted coefficient of each model based on the estimated deviation.

[0063] Assume the probability transfer matrix between different kinematic reference models is where p nn is the probability of transferring from model n to model n. The initial probability of each model is initialized to meet Ui =[u1,u2,…,u n ], taking model j as an example, solve the normalization constant to satisfy The transfer mixing probability between models i and j satisfies After solving the probabilities of all models, we can weight the mixed state estimate of the target and mixed covariance estimation

[0064] After each kinematic reference model uses a filtering algorithm to implement one-step prediction and parallel filtering, the weighted coefficients of each kinematic reference model are updated using a likelihood function, that is, a probability update, where the likelihood function satisfies:

[0065]

[0066] Among them, j To measure the deviation between the update and the further prediction, Τ j (t) is the weighted matrix in the measurement update process. The probability of each model can be updated as:

[0067]

[0068] Among them, the total normalization coefficient

[0069] Based on the above probability update of the model, the probability of each model can be used to achieve weighted prediction of the state of the enemy target individual in the next time period And the new covariance matrix P(t|t) satisfies: n=3, and the state after the weighting of the three models is calculated That is, the behavior of the enemy target individual in the future is obtained through weighted prediction, as follows

[0070]

[0071] It should be noted that the interactive multi-model filtering algorithm in this application processes multiple pre-set estimation system models, and the parallel filtering step is not limited to one filtering method. Therefore, the Kalman filter can be replaced by other filtering methods, or different filtering methods can be applied in a system for multi-model interaction, so as to better estimate the motion state of the enemy target individual.

[0072] The above scheme realizes the prediction of the motion state of enemy target individuals without relying on prior knowledge. Another goal of cluster game confrontation is to achieve the allocation of one's own resources. Due to the complex environment of cluster game confrontation scenarios and the large scale of individuals involved, the knowledge strategy set under a single evaluation criterion is difficult to describe the idealized solution for cluster resource allocation. For cluster game confrontation scenarios, Pareto optimality refers to an ideal state of one's own resource allocation. It is constructed based on multiple evaluation criteria and can more comprehensively describe the resource allocation of complex environments. Assuming the allocation problem of inherent disposable resources and targets to be attacked, assuming two objective functions, for a certain solution, if no other solution can be found in the variable space that can be better than this solution in all aspects, then this solution is the Pareto optimal solution. Therefore, for multi-objective optimization problems, there are generally multiple Pareto optimal solutions, which constitute the Pareto optimal frontier.

[0073] The constraints of target allocation in cluster game confrontation can be divided into two categories: environmental constraints and model constraints. Based on this, this application comprehensively considers the actual combat space situation such as safety avoidance, target damage threshold and advantage assignment, threat index, etc., takes the utilization of one's own resources and the damage of the enemy's incoming targets as the objective function, and establishes a multi-objective optimization model under multi-constraint conditions under the target cluster, that is, a multi-constraint multi-objective weapon-target allocation model, the purpose is to solve the Pareto solution set in the constrained feasible region with the objective function as a variable, as a candidate set of game confrontation expert knowledge (that is, a game confrontation decision rule library) for use in the decision-making link. This application constructs a variety of evaluation criteria for evaluating the cluster game confrontation environment. Based on the Pareto optimality theory, the generation of the game confrontation decision rule library is transformed into an optimization problem under multi-objective constraints, and the solution set under the joint action of different evaluation criteria is solved to generate a game confrontation expert knowledge strategy set.

[0074] In the embodiment of the present application, before executing step S120, the method further includes: a step of constructing a multi-objective optimization model under the multi-constraint conditions, specifically including:

[0075] The two mutually coupled objective functions of the global combat space situation function and the number of resources used by one's own side are selected for simultaneous optimization, and the problem is modeled as a multi-objective optimization problem. The multi-objective optimization problem can be defined as:

[0076] minV(x)=(v1(x),v2(x),…,v n (x)

[0077] stx∈Π

[0078] Among them, x is the decision-related variable, Π is the decision space, V:Π→R mThis application takes the expected survival of the global goal and the consumption of one's own individual resources as the optimization goal. Assuming that one's own individual resources are n and the enemy's targets to be attacked are m, this type of weapon-target allocation problem can be modeled as follows: the resource set within one's own cluster W = {w1, w2, ..., w n}, enemy target set T = {t1, t2, …, t m}, the target position set P = {p1,p2,…,p m}, target individual value weight set Δ=[Δ1,Δ2,…,Δ m ]; the corresponding decision variable is: the resource usage of one’s own cluster X=[x1,x2,…,x n ], the set of attack points of one’s own weapons Z = {z1,z2,…,z n}; then the multi-objective optimization function under multiple constraints can be modeled as follows:

[0079]

[0080] Among them, X represents the resource usage set of our own cluster, Z represents the set of our own weapon strike points, f is the normalized environmental evaluation index, f = T1 + T2 + T3 + ..., T1, T2 and T3 represent different threat functions, i represents the number of our own resources, j represents the number of enemy target individuals, x i Indicates the resources of one's own cluster. v1 measures the survival probability of the global enemy target, and v2 measures the usage of one's own cluster resources.

[0081] In one embodiment, the threat function includes: an angle threat function, a speed threat function, a distance threat function and a height threat function, as shown below:

[0082] T angle =(Δe) / 360°

[0083]

[0084] Among them, T angle is the angle threat, Δe is the angle deviation; T vel For speed threat, v m is the enemy target speed measure taking into account the own speed, v t is the enemy target speed; T dis is the distance threat, H is the risk-free distance, L is the high-risk distance; T high High threat, h d is the height difference between the target and the player, X represents the distance, X h Indicates the upper limit of the distance threshold when the function is segmented, X L Indicates the lower limit of the distance threshold when dividing the function into segments.

[0085] The multi-objective optimization model under multi-constraint conditions described in the embodiment of the present application is characterized in that, for the cluster game confrontation environment, the decision matrix X and the objective function V(X) are both discrete variables, and solving their theoretical optimal solutions is only limited by the scale of discrete variables; while the model's action point set Z and objective function f are continuous functions.

[0086] Considering the safety avoidance constraints on the own protection area, the enemy's special dynamic target damage threshold constraints, the superior weapon preference assignment constraints, the confrontation maneuver situation constraints and other factors in the game confrontation link, in the multi-objective optimization function under the multi-constraint conditions of cluster game confrontation, the specific expert knowledge strategy rules (confrontation decision rules) are defined as follows:

[0087] ||X||≤n,x i ∈{0,1}

[0088]

[0089] a ij R i >dis;

[0090] Where ||X||≤n,x i ∈{0,1} is a quantity constraint, and the resource scheduling of one's own side must be an integer; As a survival probability constraint, for some key target nodes in the enemy's incoming targets, such as enemy units that execute offensive strategies, our decision-making system should ensure that such high-threat targets have a lower survival probability in the game confrontation link, ρ j is the expected survival probability; a ij R i >dis is the weapon preference constraint, Ri is the attack range, dis is the distance between the enemy target and the friendly target, i represents the friendly target, and j represents the enemy target; when facing different terrains, weapon types, or heterogeneous target types, prior knowledge shows the degree of match between friendly weapons and incoming enemy targets, or the decision maker assigns a certain type of weapon to a certain enemy target, a ij ∈[0,1] represents whether there is a dominant match.

[0091] Step S130 is a process of solving a multi-objective optimization model under multiple constraints.

[0092] This application constructs a data screening algorithm based on the coding evolution mechanism according to the characteristics of cluster game confrontation scenarios and the properties of multi-objective optimization models under multiple constraints. Population individuals represent weapon usage and strike coordinates, using continuous real number coding Z = {z1, z2, ..., z w}, component z iIndicates the position of the action point of one's own weapon resource i in the battlefield environment. Taking the three-dimensional combat space as an example, When weapon i is not used, Therefore, the individual initialization algorithm is partially set up as follows:

[0093] Input: Population size pop; Own group size n; Enemy target size m; Battlefield range B: Combined range of own weapon strike point set Z

[0094] Output: Initialize population data P = {SP1, SP2, ..., SP m}, where SP i For subgroup i

[0095]

[0096]

[0097] Return initialize population information

[0098] Due to the existence of multiple constraints in the multi-objective optimization model, a large number of individuals that violate the constraints will be generated during population evolution, that is, infeasible solutions. The degree of violation of infeasible solutions is used as the basis for dominance ranking, satisfying:

[0099] Individual i constrains individual j if and only if any of the following conditions is true:

[0100] 1. Individual i and individual j are both feasible solutions, and individual i dominates individual j.

[0101] 2. Individual i is a feasible solution, and individual j is an infeasible solution.

[0102] 3. Individual i and individual j are both infeasible solutions, but individual i has a smaller constraint violation value.

[0103] Reference Figure 2 As shown in the figure, the cluster game confrontation decision rule library construction process is completed based on the multi-objective optimization algorithm under the multi-constraint conditions of NSGA-II. Under this algorithm flow, the cluster game confrontation decision rule library (corresponding to Figure 2 The cluster game confrontation expert knowledge strategy rule base) realizes the process transformation from raw data to the allocation and attack strategy corresponding to different target scales.

[0104] The main features of the algorithm in this application are: 1. Establish a cluster target allocation problem model under multiple constraints, and the evaluation indicator is to minimize the strike loss; 2. Use a multi-objective optimization algorithm to solve the problem and obtain alternative solutions to assist decision-making; 3. Use a violation value constraint algorithm to ensure the completeness of the Pareto optimal solution; 4. Use a cooperative evolution mechanism to enhance convergence.

[0105] Step S140 is to generate actual strike decisions according to the target allocation schemes under different decision spaces. According to the input parameters, the obtained data table can be determined to be a table of pop*(n+3*n+2+1+1+1+1), where the number of rows pop is the preset population number, corresponding to the dominant pop schemes screened out after the number of iterations;

[0106]

[0107] The meanings of the column numbers are as follows:

[0108] Column n represents the weapon usage, 1 indicates use and 0 indicates no use;

[0109] 3*n columns are weapon attack coordinates, in the format of [x1, y1, z1,…, x n ,y n ,z n ];

[0110] Column 2 is the optimization objective function value, v1 is used to measure the survival probability of the global enemy target, and v2 is used to measure the resource usage of the own cluster;

[0111] 1 is a feasible solution judgment, 1 is a solution that satisfies multiple constraints, and 0 is a solution that violates the constraints;

[0112] Column 1 is the violation value constraint value. This column is not 0 only for infeasible solutions, indicating the degree of violation of multiple constraints.

[0113] Column 1 is the ranking value, which represents the solution with the highest dominance in the Pareto optimal solution set. Each number corresponds to at most n groups of results. Those that do not meet the constraints will be screened out and not recorded;

[0114] Column 1 is the crowding degree, which represents the data divergence under the same dominance level.

[0115] According to the output data results, the corresponding plans for different quantities of weapon resources under each decision space are selected as the target allocation plan for the game confrontation link to support the game confrontation decision.

[0116] In order to prove the practicability and effectiveness of the method described in this application, the method described in this application was used to conduct simulation tests on target behavior prediction and weapon target allocation.

[0117] First is the target behavior prediction part. Suppose the initial state of an enemy target in a three-dimensional battlefield environment is x0 = [20, 20, 0, 20, 20, 0, 20, 0] T, the node will move at a constant speed within t = 1-10s, move at a constant acceleration within t = 10-40s, move at a constant speed within t = 40-60s, turn at a constant rate within t = 60-100s, and move at a constant speed within t = 100-120s. The sampling time is ΔT = 0.1s. There is noise interference from the outside world. The schematic diagram of the predicted trajectory of the enemy target individual behavior in the simulation test is as follows Figure 4 As shown in the figure, the error change in the process of predicting the enemy target individual behavior is as follows Figure 5 As shown in , the schematic diagram of the predicted probability change of each kinematic reference model is as follows Figure 6 As shown in the above figure, it can be seen that the unmanned cluster combat decision-making method described in this application can achieve independent of pre-war prior information and effectively predict the movement state of enemy target individuals.

[0118] Regarding the resource allocation part of our own cluster (weapon target allocation), taking the result under a certain rule of n=5,m=20 as an example, the enemy individual distribution diagram in the target resource allocation link is as follows: Figure 7 As shown in the figure, the target resource strike area diagram is as follows Figure 8 As shown in , the loss change diagram of the optimization algorithm is as follows Fig. 9 As shown in the figures, it can be explained that the unmanned cluster combat decision-making method described in the embodiment of the present application can generate effective attack decisions without simplifying the cluster game confrontation problem.

[0119] The present application also provides an unmanned cluster battle decision-making device. The schematic diagram of the device structure is as follows: Fig.10 As shown in , the device comprises:

[0120] The first unit 101 is used to obtain current status information of multiple enemy target individuals and own resource information in a cluster battle;

[0121] The second unit 102 is used to estimate the state information of the multiple enemy target individuals in the next time period by using an interactive multiple model filtering algorithm based on the current state information of the multiple enemy target individuals;

[0122] The third unit 103 is used to input the state information of the multiple enemy target individuals in the next time period and the own resource information into the multi-objective optimization model under the constructed multi-constraint conditions;

[0123] The fourth unit 104 is used to solve the multi-objective optimization model under the multi-constraint conditions based on the multi-objective optimization algorithm to obtain output target allocation information;

[0124] The fifth unit 105 is used to generate a friendly attack decision according to the target allocation information in different decision spaces.

[0125] It should be understood that the specific features, operations and details described hereinabove with respect to the method of the present application may also be similarly applied to the device and system of the present application, or vice versa. In addition, each step of the method of the present application described above may be performed by a corresponding component or unit of the device or system of the present application.

[0126] It should be understood that each module / unit of the device of the present application can be implemented in whole or in part by software, hardware, firmware or a combination thereof. Each module / unit can be embedded in the processor of the electronic device or independent of the processor in the form of hardware or firmware, or can be stored in the memory of the electronic device in the form of software for the processor to call to perform the operation of each module / unit. Each module / unit can be implemented as an independent component or module, or two or more modules / units can be implemented as a single component or module.

[0127] like Fig.11 As shown, the present application provides an electronic device 400, which includes a processor 401 and a memory 402 storing computer program instructions. When the processor 401 executes the computer program instructions, each step of the above-mentioned data detection method is implemented. The electronic device 400 can be a server, a terminal, or any other electronic device with necessary computing and / or processing capabilities in a broad sense.

[0128] In one embodiment, the electronic device 400 may include a processor, a memory, a network interface, a communication interface, etc. connected through a system bus. The processor of the electronic device 400 may be used to provide necessary computing, processing and / or control capabilities. The memory of the electronic device 400 may include a non-volatile storage medium and an internal memory. The non-volatile storage medium may store an operating system, a computer program, etc. The internal memory may provide an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface and the communication interface of the electronic device 400 may be used to connect and communicate with external devices through a network. When the computer program is executed by the processor, the steps of the method of the present application are executed.

[0129] The present application provides a computer-readable storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the above-mentioned unmanned cluster confrontation decision-making method is implemented.

[0130] Those skilled in the art will appreciate that the method steps of the present application can be completed by instructing related hardware such as the electronic device 400 or a processor through a computer program, and the computer program can be stored in a non-temporary computer-readable storage medium, and the computer program causes the steps of the present application to be executed when it is executed. Depending on the circumstances, any reference to memory, storage, or other media herein may include non-volatile or volatile memory. Examples of non-volatile memory include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk, etc. Examples of volatile memory include random access memory (RAM), external cache memory, etc.

[0131] The various technical features described above can be combined arbitrarily. Although all possible combinations of these technical features are not described, any combination of these technical features should be considered to be covered by this specification as long as there is no contradiction in such combination.

[0132] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. An unmanned swarm combat decision-making method, characterized in that: include: Obtain the current status information of multiple enemy target individuals and own resource information in cluster combat; Based on the current state information of the multiple enemy target individuals, using an interactive multiple model filtering algorithm to estimate the state information of the multiple enemy target individuals in the next time period; Inputting the state information of the multiple enemy target individuals in the next time period and the own resource information into the constructed multi-objective optimization model under the multi-constraint conditions; Solving the multi-objective optimization model under the multi-constraint conditions based on a multi-objective optimization algorithm to obtain output target allocation information; Generate your own strike decision based on the target allocation information in different decision spaces.

2. The method according to claim 1, characterized in that The method of estimating the state information of the multiple enemy target individuals in the next time period by using an interactive multiple model filtering algorithm based on the current state information of the multiple enemy target individuals includes: For each enemy target individual, inputting current state information into a plurality of kinematic reference models preset for the corresponding enemy target individual; Acquire predicted motion state vectors respectively output by the plurality of kinematic reference models; The weighted coefficients of the predicted motion state vectors output by the multiple kinematic reference models are determined, thereby obtaining a mixed state estimate and a mixed covariance estimate of the enemy target individual as state information of the enemy target individual in the next time period.

3. The method according to claim 2, characterized in that After obtaining the mixed state estimation and mixed covariance estimation of the enemy target individual, the method further includes: The weighting coefficients of the multiple kinematic reference models are updated using a likelihood function.

4. The method according to claim 2, characterized in that: The kinematic reference model includes: Uniform motion model, uniform acceleration motion model and constant rate turning motion model.

5. The method according to claim 1, characterized in that The method further comprises: a step of constructing a multi-objective optimization model under the multi-constraint conditions, specifically comprising: The multi-objective optimization model under the multi-constraint conditions is constructed by taking the environmental constraints and the model constraints as constraints as follows Among them, X represents the resource usage set of our own cluster, Z represents the set of our own weapon strike points, f is the normalized environmental evaluation index, f = T1 + T2 + T3 + ..., T1, T2 and T3 represent different threat functions, i represents the number of our own resources, j represents the number of enemy target individuals, x i Represents the own cluster resources, v1 measures the survival probability of the global enemy target, and v2 measures the usage of the own cluster resources.

6. The method according to claim 5, characterized in that The threat functions include: Angle threat function, speed threat function, distance threat function and height threat function.

7. An unmanned cluster combat decision-making device, characterized in that: include: The first unit is used to obtain the current status information of multiple enemy target individuals and the own resource information in the cluster battle; The second unit is used to estimate the state information of the multiple enemy target individuals in the next time period by using an interactive multiple model filtering algorithm based on the current state information of the multiple enemy target individuals; The third unit is used to input the state information of the multiple enemy target individuals in the next time period and the own resource information into the multi-objective optimization model under the constructed multi-constraint conditions; A fourth unit is used to solve the multi-objective optimization model under the multi-constraint conditions based on a multi-objective optimization algorithm to obtain output target allocation information; The fifth unit is used to generate a friendly attack decision according to the target allocation information in different decision spaces.

8. The device according to claim 7, characterized in that The device also includes: The sixth unit is used to construct the multi-objective optimization model under the multi-constraint conditions, specifically including: taking the environmental constraints and the model constraints as constraints to construct the multi-objective optimization model under the multi-constraint conditions as follows Among them, X represents the resource usage set of our own cluster, Z represents the set of our own weapon strike points, f is the normalized environmental evaluation index, f = T1 + T2 + T3 + ..., T1, T2 and T3 represent different threat functions, i represents the number of our own resources, j represents the number of enemy target individuals, x i Represents the own cluster resources, v1 measures the survival probability of the global enemy target, and v2 measures the usage of the own cluster resources.

9. An electronic device, characterized in that: include: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the steps of the method according to any one of claims 1 to 6.

Images