Evolutionary game-based combat simulation experiment design method, device and equipment

Abstract

An evolutionary game-based combat simulation experiment design optimization method is provided, including: initializing a blue party population and a red party population; and iteratively evolving the red party population, including the following steps: evaluating the fitness of the current red party population against the blue party population using a fitness evaluation machine learning model to obtain a fitness evaluation result, wherein the fitness evaluation machine learning model is trained such that the fitness indicates the degree of superiority or inferiority of the red combat experiment scheme to the blue combat experiment scheme; in response to the fitness evaluation result being less than a preset threshold, performing an evolutionary operation on the current red party population to evolve the current red party population, and repeating the step of evaluating the fitness of the current red party population against the blue party population using the fitness evaluation machine learning model using the evolved current red party population; and in response to the fitness evaluation result being greater than the preset threshold, taking the red combat experiment scheme represented by the current red party population as the target red combat experiment scheme.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to a method, apparatus, computer equipment, medium and program product for designing combat simulation experiments based on evolutionary game theory. Background Technology

[0002] Operational experimental research faces the following difficulties and challenges: First, it needs to address the uncertainty of the adversary. Due to the fog of war, the key experimental factors and levels of the enemy are generally difficult to determine in experimental scenarios. Their values ​​can only be assumed and speculated to be within a certain range. Therefore, the uncertainty of the adversary needs to be considered in experimental design, requiring our operational experimental schemes to have good robustness and generalization ability. Second, it needs to resolve the contradiction between the long duration of simulation experiments and the large experimental search space. Operational simulation experiments at the campaign or joint tactical level require several hours or days of actual physical time, and even a single simulation experiment requires several minutes, tens of minutes, or even longer. In such cases, the time required to completely search the vast experimental space is unacceptable.

[0003] In related technologies, many simulation experiment methods have been proposed to reduce the number of optimization attempts required for experimental schemes in combat simulation experiments. For example, methods such as orthogonal design, Latin hypercube design, and random sampling, which can be used for any simulation system, have been proposed for combat simulation experiment design. However, these methods do not consider the characteristics of the simulation system itself, are unidirectional and one-time, generally require a large number of experiments, and have poor optimization effects on combat simulation experiment design. Summary of the Invention

[0004] This disclosure provides a method, apparatus, computer equipment, medium, and program product for identifying alias entities.

[0005] According to one aspect of this disclosure, a method for designing and optimizing combat simulation experiments based on evolutionary game theory is provided, comprising: initializing a blue team population, wherein the blue team population represents a blue team combat experiment scheme in the combat simulation experiment, wherein the blue team population includes multiple blue team combat experiment factor individuals, each blue team combat experiment factor individual being associated with combat effectiveness in the combat simulation experiment; initializing a red team population, wherein the red team population represents a red team combat experiment scheme in the combat simulation experiment, wherein the red team population includes multiple red team combat experiment factor individuals, each red team combat experiment factor individual being associated with combat effectiveness in the combat simulation experiment; and iteratively evolving the red team population, comprising the following steps: using a fitness evaluation machine learning model. The fitness of the current red team population against the blue team population is evaluated to obtain a fitness evaluation result, wherein the fitness evaluation machine learning model is trained such that the fitness indicates the superiority or inferiority of the red team's combat experimental scheme against the blue team's combat experimental scheme; in response to the fitness evaluation result being less than a preset threshold, an evolutionary operation is performed on the current red team population to evolve the current red team population, and the step of evaluating the fitness of the current red team population against the blue team population using the fitness evaluation machine learning model is repeated using the evolved current red team population; and in response to the fitness evaluation result being greater than the preset threshold, the red team's combat experimental scheme represented by the current red team population is taken as the target red team's combat experimental scheme.

[0006] According to another aspect of this disclosure, a combat simulation experiment design optimization device based on evolutionary game theory is provided, comprising: a first initialization unit for initializing a blue team population, wherein the blue team population represents a blue team combat experiment scheme in the combat simulation experiment, wherein the blue team population includes multiple blue team combat experiment factor individuals, each blue team combat experiment factor individual being associated with combat effectiveness in the combat simulation experiment; a second initialization unit for initializing a red team population, wherein the red team population represents a red team combat experiment scheme in the combat simulation experiment, wherein the red team population includes multiple red team combat experiment factor individuals, each red team combat experiment factor individual being associated with combat effectiveness in the combat simulation experiment; and a population evolution unit for iteratively evolving the red team population, including the following steps. The following steps are performed: 1) Evaluate the fitness of the current red team population against the blue team population using a fitness evaluation machine learning model to obtain a fitness evaluation result. The fitness evaluation machine learning model is trained such that the fitness indicates the superiority or inferiority of the red team's combat experimental plan against the blue team's combat experimental plan. 2) In response to the fitness evaluation result being less than a preset threshold, perform an evolutionary operation on the current red team population to evolve it, and repeat the step of evaluating the fitness of the current red team population against the blue team population using the evolved current red team population. 3) In response to the fitness evaluation result being greater than a preset threshold, use the red team's combat experimental plan represented by the current red team population as the target red team's combat experimental plan.

[0007] According to another aspect of this disclosure, a computer device is provided, comprising: at least one processor; and a memory having a computer program stored thereon, wherein the computer program, when executed by the processor, causes the processor to perform the method described above.

[0008] According to another aspect of this disclosure, a computer-readable storage medium is provided that stores a computer program thereon, which, when executed by a processor, causes the processor to perform the methods described above.

[0009] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, causes the processor to perform the methods described above.

[0010] According to one or more embodiments of this disclosure, by introducing an evolutionary game theory algorithm, the sample space of adversarial schemes for the blue side is first determined. Then, a proxy model is used to implement simulated game adversarial competition between the red and blue sides to evaluate fitness. This can eliminate experimental schemes for the red side with low fitness, and update and optimize some of the better combat schemes for the red side through evolutionary operations, thereby improving the fitness of the red side's combat experimental schemes. Based on this, the number of experiments can be effectively reduced, experimental efficiency can be improved, and the difficulties and challenges faced in the design of complex combat simulation experiments, such as uncertain opponents, long simulation experiment time, and large experimental search space, can be overcome, thus optimizing the combat simulation experiment design process.

[0011] These and other aspects of this disclosure will be apparent from the embodiments described below, and will be elucidated with reference to the embodiments described below. Attached Figure Description

[0012] Further details, features, and advantages of this disclosure are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which:

[0013] Figure 1 This is a schematic diagram illustrating an example system in which various methods described herein may be implemented according to exemplary embodiments;

[0014] Figure 2 This is a flowchart illustrating an evolutionary game-based combat simulation experiment design optimization method according to an exemplary embodiment;

[0015] Figure 3 This is a flowchart illustrating the fitness evaluation of an evolutionary game-based combat simulation experiment design optimization method according to an exemplary embodiment;

[0016] Figure 4 This is a schematic diagram illustrating the game evolution process of the evolutionary game-based combat simulation experiment design optimization method according to an exemplary embodiment;

[0017] Figure 5 This is a partial flowchart illustrating an evolutionary game-based combat simulation experiment design optimization method according to another exemplary embodiment;

[0018] Figure 6 This is a flowchart illustrating the training process of a fitness evaluation machine learning model according to an exemplary embodiment;

[0019] Figure 7 This is a schematic block diagram illustrating an evolutionary game-based combat simulation experiment design optimization device according to an exemplary embodiment;

[0020] Figure 8 This is a block diagram illustrating an exemplary computer device that can be applied to an exemplary embodiment. Detailed Implementation

[0021] In this disclosure, unless otherwise stated, the use of terms such as "first," "second," etc., to describe various elements is not intended to limit the positional, temporal, or importance relationships of these elements; such terms are merely used to distinguish one element from another. In some examples, the first element and the second element may refer to the same instance of that element, while in other cases, based on the context, they may refer to different instances.

[0022] The terminology used in the description of the various examples in this disclosure is for the purpose of describing particular examples only and is not intended to be limiting. Unless the context expressly indicates otherwise, an element may be one or more unless the number of elements is specifically limited. As used herein, the term "multiple" means two or more, and the term "based on" should be interpreted as "at least partially based on". Furthermore, the terms "and / or" and "at least one of..." cover any one of the listed items and all possible combinations thereof.

[0023] In related technologies, many simulation experiment methods have been proposed to reduce the number of optimization attempts required for experimental schemes in combat simulation experiments. For example, methods such as orthogonal design, Latin hypercube design, and random sampling, which can be used for any simulation system, have been proposed for combat simulation experiment design. However, these methods do not consider the characteristics of the simulation system itself, are unidirectional and one-time, generally require a large number of experiments, and have poor optimization effects on combat simulation experiment design.

[0024] Based on this, this disclosure proposes an evolutionary game theory-based optimization method for combat simulation experiment design. By introducing an evolutionary game theory algorithm, the sample space of adversarial schemes for the blue side is first determined. Then, a proxy model is used to implement simulated game adversarial competition between the red and blue sides to evaluate fitness. This method can eliminate red side experimental schemes with low fitness and update and optimize some of the better red side's combat schemes through evolutionary operations, thereby improving the fitness of the red side's combat experimental schemes. Based on this, the number of experiments can be effectively reduced, experimental efficiency can be improved, and the difficulties and challenges faced in designing complex combat simulation experiments, such as uncertain opponents, long simulation experiment time, and large experimental search space, can be overcome, thus optimizing the combat simulation experiment design process.

[0025] Exemplary embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0026] Figure 1 This is a schematic diagram illustrating an example system 100 in which various methods described herein may be implemented according to exemplary embodiments.

[0027] refer to Figure 1The system 100 includes a client device 110, a server 120, and a network 130 that communicatively couples the client device 110 and the server 120.

[0028] Client device 110 includes a display 114 and a client application (APP) 112 that can be displayed on the display 114. Client application 112 can be an application that needs to be downloaded and installed before running, or a lightweight application (liteapp). If client application 112 is an application that needs to be downloaded and installed before running, client application 112 can be pre-installed on client device 110 and activated. If client application 112 is a mini-app, user 102 can directly run client application 112 on client device 110 without installing it, by searching for client application 112 in the host application (e.g., by the name of client application 112) or scanning the graphic code of client application 112 (e.g., barcode, QR code, etc.). In some embodiments, client device 110 can be any type of mobile computing device, including mobile computers, mobile phones, wearable computing devices (e.g., smartwatches, head-mounted devices including smart glasses, etc.), or other types of mobile devices. In some embodiments, the client device 110 may alternatively be a fixed computer device, such as a desktop computer, server computer, or other type of fixed computer device.

[0029] Server 120 is typically a server deployed by an Internet Service Provider (ISP) or Internet Content Provider (ICP). Server 120 can represent a single server, a cluster of multiple servers, a distributed system, or a cloud server providing basic cloud services (such as cloud databases, cloud computing, cloud storage, and cloud communications). It will be understood that, although... Figure 1 The diagram shows that server 120 communicates with only one client device 110, but server 120 can provide background services to multiple client devices simultaneously.

[0030] Examples of network 130 include combinations of local area networks (LANs), wide area networks (WANs), personal area networks (PANs), and / or communication networks such as the Internet. Network 130 can be wired or wireless. In some embodiments, technologies and / or formats including Hypertext Markup Language (HTML), Extensible Markup Language (XML), etc., are used to process data exchanged through network 130. Furthermore, encryption technologies such as Secure Sockets Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), and Internet Protocol Security (IPsec) can be used to encrypt all or some of the links. In some embodiments, custom and / or dedicated data communication technologies can be used to replace or supplement the aforementioned data communication technologies.

[0031] For the purposes of this disclosure's embodiments, Figure 1 In the example, client application 112 can be a simulated combat experiment design program that can provide various functions based on combat simulation experiment design optimization, such as experiment scheme design, fitness calculation of experiment factors, evolutionary updates of experiment schemes, etc. Correspondingly, server 120 can be a server used with the combat simulation experiment design program. Server 120 can provide online combat simulation experiment design optimization services to client application 112 running on client device 110. Alternatively, local combat simulation experiment design optimization services can also be provided by client application 112 running on client device 110. As a non-limiting example, the method described below can be... Figure 1 The system 100 is running.

[0032] Figure 2 This is a flowchart illustrating an evolutionary game-based combat simulation experiment design optimization method 200 according to an exemplary embodiment. Method 200 can be implemented on a client device (e.g., Figure 1 The execution is performed at the client device 110 shown, that is, the execution entity of each step of method 200 can be... Figure 1 The client device 110 shown. In some embodiments, method 200 can be performed on a server (e.g., Figure 1 The method 200 is executed at server 120 (as shown in the figure). In some embodiments, the method 200 may be executed in combination by a client device (e.g., client device 110) and a server (e.g., server 120).

[0033] refer to Figure 2 The method for designing and optimizing combat simulation experiments based on evolutionary game theory includes the following steps.

[0034] Step 210: Initialize the Blue Team population. The Blue Team population represents the Blue Team combat experiment scheme in the combat simulation experiment. The Blue Team population includes multiple Blue Team combat experiment factor individuals, and each Blue Team combat experiment factor individual is associated with the combat effectiveness in the combat simulation experiment.

[0035] Step 220: Initialize the Red Team population. The Red Team population represents the Red Team combat experiment scheme in the combat simulation experiment. The Red Team population includes multiple Red Team combat experiment factor individuals, and each Red Team combat experiment factor individual is associated with the combat effectiveness in the combat simulation experiment.

[0036] In some embodiments, the experimental factors for both the red and blue sides and the range of factor levels for each experimental factor can be determined according to experimental requirements. The experimental factors are initialized as individuals, factor types, and factor levels as genes, thereby generating the blue and red populations suitable for the evolutionary algorithm.

[0037] In some embodiments, the experimental factor is the experimental variable, which may include, but is not limited to, elements related to combat effectiveness such as weapon strength, defensive capability, and attack frequency.

[0038] According to some embodiments, each of the multiple red team operational experimental factor individuals has a factor type and a factor level, each of the multiple blue team operational experimental factor individuals has a factor type and a factor level, and at least two blue team operational experimental factors with the same factor type have different factor levels.

[0039] In some embodiments, since the force elements of the Blue Force's combat system are relatively certain, but the factor levels and the confrontation process are uncertain, the Blue Force's combat experimental schemes can be treated as an independent population. Each individual corresponds to a factor type, and experimental factors of the same factor type have different factor levels. This enables the identification of the optimal combination of experimental factors and factor levels for the Red Force's combat experimental scheme against the Blue Force's combat experimental scheme, so that the target Red Force combat experimental scheme corresponding to the obtained Red Force population can have the best possible combat effectiveness against the Blue Force's combat experimental scheme.

[0040] In some embodiments, the factor level of each experimental factor can be used to indicate the magnitude or degree of the corresponding experimental variable. For example, the blue team's combat experiment scheme includes 5 experimental factors, and each experimental factor includes 3 factor levels: "strong", "medium", and "weak". Alternatively, the red team's combat experiment method includes 3 experimental factors, and the value range of the factor level of each experimental factor is [0,1], without any restriction.

[0041] In some embodiments, the evolutionary game-based combat simulation experiment design optimization of this disclosure is also applicable to other experimental schemes designed based on sequential experimental design methods. In one example, the method of this disclosure can be used to design and optimize a rice production experiment, where the experimental factors may include light intensity, and the factor level of light intensity may include 10,000 lux or 20,000 lux. In addition, experimental factors may also include humidity, carbon dioxide content, etc., without specific limitations.

[0042] Step 230 allows the red population to evolve iteratively.

[0043] In some embodiments, the Red team's experimental scheme is the main focus of the optimization study. Therefore, the Red team population is subjected to evolutionary iterations to explore the impact of experimental factor levels on combat effectiveness when playing against the Blue team population. Through repeated game interactions, the Red and Blue teams continuously learn and optimize the Red team population, forming a Red team population with a stable strategy and interests, in order to obtain a target Red team combat experimental scheme that meets the requirements.

[0044] In some embodiments, the evolution of the red population can be implemented based on a genetic algorithm that enhances elite preservation and a differential evolution algorithm.

[0045] Step 230 includes steps 2310 to 2330.

[0046] Step 2310: Use a fitness evaluation machine learning model to evaluate the fitness of the current red population against the blue population and obtain the fitness evaluation result. The fitness evaluation machine learning model is trained so that the fitness indicates the superiority or inferiority of the red team's combat experimental plan against the blue team's combat experimental plan.

[0047] In some embodiments, the fitness assessment results indicate the superiority or inferiority of the Red Team's combat experiment scheme relative to the Blue Team's combat experiment scheme. The higher the fitness, the better the corresponding Red Team combat experiment scheme for the current Red Team population.

[0048] Figure 3 A flowchart of fitness evaluation in step 2310 is shown.

[0049] refer to Figure 3 According to some embodiments, step 2310 includes the following steps.

[0050] Step 3311: Input the red team population and the blue team population into the fitness evaluation machine learning model to obtain the first fitness of each red team combat experiment factor individual in the current red team population relative to each blue team combat experiment factor individual in the blue team population, so as to obtain multiple first fitness values.

[0051] Step 3312: Calculate the second fitness of each red team combat experiment factor individual in the current red team population relative to the blue team population based on multiple first fitness values, and obtain multiple second fitness values.

[0052] According to some embodiments, the second fitness of each individual Red Team combat experiment factor in the Red Team population relative to the Blue Team population is calculated using the following formula:

[0053]

[0054]

[0055] Among them, PiB α represents the second fitness of the i-th red team experimental factor individual in the red team population relative to the blue team population. k p represents the weight of each individual Blue Team combat experiment factor in the Blue Team population. ik N represents the fitness of the i-th red team combat experiment factor individual in the red team population relative to the k-th blue team combat experiment factor individual in the blue team population, as output by the fitness evaluation machine learning model. B This indicates the number of Blue Team combat experiment factor individuals in the Blue Team population.

[0056] Step 3313: Calculate the average of multiple second fitness values.

[0057] Step 3314: Determine the fitness assessment result based on the average value.

[0058] In some embodiments, since it is necessary to determine whether the Red Team population needs to continue evolving, the fitness calculation is performed primarily on the Red Team population, thereby improving the relevance and effectiveness of the fitness assessment results and enabling a more accurate search for the Red Team population that can provide optimal combat effectiveness.

[0059] Step 2320: In response to the fitness assessment result being less than a preset threshold, perform an evolution operation on the current red team population to evolve the current red team population, and repeat step 2310 using the evolved current red team population.

[0060] In some embodiments, the above evolutionary operations include, but are not limited to, crossover, mutation, selection, and elimination, and there are no restrictions on these operations.

[0061] Step 2330: In response to the fitness assessment result being greater than a preset threshold, the red team's combat experiment scheme represented by the current red team population is taken as the target red team's combat experiment scheme.

[0062] Figure 4 The figure illustrates a schematic diagram of the game-playing evolution process of the red team's population against the blue team's population, according to an embodiment of this disclosure. (Reference) Figure 4 The blue team does not evolve, while the red team continuously evolves by engaging in game-playing and confrontation with the blue team and performing evolutionary operations on the red team whose fitness assessment results are greater than a preset threshold. In this way, the target red team combat experimental scheme that meets the requirements can be obtained in the end.

[0063] According to some embodiments, such as Figure 5 As shown, in addition to steps 510 to 5330 corresponding to steps 210 to 2330 above, the combat simulation experiment design optimization method based on evolutionary game theory in this embodiment of the present disclosure also includes the following steps 540 to 560.

[0064] Step 540: Use the simulation combat experiment system to conduct a game-like confrontation between the target red team's combat experiment scheme and the blue team's combat experiment scheme represented by the blue team population.

[0065] Step 550: Obtain the simulation results of the simulated combat experimental system.

[0066] In some embodiments, the above simulation results include, but are not limited to, evaluation indicators such as combat capability, combat effectiveness, and contribution rate, and there are no restrictions on these.

[0067] Step 560: Evaluate the merits of the target red force combat experiment scheme based on the simulation results.

[0068] In some embodiments, the target red force's combat experiment scheme can be input into the simulation combat experiment system for combat simulation, so as to effectively evaluate the performance of the fitness evaluation machine learning model and the optimization effect of the above-mentioned combat simulation experiment design optimization method based on evolutionary game theory based on the simulation results.

[0069] In one example, by analyzing the target red team's combat experiment scheme and the corresponding simulation results, we can obtain the upper and lower bounds, probability distributions, and inflection points of the corresponding experimental factors, so as to better realize the optimized design of subsequent experiments.

[0070] According to the evolutionary game theory-based combat simulation experiment design optimization method of this disclosure, by introducing an evolutionary game theory algorithm, the sample space of the blue side's adversarial schemes is first determined. Then, a proxy model is used to realize the simulated game adversarial between the red and blue sides to evaluate fitness. This can eliminate red side experimental schemes with low fitness, and update and optimize some of the red side's better combat schemes through evolutionary operations to improve the fitness of the red side's combat experimental schemes. Therefore, it can effectively reduce the number of experiments, improve experimental efficiency, and overcome the difficulties and challenges of complex combat simulation experiment design, such as uncertain opponents, long simulation experiment time, and large experimental search space, thereby optimizing the combat simulation experiment design process.

[0071] Furthermore, evolutionary algorithms possess strong global optimization capabilities, performing adaptive probabilistic search across the entire sample space to ensure globally optimal performance. Moreover, operations such as selection, crossover, and mutation are performed probabilistically, increasing the flexibility of the search process and enabling convergence to the optimal solution with a higher probability. This ensures the representativeness of the small sample space obtained through evolutionary game theory to the entire sample space. Further, evolutionary algorithms begin the search for the optimal solution from an initial population composed of many individuals, rather than starting with a single individual. The selection, crossover, and mutation operations performed on the initial population dynamically generate a new generation of individuals, incorporating much population information. This information helps avoid searching for unnecessary points, thus preventing getting trapped in local optima and gradually approaching the global optimum.

[0072] Figure 6 A flowchart illustrating the training process of a fitness evaluation machine learning model according to an embodiment of this disclosure is shown.

[0073] According to some embodiments, such as Figure 6 As shown, the training process of the above-mentioned fitness evaluation machine learning model includes the following steps.

[0074] Step 610: Obtain the current Blue Force combat plan. The current Blue Force combat plan includes multiple Blue Force combat experiment factors. Each Blue Force combat experiment factor is associated with the combat effectiveness in the combat simulation experiment. Each Blue Force combat experiment factor has a factor type and a factor level.

[0075] Step 620: Obtain the current Red Force combat plan. The current Red Force combat plan includes multiple Red Force combat experiment factors. Each Red Force combat experiment factor is associated with the combat effectiveness in the combat simulation experiment. Each Red Force combat experiment factor has a factor type and a factor level.

[0076] In some embodiments, the aforementioned current blue team combat plan and red team combat plan can be generated randomly or pre-designed according to experimental requirements, without any limitation.

[0077] Step 630: Conduct a sequential simulation combat experiment using the current Blue Force combat plan and the current Red Force combat plan to obtain a sample training dataset.

[0078] It should be noted that the sequential simulation combat experiment here refers to combat simulation experiment design based on the sequential experimental design method. Specifically, in the sequential experimental design method, the number of experiments is not predetermined. Instead, after each one or several experiments, it is determined whether the stopping rule has been met. In response to meeting the stopping rule, a decision analysis is performed. Based on this, by introducing the sequential experimental design method, the simulation combat experiment design can be viewed as a multi-iterative process, thereby achieving the goal of experiment design optimization with as few experiments as possible.

[0079] Step 630 includes steps 6310 to 6360.

[0080] Step 6310: Adjust the factor type and / or factor level of at least one Blue Force operational experiment factor in the current Blue Force operational plan to obtain the sequential Blue Force operational plan.

[0081] Step 6320: Adjust the factor type and / or factor level of at least one Red Force operational experimental factor in the current Red Force operational plan to obtain the sequential Red Force operational plan.

[0082] It should be noted that in each iteration, only step 6310, only step 6320, or both steps 6310 and 6320 can be executed; there are no restrictions on this. Adjustments to the red and blue team's combat plans can make the generated training data more diverse, thereby improving the performance of the trained model.

[0083] Step 6330: Input the sequential blue team combat plan and the sequential red team combat plan into the simulation combat experiment system for simulated game confrontation, and obtain sample simulation results. The sample simulation results indicate the degree of superiority or inferiority of the sequential red team combat plan against the sequential blue team combat plan.

[0084] Step 6340: Add the sequential blue team combat plan, the sequential red team combat plan, and the sample simulation results as a set of sample data to the sample training dataset.

[0085] Step 6350: In response to the number of sample data groups in the sample training dataset being less than or equal to a preset threshold, the sequential blue team combat plan is taken as the current blue team combat plan and the sequential red team combat plan is taken as the current red team combat plan, and step 6310 is repeated.

[0086] Step 6360: In response to the number of sample data groups in the sample training dataset being greater than a preset threshold, step 640 is executed.

[0087] Step 640: Train the initial machine learning model using the sample training dataset to obtain the fitness evaluation machine learning model.

[0088] In some embodiments, the initial machine learning model described above includes, but is not limited to, a deep neural network construction model, such as a multi-input, multi-output (MIMO) multilayer feedforward neural network structure. Furthermore, the number of network layers and neurons in the model is determined based on the number of input and output features and the model complexity. Based on this, a complex function mapping relationship from input to output can be established for a multi-input, multi-output dynamic system, enabling learning and training on large datasets to approximate real-world relationships as closely as possible.

[0089] In some embodiments, mean squared error (MSE), root mean square error (RMSE), and R0 can be used during model training. 2 The score is used as the fitting evaluation index of the machine learning model, and an appropriate error threshold is determined according to the needs of the simulation experiment in order to obtain a high-performance fitness evaluation machine learning model.

[0090] According to embodiments of this disclosure, employing a sequential experimental design method to form sequential experimental schemes can effectively reduce the time required to design operational experimental schemes. Simultaneously, it can ensure that the generated operational experimental schemes have strong generality, avoiding situations where the trained model cannot accurately evaluate the merits of a scheme due to excessive emphasis on a particular element during the design process. Furthermore, by inputting the generated sequential experimental schemes of both red and blue teams into a simulation experimental system for game-based competition to generate training sample data, the evaluation accuracy of the trained fitness evaluation machine learning model can be effectively improved, thereby enhancing the overall effectiveness of the experimental optimization design.

[0091] Figure 7 This is a schematic block diagram illustrating a combat simulation experiment design optimization device 700 based on evolutionary game theory, according to an exemplary embodiment. Figure 7 As shown, the device 700 includes: a first initialization unit 710, used to initialize a blue team population, the blue team population representing a blue team combat experiment scheme in a combat simulation experiment, wherein the blue team population includes multiple blue team combat experiment factor individuals, each blue team combat experiment factor individual being associated with combat effectiveness in the combat simulation experiment; a second initialization unit 720, used to initialize a red team population, the red team population representing a red team combat experiment scheme in a combat simulation experiment, wherein the red team population includes multiple red team combat experiment factor individuals, each red team combat experiment factor individual being associated with combat effectiveness in the combat simulation experiment; and a population evolution unit 730, used to iteratively evolve the red team population, the population evolution unit 730 including: a fitness calculation subunit 731, used to utilize fitness to evaluate machine learning models. The system evaluates the fitness of the current red team population against the blue team population to obtain a fitness evaluation result. The fitness evaluation machine learning model is trained so that the fitness indicates the superiority or inferiority of the red team's combat experimental plan against the blue team's combat experimental plan. An evolutionary operation subunit 732 is used to perform an evolutionary operation on the current red team population in response to the fitness evaluation result being less than a preset threshold, causing the current red team population to evolve. The evolved current red team population is then used to repeat the step of evaluating the fitness of the current red team population against the blue team population using the fitness evaluation machine learning model. A scheme determination subunit 733 is used to select the red team combat experimental plan represented by the current red team population as the target red team combat experimental plan in response to the fitness evaluation result being greater than a preset threshold.

[0092] It should be understood that Figure 7 The various modules of the device 700 shown can be connected to the reference. Figure 2The steps in method 200 described correspond to each other. Therefore, the operations, features, and advantages described above for method 200 also apply to apparatus 700 and its included modules. For the sake of brevity, some operations, features, and advantages will not be repeated here.

[0093] While specific functions have been discussed above with reference to specific modules, it should be noted that the functions of the modules discussed herein can be divided into multiple modules, and / or at least some functions of multiple modules can be combined into a single module. The specific module performing an action discussed herein includes the specific module itself performing the action, or alternatively, the specific module calling or otherwise accessing another component or module that performs the action (or performs the action in conjunction with the specific module). Therefore, a specific module performing an action can include the specific module performing the action itself and / or another module that performs the action, called or otherwise accessed by the specific module. For example, the fitness calculation subunit 731, evolutionary operation subunit 732, and scheme determination subunit 733 described above can be combined into a single module in some embodiments.

[0094] It should also be understood that this article can describe various technologies in the general context of software and hardware components or program modules. The above regarding... Figure 7 The various modules described can be implemented in hardware or in hardware in combination with software and / or firmware. For example, these modules can be implemented as computer program code / instructions configured to execute in one or more processors and stored in a computer-readable storage medium. Alternatively, these modules can be implemented as hardware logic / circuit. For example, in some embodiments, one or more of the first initialization unit 710, the second initialization unit 720, the population evolution unit 730, the fitness calculation subunit 731, the evolution operation subunit 732, and the scheme determination subunit 733 can be implemented together in a system on a chip (SoC). The SoC may include an integrated circuit chip (which includes a processor (e.g., a central processing unit (CPU), microcontroller, microprocessor, digital signal processor (DSP), etc.), memory, one or more communication interfaces, and / or one or more components of other circuitry) and may optionally execute received program code and / or include embedded firmware to perform functions.

[0095] According to one aspect of this disclosure, a computer device is provided, including a memory, a processor, and a computer program stored in the memory. The processor is configured to execute the computer program to implement the steps of any of the method embodiments described above.

[0096] According to one aspect of this disclosure, a non-transitory computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of any of the method embodiments described above.

[0097] According to one aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the steps of any of the method embodiments described above.

[0098] In the following text, combined with Figure 8 Illustrative examples describing such computer devices, non-transitory computer-readable storage media, and computer program products.

[0099] Figure 8 An example configuration of a computer device 800 that can be used to implement the methods described herein is shown. For example, Figure 1 The server 120 and / or client device 110 shown may include an architecture similar to computer device 800. The aforementioned alias entity identification device may also be implemented wholly or at least partially by computer device 800 or similar devices or systems.

[0100] Computer device 800 can be a variety of different types of devices. Examples of computer device 800 include, but are not limited to: desktop computers, server computers, laptop or netbook computers, mobile devices (e.g., tablet computers, cellular or other wireless phones (e.g., smartphones), notebook computers, mobile stations), wearable devices (e.g., glasses, watches), entertainment devices (e.g., entertainment appliances, set-top boxes communicatively coupled to a display device, game consoles), televisions or other display devices, automotive computers, and so on.

[0101] Computer device 800 may include at least one processor 802, memory 804, multiple communication interfaces 806, display device 808, other input / output (I / O) devices 810, and one or more mass storage devices 812 capable of communicating with each other, such as via system bus 814 or other suitable connections.

[0102] Processor 802 may be a single processing unit or multiple processing units, and all processing units may include single or multiple computing units or multiple cores. Processor 802 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuits, and / or any device that manipulates signals based on operating instructions. Among other capabilities, processor 802 may be configured to acquire and execute computer-readable instructions stored in memory 804, mass storage device 812, or other computer-readable media, such as program code of operating system 816, program code of application program 818, program code of other program 820, etc.

[0103] Memory 804 and mass storage device 812 are examples of computer-readable storage media for storing instructions executed by processor 802 to perform the various functions described above. For example, memory 804 may generally include both volatile and non-volatile memory (e.g., RAM, ROM, etc.). Furthermore, mass storage device 812 may generally include hard disk drives, solid-state drives, removable media, including external and removable drives, memory cards, flash memory, floppy disks, optical disks (e.g., CDs, DVDs), storage arrays, network-attached storage, storage area networks, etc. Both memory 804 and mass storage device 812 may be collectively referred to herein as memory or computer-readable storage media, and may be non-transitory media capable of storing computer-readable, processor-executable program instructions as computer program code, which may be executed by processor 802 as a specific machine configured to perform the operations and functions described in the examples herein. Multiple programs may be stored on mass storage device 812. These programs include an operating system 816, one or more application programs 818, other programs 820, and program data 822, and they can be loaded into memory 804 for execution.

[0104] Although Figure 8 The modules 816, 818, 820, and 822, or portions thereof, are illustrated as being stored in memory 804 of computer device 800; however, modules 816, 818, 820, and 822 may be implemented using any form of computer-readable medium accessible by computer device 800. As used herein, “computer-readable medium” includes at least two types of computer-readable media: computer-readable storage media and communication media.

[0105] Computer-readable storage media include volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. Computer-readable storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, DVD, or other optical storage devices, magnetic cassettes, magnetic tapes, disk storage devices or other magnetic storage devices, or any other non-transmission medium that can be used to store information for access by computer devices. In contrast, communication media can embody computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms. Computer-readable storage media as defined herein do not include communication media.

[0106] One or more communication interfaces 806 are used for exchanging data with other devices, such as via a network, direct connection, etc. Such communication interfaces can be one or more of the following: any type of network interface (e.g., a network interface card (NIC)), wired or wireless (such as IEEE 802.11 Wireless LAN (WLAN)) wireless interface, Wi-MAX interface, Ethernet interface, Universal Serial Bus (USB) interface, cellular network interface, Bluetooth. TM Interfaces include near-field communication (NFC) interfaces, etc. Communication interface 806 facilitates communication across various network and protocol types, including wired networks (e.g., LAN, cable, etc.) and wireless networks (e.g., WLAN, cellular, satellite, etc.), the Internet, etc. Communication interface 806 can also provide communication with external storage devices (not shown) such as storage arrays, network-attached storage, storage area networks, etc.

[0107] In some examples, a display device 808, such as a monitor, may be included for displaying information and images to the user. Other I / O devices 810 may be devices that receive various inputs from the user and provide various outputs to the user, and may include touch input devices, gesture input devices, cameras, keyboards, remote controls, mice, printers, audio input / output devices, and so on.

[0108] The techniques described herein can be supported by these various configurations of computer device 800, and are not limited to specific examples of the techniques described herein. For example, the functionality can also be implemented wholly or partially on a “cloud” using a distributed system. A cloud includes and / or represents a platform for resources. The platform abstracts the underlying functionality of the cloud’s hardware (e.g., servers) and software resources. Resources may include applications and / or data that can be used when performing computational processing on a server remote from computer device 800. Resources may also include services provided via the Internet and / or via subscriber networks such as cellular or Wi-Fi networks. The platform can abstract resources and functionality to connect computer device 800 to other computer devices. Therefore, the implementation of the functionality described herein can be distributed throughout the cloud. For example, the functionality may be implemented partly on computer device 800 and partly through a platform that abstracts the functionality of the cloud.

[0109] Although this disclosure has been described and illustrated in detail in the accompanying drawings and the foregoing description, such description and illustration should be considered illustrative and suggestive, not restrictive; this disclosure is not limited to the disclosed embodiments. By studying the drawings, the disclosure, and the appended claims, those skilled in the art will be able to understand and implement variations of the disclosed embodiments in practice with respect to the claimed subject matter. In the claims, the word "comprising" does not exclude other elements or steps not listed, the indefinite article "a" or "an" does not exclude a plurality, the term "a plurality" means two or more, and the term "based on" should be interpreted as "at least partially based on". The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be beneficial.

Claims

1. A method for designing and optimizing combat simulation experiments based on evolutionary game theory, comprising: Initialize the blue team population, which represents the blue team combat experiment scheme in the combat simulation experiment. The blue team population includes multiple blue team combat experiment factor individuals, and each blue team combat experiment factor individual is associated with the combat effectiveness in the combat simulation experiment. Initialize the Red Force population, which represents the Red Force combat experiment scheme in the combat simulation experiment. The Red Force population includes multiple Red Force combat experiment factor individuals, each of which is associated with combat effectiveness in the combat simulation experiment. The process of iteratively evolving the red population includes the following steps: The fitness of the current red team population against the blue team population is evaluated using a fitness evaluation machine learning model to obtain a fitness evaluation result. The fitness evaluation machine learning model is trained such that the fitness indicates the degree of superiority or inferiority of the red team's combat experimental plan against the blue team's combat experimental plan. In response to the fitness assessment result being less than a preset threshold, an evolutionary operation is performed on the current red team population to evolve it, and the step of evaluating the fitness of the current red team population relative to the blue team population using a fitness assessment machine learning model is repeated using the evolved current red team population; and In response to the fitness assessment result being greater than a preset threshold, the red team combat experiment scheme represented by the current red team population is taken as the target red team combat experiment scheme.

2. The method according to claim 1, wherein, The step of evaluating the fitness of the current red population against the blue population using a fitness evaluation machine learning model includes: The red team population and the blue team population are input into the fitness evaluation machine learning model to obtain the first fitness of each red team combat experiment factor individual in the current red team population relative to each blue team combat experiment factor individual in the blue team population, so as to obtain multiple first fitnesss. The second fitness of each red team combat experiment factor individual in the current red team population relative to the blue team population is calculated based on the multiple first fitness values, resulting in multiple second fitness values; Calculate the average of the plurality of second fitness values; and The fitness assessment result is determined based on the average value.

3. The method according to claim 2, wherein, The second fitness of each individual Red Team combat experimental factor in the Red Team population relative to the Blue Team population is calculated using the following formula: Among them, P iB α represents the second fitness of the i-th red team combat experiment factor individual in the red team population relative to the blue team population. k p represents the weight of each individual Blue Team combat experiment factor in the Blue Team population. ik N represents the fitness of the i-th red team combat experiment factor individual in the red team population relative to the k-th blue team combat experiment factor individual in the blue team population, as output by the fitness evaluation machine learning model. B This indicates the number of individual Blue Team combat experiment factor individuals in the Blue Team population.

4. The method according to claim 1, wherein, Each of the multiple Red Team operational experimental factor individuals has a factor type and a factor level, and each of the multiple Blue Team operational experimental factor individuals has a factor type and a factor level. At least two Blue Team operational experimental factors with the same factor type have different factor levels.

5. The method according to claim 1, wherein, The training process of the fitness evaluation machine learning model includes the following steps: Obtain the current Blue Force combat plan, which includes multiple Blue Force combat experiment factors. Each Blue Force combat experiment factor is associated with combat effectiveness in combat simulation experiments. Each Blue Force combat experiment factor has a factor type and a factor level. Obtain the current Red Force combat plan, which includes multiple Red Force combat experiment factors. Each Red Force combat experiment factor is associated with combat effectiveness in combat simulation experiments. Each Red Force combat experiment factor has a factor type and a factor level. A sequential simulation combat experiment was conducted using the current Blue Force combat plan and the current Red Force combat plan to obtain a sample training dataset. as well as The initial machine learning model is trained using the sample training dataset to obtain the fitness evaluation machine learning model; The sequential simulation combat experiment using the current blue team's combat plan and the current red team's combat plan includes the following steps: Adjust the factor type and / or factor level of at least one blue force operational experiment factor in the current blue force operational plan to obtain a sequential blue force operational plan; Adjust the factor type and / or factor level of at least one Red Force operational experimental factor in the current Red Force operational plan to obtain a sequential Red Force operational plan; The sequential blue team combat plan and the sequential red team combat plan are input into the simulation combat experiment system for simulated game confrontation to obtain sample simulation results. The sample simulation results indicate the degree of superiority or inferiority of the sequential red team combat plan against the sequential blue team combat plan. The sequential blue team combat plan, the sequential red team combat plan, and the sample simulation results are added as a set of sample data to the sample training dataset; In response to the number of sample data groups in the sample training dataset being less than or equal to a preset threshold, the sequential blue team combat plan is taken as the current blue team combat plan, and the sequential red team combat plan is taken as the current red team combat plan, and the step of adjusting the factor level of at least one blue team combat experiment factor in the current blue team combat plan is repeated; and In response to the number of sample data groups in the sample training dataset being greater than the preset threshold, the step of training an initial machine learning model using the sample training dataset is executed.

6. The method according to any one of claims 1-5, further comprising: The target red team's combat test scheme and the blue team's combat test scheme represented by the blue team population were used to conduct a game-like confrontation using a simulated combat test system. Obtain the simulation results of the simulated combat experimental system; The merits and demerits of the target red force combat experiment scheme are evaluated based on the simulation results.

7. A combat simulation experiment design optimization device based on evolutionary game theory, comprising: The first initialization unit is used to initialize the blue team population, which represents the blue team combat experiment scheme in the combat simulation experiment. The blue team population includes multiple blue team combat experiment factor individuals, and each blue team combat experiment factor individual is associated with the combat effectiveness in the combat simulation experiment. The second initialization unit is used to initialize the Red Force population, which represents the Red Force combat experiment scheme in the combat simulation experiment. The Red Force population includes multiple Red Force combat experiment factor individuals, and each Red Force combat experiment factor individual is associated with the combat effectiveness in the combat simulation experiment. Population evolution unit, used to iteratively evolve the red team population, the population evolution unit comprising: The fitness calculation subunit is used to evaluate the fitness of the current red team population against the blue team population using a fitness evaluation machine learning model to obtain a fitness evaluation result, wherein the fitness evaluation machine learning model is trained such that the fitness indicates the degree of superiority or inferiority of the red team's combat experiment plan against the blue team's combat experiment plan. An evolutionary operation subunit is configured to, in response to the fitness evaluation result being less than a preset threshold, perform an evolutionary operation on the current red team population to evolve the current red team population, and repeat the step of evaluating the fitness of the current red team population relative to the blue team population using a fitness evaluation machine learning model using the evolved current red team population; and The scheme determines a sub-unit, which is used to take the red team combat experiment scheme represented by the current red team population as the target red team combat experiment scheme in response to the fitness evaluation result being greater than a preset threshold.

8. A computer device, comprising: At least one processor; as well as The memory, on which computer programs are stored, When the computer program is executed by the processor, it causes the processor to perform the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, causes the processor to perform the method of any one of claims 1 to 6.

10. A computer program product comprising a computer program that, when executed by a processor, causes the processor to perform the method of any one of claims 1 to 6.

Images