Man-machine intelligent game confrontation scene design method in war game deduction field

By designing complex and changeable human-machine intelligent game confrontation scenarios in the field of wargame deduction, combining decision-making, learning and reactive AI, and combining system engineering concepts, the universality problem of intelligent game system scenario design is solved, achieving more efficient and More applicable military decision-making support.

CN119004986APending Publication Date: 2024-11-22Chinese People's Liberation Army Cyberspace Force Information Engineering University
View PDF 0 Cites 11 Cited by

Patent Information

Application Number
CN202411120746.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2024-06-06
Filing Date
2024-08-15
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing intelligent game system scenario design lacks universality and cannot meet the needs of different fields. Especially in the military field, intelligent game systems are difficult to directly apply to the global decision-making environment, resulting in insufficient universal and effective decision-making support.

Method used

A human-computer intelligent game confrontation scenario design method in the field of wargame deduction is proposed. By setting scene themes, game roles, deduction environments and rules, complex and changeable confrontation scenarios are formed, including decision-making AI, learning AI and reactive AI The setting of the system engineering concept is combined with the system engineering concept, and the scene design is optimized to meet different needs.

Benefits of technology

It improves the intelligence level and scene authenticity of the war chess deduction system, enhances the accuracy and adaptability of the deduction, provides more scientific support for military training and decision-making analysis, and improves the universality and practicality of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004986A_ABST
    Figure CN119004986A_ABST
Patent Text Reader

Abstract

The invention discloses a man-machine intelligent game confrontation scene design method in the field of war game deduction, relates to comprehensive planning from themes to details, and aims to construct real, complex and diversified deduction scenes for a man-machine intelligent game system. Macroscopically, the basic process of scene design of the man-machine intelligent game system can be summarized into six main links including scene theme setting, role setting, environment setting, rule setting, situation setting and verification and testing. According to the method, machine intelligent agent settings with different decision-making preferences are introduced, so that intelligent design of a deduction scene is embodied. By constructing complex and diversified game confrontation roles, behavior modes and decision-making processes of different participants are simulated, so that a deduction scene is closer to an actual battlefield condition. According to the method, the deduction accuracy is improved, the effects of different strategies can be better predicted and evaluated, and a scientific basis is provided for military decision making.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of war game simulation and system engineering technology. Specifically, the present invention provides a human-machine intelligent game confrontation scenario design method for testing and improving the decision-making ability and response strategy of an intelligent system in a war game simulation. Background Art

[0002] 1. Development Background of Intelligent Gaming System

[0003] According to different human-machine confrontation tasks, researchers can model three types of games: complete information games, incomplete information games, and incomplete information multi-agent games. These different types of game models have different characteristics and uses in research and application, providing a framework for studying and solving practical problems, helping people better understand and analyze game behaviors. In particular, multi-agent games are not only of great research value in the field of artificial intelligence research, but also have great potential application value in social management, intelligent transportation, economy, military and other fields. Many scientific research institutions at home and abroad are also competing to invest in this field.

[0004] Game theory provides a theoretical framework for solving real-world problems of human-machine confrontation, while intelligent technology provides an efficient method for solving strategies. The combination of game theory and intelligent technology forms the discipline of intelligent game. The research content of intelligent game mainly focuses on two core areas: the construction of game rules and the solution of game strategies.

[0005] The construction of game rules is the foundation of intelligent games. The rules define the basic framework of the game and the way of interaction between participants. The carefully designed game rules are to simulate various complex competition and cooperation scenarios in the real world and provide a suitable experimental environment for studying game strategies. Therefore, participants must strictly follow these existing rules in the decision-making process of pursuing their own interests.

[0006] Solving game strategies is the core task of intelligent games. Given the game rules, how to find the optimal or near-optimal strategy is the key to intelligent game research. According to the profound insights of Nash's theorem, in actual games, although Nash equilibrium cannot guarantee a win in every game, in the long run, it is realistic to significantly increase the probability of winning, which provides important guidance for the formulation of intelligent game strategies.

[0007] In the study of intelligent games, the construction of game rules and the solution of game strategies complement each other. On the one hand, reasonable game rules can help simplify the difficulty of strategy solution and improve the efficiency of solution; on the other hand, efficient strategy solution methods can also reveal the potential laws and characteristics in game rules and provide guidance for the optimization and improvement of rules.

[0008] Therefore, the research content of intelligent games not only involves the basic theories and methods of game theory, but also needs to use the power of intelligent technology to promote the in-depth development of research. With the continuous advancement of intelligent technology, intelligent games are showing their unique value and potential in more fields.

[0009] 2. Scenario Design of Human-Machine Intelligent Game System

[0010] 1. The advantages and functions of human-machine intelligent game system scenario applications cannot be ignored.

[0011] First, scenario applications can significantly improve decision-making efficiency and accuracy. In a complex and ever-changing real-world environment, traditional decision-making methods often have difficulty coping with various uncertainties. The human-machine intelligent game system can quickly process massive amounts of information and accurately judge the situation through technologies such as big data analysis and deep learning, thereby making more scientific and reasonable decisions. This efficient and accurate decision-making method not only improves work efficiency, but also reduces decision-making risks, bringing tangible benefits to enterprises and society.

[0012] Secondly, scenario applications can be used to train artificial intelligence, enabling intelligent game systems to cope with complex and uncertain environments. By simulating various actual scenarios, the artificial intelligence system can learn and grow in continuous practice, which can greatly improve the generalization ability and intelligence of artificial intelligence. This training method not only makes the artificial intelligence system more adaptable to the complexity of the real world, but also enables it to respond quickly when facing unknown challenges and show a higher level of intelligence.

[0013] In addition, scenario applications can also provide personalized and customized services. Users' needs and preferences vary greatly. Through intelligent systems and algorithms, we can provide personalized services and recommendations based on the unique needs of each user. This customized service approach helps improve user experience and satisfaction.

[0014] 2. The demand for game scenarios in intelligent game systems is becoming increasingly urgent, but the scenario design methods for human-machine intelligent game systems are still in short supply.

[0015] Since the Turing test, human-machine confrontation has been an effective means and method for evaluating the level of intelligence. However, the Turing test mainly focuses on the limitations of language and behavior, as well as the complexity and uncertainty of human intelligence and cognition, which makes it difficult to evaluate machine intelligence.

[0016] The demand for game scenarios in intelligent game systems is becoming increasingly urgent, mainly due to the complex and ever-changing decision-making environment and growing demand for intelligence in the real world. Whether it is business competition, financial investment or military confrontation, decisions need to be made quickly and accurately to cope with the ever-changing situation. However, although the technology of intelligent game systems is becoming increasingly novel and the theory is constantly developing, most intelligent game scenario designs are still targeted at local professional fields. This leads to the limitations and non-universality of scenario design, which cannot meet the needs of intelligent game systems in different fields. For example, intelligent game systems in the field of electronic warfare may not be directly applied to the global military field because they face completely different decision-making environments and rules. In order to solve this problem, we urgently need a relatively general method for scenario design of human-machine intelligent game systems. This method should be able to comprehensively consider the commonalities and differences of different fields and design game scenarios that are both in line with actual conditions and universal.

[0017] In short, the lack of a relatively universal human-machine intelligent game system scenario design method is an important problem facing the current intelligent game field. In order to solve this problem, we need to deeply analyze the commonalities and differences of game scenarios in different fields, draw on existing experience and technology to innovatively propose new scenario design methods, and focus on the customizability and scalability of scenario design. Only in this way can we design intelligent game system scenarios that are more in line with actual needs and have universal applicability, and provide more efficient and intelligent decision-making support for different fields. Summary of the invention

[0018] In view of the above problems, the present invention proposes a method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation, which realizes the design of human-machine intelligent game confrontation scenarios in the field of war game simulation. This method not only standardizes the various elements of the game scenario, but also designs the organization and application of the game elements, and finally forms a combat scenario that meets the needs of the designer. This design method helps to improve the decision-making ability, adaptability and strategic thinking of human-machine intelligent bodies, and provides more realistic and comprehensive support for military training and decision analysis.

[0019] In order to achieve the above object, the present invention adopts the following technical solutions:

[0020] The present invention proposes a method for designing a human-machine intelligent game confrontation scenario in the field of war game simulation, comprising:

[0021] Step 1: Establish the scenario theme based on the specific needs and goals of the war game, and refine the scenario theme based on the needs;

[0022] Step 2: On the basis of the scene theme, the game roles are set, and the game roles include human agents, machine agents, and human-machine intelligence hybrids; the human agent is an agent played by humans, the machine agent is an AI agent or multiple AI composite agents, and the human-machine intelligence hybrid is a composite agent composed of humans and one or more AI agents;

[0023] Step 3: Set the intensity of the game and the specific time and space background according to the needs of the game, complete the game environment setting, and provide an action framework for the game characters;

[0024] Step 4: According to the simulation objectives and scenario characteristics, formulate detailed combat rules, victory and defeat criteria, and constraints in the game process to provide decision-making basis for game players;

[0025] Step 5: Determine the initial situation in the simulation scenario based on the scenario theme, game roles, and simulation environment settings. The initial situation includes intelligence data, command relationships, and force deployment, and the initial situation is used as the starting point for the simulation.

[0026] Step 6: Verify and test the designed human-machine intelligent game confrontation scenario. Through simulation system simulation or actual human-machine confrontation experiments, verify the rationality and effectiveness of the scenario design. During the verification and testing process, collect performance data and feedback from the human-machine intelligent agent, optimize and improve the scenario, and ensure that it meets the needs and goals of the war game simulation.

[0027] Furthermore, the scenario themes include verifying operational concepts, supporting operational training, evaluating equipment effectiveness, and testing innovative operational methods.

[0028] Furthermore, step 2 also includes: setting different levels of intelligence according to the importance of the machine agent in the game role in order to formulate strategies from basic to advanced.

[0029] Furthermore, the machine intelligent agent includes decision-making AI, learning AI and reactive AI. Decision-making AI is responsible for global strategic planning and decision-making, learning AI is responsible for local collection and analysis of data to optimize decisions, and reactive AI is responsible for autonomously performing corresponding game tasks according to its own specific game rules.

[0030] Furthermore, the decision-making AI is set as follows:

[0031] 1) Define goals and tasks

[0032] Set strategic goals and break them down into a series of executable subtasks;

[0033] 2) Design Algorithm

[0034] Design decision-making algorithms: Combine short-term tactical goals with long-term strategic goals to make comprehensive decisions; the decision-making algorithm operates in a dynamic battlefield environment, traverses the decision tree under the current situation, and finds the optimal strategy to achieve short-term tactical goals through trial and error; on this basis, conduct situation assessment, evaluate the changes in the battlefield situation after executing this optimal strategy for tactical goals, and comprehensively consider this decision-making strategy based on the contribution of its execution results to long-term strategic goals, so as to achieve the organic coordination of short-term tactical goals and long-term strategic goals;

[0035] Design a decision-making algorithm evolution mechanism: Use the state-action-reward model to define the state, action, and corresponding reward in the decision-making process to guide the optimization of decision-making AI strategies; adjust strategies in real time based on instant feedback from battle results and resource changes; at the same time, optimize and upgrade long-term strategies based on previous decision-making results and long-term feedback from game results; modify the decision algorithm by modifying the decision tree and iterative optimization to achieve decision algorithm evolution;

[0036] 3) Establish a knowledge base and rule system

[0037] Use past game records to build a knowledge base and set decision rules and priorities based on experience and domain knowledge;

[0038] 4) Make decisions based on subtasks

[0039] High-level decision making: developing and modifying overall strategies and long-term plans in real time;

[0040] Mid-level decision-making: Based on the strategic goals determined by high-level decisions, sub-tasks are divided, resources are allocated, and specific mid-term tasks are formulated;

[0041] Low-level decision-making: Specific actions and actual use of resources for executing subtasks based on mid-term tasks determined by mid-level decisions.

[0042] Furthermore, the learning AI is configured as follows:

[0043] 1) Define learning objectives

[0044] Define learning content, including pattern recognition and dynamic adaptation to the environment; set prediction accuracy and response time to measure how close the learning AI is to the learning goal;

[0045] 2) Choose a learning method

[0046] Select the following learning algorithms based on the data type: supervised learning, unsupervised learning, or semi-supervised learning; introduce online learning mechanisms to collect feedback from actual operations and update and improve the algorithms in real time;

[0047] 3) Build a dataset

[0048] Perform data cleaning, automatically identify and extract important features, reduce data processing dimensions, collect and construct datasets for training and validation from data input sources;

[0049] 4) Training and validating the model

[0050] Use the constructed data set to train the model and adjust the model parameters to optimize performance; verify the model and continuously iterate and adjust the model parameters;

[0051] 5) Deployment and iterative optimization

[0052] Deploy trained models in actual game environments, conduct real-time data analysis and decision support, and integrate learning AI into the overall decision-making system.

[0053] Furthermore, the reactive AI is configured as follows:

[0054] 1) Define reaction rules

[0055] Define the types of actions that reactive AI will perform and clarify the triggering conditions for the actions;

[0056] 2) Matching state-action pairs

[0057] Use state machines to define different states and their transition conditions, and construct behavior trees to represent behaviors and action combinations using a hierarchical tree structure.

[0058] Define the actions to be executed and their specific effects, and set the execution conditions so that the reactive AI can select the appropriate actions to execute based on the current state and environmental conditions;

[0059] 3) Adjustment and optimization

[0060] Adjust operating rules and priorities in real time based on actual conditions and feedback, conduct various stress tests and scenario simulations regularly, monitor the performance of reactive AI, and continuously optimize response strategies through trial and error and reward mechanisms.

[0061] Furthermore, the step 2 further includes: setting a unique decision-making style for each machine agent; according to different decision-making styles, the machine agents include adventurous machine agents, rational machine agents and conservative machine agents;

[0062] Set up the adventurous machine agent as follows:

[0063] Build risk assessment models to quantify the risk level of different decisions, but at the same time ensure that the risk assessment model has a high tolerance for high-risk strategies; design a reward mechanism that prefers future returns and gives high weight to the long-term returns of high-risk actions; enable AI agents to explore multiple strategy paths simultaneously, evaluate potential returns under different risk strategies, and optimize selection opportunities; and embed risk management mechanisms;

[0064] Set up the Intelligent Machine Agent as follows:

[0065] Establish a risk assessment system that considers both potential benefits and risks; adjust the balance between algorithm benefits and risks and set preference parameters to maximize benefits while controlling risks; use historical and current data for analysis; and predict decision outcomes through scenario simulation or scenario analysis, and evaluate the possible consequences of various decision options in different scenarios; introduce a continuous feedback mechanism so that AI agents can learn from each decision outcome and dynamically adjust decision strategies to enhance the benefits of future decisions;

[0066] Set up the conservative machine agent as follows:

[0067] Establish a risk warning system to identify and evaluate possible risks at an early stage; set up a risk avoidance mechanism to reward decisions with low risks and stable returns, and ensure that strategies tend to avoid major risks through risk weight adjustments; at the same time, select or develop algorithms that tend to act robustly with the goal of stable growth; simulate the results of games under extremely unfavorable conditions, and prepare buffer strategies for possible losses based on the results of games.

[0068] Furthermore, step 3 includes: setting the game intensity and setting the game dimension; setting the game dimension includes setting the physical domain, cognitive domain, information domain and time and space constraints; the physical domain includes terrain, weather and facilities; the cognitive domain includes the psychological state, cultural knowledge background and experience accumulation of the participants; the information domain includes the areas of information transmission, processing and utilization in the game; the time and space constraints include combat time, battlefield scope and action time limit.

[0069] Furthermore, the step 4 comprises:

[0070] Setting the rules of military confrontation: Formulate military action norms so that the game players can play the game in accordance with military laws, including activity space restrictions, fire damage effects, and weapon equipment performance;

[0071] Set resource scheduling rules: clarify the acquisition methods, allocation principles, dynamic adjustments and usage restrictions of manpower, materials and equipment, including game resources, to simulate the principle of limited resources in the game process;

[0072] Setting rules for friendly and enemy organizational relationships: defining alliances, hostility or neutral relationships between game characters and their organizational structures, providing a basis for teamwork and confrontation during the game, and influencing the game dynamics;

[0073] Set system engineering rules: stipulate the comprehensive effects between game roles and between game roles and the game environment, analyze the game system as a whole, and comprehensively consider the interrelationships and mutual constraints between various parts.

[0074] Compared with the prior art, the present invention has the following beneficial effects:

[0075] The present invention proposes a human-machine intelligent game confrontation scenario design method in the field of war game simulation. The method aims to improve the intelligence level of the war game simulation system, enhance the authenticity and complexity of the simulation scenario, and create a series of highly realistic, complex and changeable confrontation scenarios. These scenarios can fully simulate the complexity and dynamics of the real battlefield environment, thereby better simulating the actual battlefield environment, and at the same time provide strong support for military training and decision analysis.

[0076] The present invention embodies the intelligent design of the deduction scene by introducing machine agent settings with different decision preferences. By constructing complex and diverse game confrontation roles and simulating the behavior patterns and decision-making processes of different participants, the deduction scene is made closer to the actual battlefield situation. This method not only improves the accuracy of the deduction, but also can better predict and evaluate the effects of different strategies, providing a scientific basis for military decision-making.

[0077] The present invention focuses on the diversity and customizability of the deduction scenarios. Through various theme presets and flexible rule adjustment mechanisms, users can customize different deduction scenarios according to specific needs, including game themes, game role settings, game intensity, game rules, etc. This design method enables the deduction system to adapt to different training and research needs, improving the versatility and practicality of the system.

[0078] The present invention also introduces the concept of system engineering. Introducing the concept of system engineering into the internal rules of the scene can eliminate the system errors caused by each local coupling process, making the scene effect more in line with reality. Introducing the concept of system engineering into the scene design process of the entire human-machine intelligent game system can make the design process form a closed loop that can be continuously optimized, and realize the self-correction of the scene through its own cycle.

[0079] The present invention plays a role in training machine agents. By playing games with machine agents in different scenarios, the weaknesses and deficiencies in their decision-making process can be fully exposed, and their algorithms and models can be optimized and improved in a targeted manner. At the same time, the diverse scenarios can also improve the adaptability of machine agents, enabling them to flexibly respond to various challenges in different environments and conditions.

[0080] In summary, the present invention allows users to adjust scenario parameters according to actual needs and generate confrontation scenarios of different difficulty and complexity, thereby meeting training and research requirements of different levels and needs, significantly improving the practicality of the war game simulation system, and is of great significance for improving the level of military training and strategic research. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 The present invention is a flowchart of a method for designing a human-machine intelligent game confrontation scenario in the field of war game simulation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0082] The present invention will be further explained below with reference to the accompanying drawings and specific embodiments:

[0083] like Figure 1 As shown, the present invention is a method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation, which involves all-round planning from themes to details, and aims to build real, complex and diverse simulation scenarios for human-machine intelligent game systems. Macroscopically, the basic process of human-machine intelligent game system scenario design can be summarized into six main steps: scenario theme setting, role setting, environment setting, rule setting, situation setting, verification and testing. In the specific implementation process, this method can be combined with the current war game simulation, and the scene design of the user's intelligent game system can be set according to the six links of this method to standardize the various elements of the game scene, and design the organization and application of the game elements, and finally form a combat scene that meets the needs of the designer. The design method includes the following steps performed in sequence:

[0084] Step 1: Establish the theme of the scene

[0085] According to the specific needs and goals of the war game, the theme of the game scenario should be clarified first. These themes should cover multiple aspects, from verifying operational concepts to supporting operational training, to evaluating equipment effectiveness and testing innovative tactics. At the same time, the theme should be refined according to the needs, such as selecting different types of wars such as mountain offensive and defensive warfare, urban street fighting, and maritime blockade warfare. The determination of the theme will provide a clear guiding direction for the subsequent scenario design.

[0086] Step 2: Set up game roles

[0087] On the basis of the theme, the game roles are set. The game roles can be human agents, machine agents, or human-machine intelligence hybrids. Each game role must have unique characteristics such as intelligence, decision-making preferences, role responsibilities and permissions to simulate the specific performance of various participants in the actual battlefield.

[0088] Step 3: Build a gaming environment

[0089] Next, the intensity of the game and the specific time and space background are set according to the deduction requirements, providing a clear action framework for participants and agents. The setting of the environment will dominate the formulation of game rules and the final development direction of the game results.

[0090] Step 4: Develop detailed rules

[0091] According to the simulation objectives and scenario characteristics, detailed combat rules, victory and defeat criteria, and constraints in the game process are formulated. These rules provide a basis for decision-making for the game characters and ensure the rationality and effectiveness of the simulation scenario.

[0092] Step 5: Set the initial posture

[0093] After the rules are formulated, the situation is set. This includes determining the initial situation of intelligence data, command relations, force deployment, etc. in the game scenario based on the theme, role, and environment. These initial situations will serve as the starting point for the game.

[0094] Step 6: Verification and test scenario design

[0095] Finally, the designed scenario is verified and tested. The rationality and effectiveness of the scenario design are verified through simulation system simulation or actual human-machine confrontation experiments. During the verification and testing process, the performance data and feedback of the human-machine agent are collected, and the scenario is continuously optimized and improved to ensure that the needs and goals of the war game are met.

[0096] By sequentially carrying out the above six steps, the present invention realizes the design of human-machine intelligent game confrontation scenarios in the field of war game simulation. This method not only standardizes the various elements of the game scenario, but also designs the organization and application of the game elements, and finally forms a combat scenario that meets the needs of the designer. This design method helps to improve the decision-making ability, adaptability and strategic thinking of human-machine intelligent bodies, and provides more realistic and comprehensive support for military training and decision-making analysis.

[0097] The method specifically includes:

[0098] Step 1: Scene Theme Setting

[0099] The setting of scenario theme is the starting point of scenario design, and it plays a vital role in guiding the entire design process. These themes cover multiple aspects, from verifying operational concepts to supporting operational training, to evaluating equipment effectiveness and testing innovative tactics, involving operational elements such as people and their organizational methods, weapons and equipment, and operational thinking. Each theme carries specific goals and expectations.

[0100] The topics of human-machine intelligent game system scenarios include but are not limited to verifying combat concepts, supporting combat training, evaluating equipment effectiveness, and innovating combat methods.

[0101] Theme 1: Validating operational concepts

[0102] Under the theme of verifying the operational concept, the main purpose of designing a scenario is to test the feasibility and effectiveness of a new operational concept in actual use. This requires us to fully consider the characteristics and requirements of the operational concept in the scenario design and simulate the corresponding operational environment and conditions so as to truly reflect the actual effect of the operational concept.

[0103] Theme 2: Supporting combat training

[0104] The theme of combat support training focuses more on improving the actual combat capabilities of participants. By designing scenarios close to actual combat, participants can conduct actual combat exercises in a simulated combat environment, thereby enhancing their actual combat experience and response capabilities. This requires us to focus on authenticity and complexity in scenario design, and simulate various possible situations in actual combat as much as possible.

[0105] Theme 3: Evaluating Equipment Effectiveness

[0106] The theme of evaluating equipment effectiveness is to evaluate the performance of a certain type of equipment or a certain type of equipment in actual combat. By designing a specific combat scenario and observing the performance of the equipment in the scenario, the performance evaluation results can be obtained. This requires that the characteristics of the equipment and the use environment be fully considered in the scenario design so that its performance can be accurately evaluated.

[0107] Theme 4: Testing innovative combat methods

[0108] Innovative combat methods are also of great significance for improving the combat effectiveness of the army, optimizing combat effectiveness, and coping with complex and changing battlefield environments. The human-machine intelligent game system can break through the traditional mindset by simulating and analyzing a large amount of combat data, examine and solve combat problems from a new perspective and ideas, and dig out potential combat laws and trends. By setting a simulation environment close to the real military struggle scene, setting the required game resources, and appropriately adding pre-set conditions, the game system can play games autonomously with minimal or no intervention, explore efficient tactics in specific environments, and open up new combat methods. Provide new ideas and strategies for combat commanders.

[0109] These themes not only provide a clear direction for scene design, but also provide a foundation for subsequent links. After clarifying the scene theme, we can carry out subsequent work such as role setting, environment setting, situation setting, and rule setting according to the requirements and characteristics of the theme. The work of these links is all centered around the scene theme, and is all for better achieving the goals and expectations set by the scene theme.

[0110] Step 2: Role Setting

[0111] The role setting link plays a pivotal role in the scene design of the human-machine intelligent game system. This link is related to the specific characteristics of various participants in the system. By carefully setting the roles, a more realistic and challenging game scene can be constructed, thus providing strong support for the subsequent design of the human-machine intelligent game system.

[0112] S101, select the game role type:

[0113] Human agent: An agent played by humans, whose experience level and cognitive level are determined by real people. The decision-making authority of human agents is superior to that of machine agents.

[0114] Machine agent: an AI agent or multiple AI composite agents.

[0115] Human-machine intelligence hybrid: a composite agent consisting of a human and one or more AI agents (systems).

[0116] S101.1, select the above three types of intelligence to form two or more parties in the game. For example:

[0117] Human-machine-machine: One side is a human-machine intelligence hybrid, and the other side is a machine intelligence, and the two sides engage in a game.

[0118] Machine-machine-machine: The machine agents each represent three interest groups in the game.

[0119] S102, assigning machine agent attributes:

[0120] S102.1, Setting the degree of intelligence: According to the importance of the machine agent in the game, different levels of intelligence are set to formulate strategies from basic to advanced. For example, decision-making AI is responsible for global strategic planning and decision-making, learning AI is responsible for local data collection and analysis to optimize decision-making, and reactive AI is responsible for autonomously performing corresponding game tasks within its own specific game rules according to the preset requirements of advanced AI.

[0121] Specifically:

[0122] (1) Setting method of decision-making AI

[0123] The focus of decision-making AI is on global strategic planning and decision-making. This type of AI has a higher degree of intelligence because its decisions have a direct impact on the overall outcome of the game.

[0124] (1.1) Define goals and tasks

[0125] Set strategic objectives: Define high-level strategic objectives to achieve specific campaign goals, maximize resource utilization efficiency, or achieve favorable development of the battlefield situation under resource constraints.

[0126] Task decomposition: Break down strategic objectives into a series of executable subtasks, such as close reconnaissance, consolidation of defense, launching an attack, resource collection and management.

[0127] (1.2) Design algorithm

[0128] Design decision-making algorithms: Combine short-term tactical goals with long-term strategic goals to make comprehensive decisions, such as concentrating firepower to defeat enemy forces in the short term and ensuring that the battlefield situation tends to be favorable in the long term. The decision-making algorithm operates in a dynamic battlefield environment, traverses the decision tree under the current situation, and finds the optimal strategy to achieve short-term tactical goals through trial and error. On this basis, a situation assessment is conducted to evaluate the changes in the battlefield situation after the implementation of this tactical goal optimal strategy. The decision-making strategy is comprehensively considered based on the contribution of its execution results to the long-term strategic goals, so as to achieve the organic coordination of short-term tactical goals and long-term strategic goals.

[0129] Design a decision-making algorithm evolution mechanism: Use the state-action-reward model to define the state, action, and corresponding reward in the decision-making process to guide AI strategy optimization. Adjust strategies in real time based on immediate feedback such as battle results and resource changes. At the same time, optimize and upgrade long-term strategies based on long-term feedback such as previous decision results and game results. Solve the optimization problem of complex decision-making algorithms based on this, modify the decision algorithm by modifying the decision tree and iterative optimization, and realize the evolution of the decision algorithm.

[0130] (1.3) Establish a knowledge base and rule system

[0131] Build a knowledge base: Use past game records to build a knowledge base covering enemy strategies, resource distribution, geographic information, etc.

[0132] Design a rule system: Set a series of decision-making rules and priorities based on experience and domain knowledge, such as adopting different tactics when facing different enemies and giving priority to protecting core resource points.

[0133] (1.4) Decision making based on subtask division

[0134] High-level decision-making: Develop and modify overall strategies and long-term plans in real time, such as mission determination, regional expansion direction, and sub-task deadlines.

[0135] Mid-level decision-making: Based on the strategic goals determined by high-level decisions, sub-tasks are divided, resources are allocated, and specific mid-term tasks are formulated, such as troop deployment and resource collection.

[0136] Low-level decision-making: Responsible for executing specific actions of sub-tasks and actual use of resources, such as troop movements, specific attack and defense measures, based on the force deployment, game conditions and resource allocation determined by the middle-level decisions.

[0137] (2) Setting method of learning AI

[0138] Learning AI is mainly responsible for local data collection and analysis to optimize decision-making. Its intelligence is mainly reflected in data processing and pattern recognition.

[0139] (2.1) Define learning objectives

[0140] The learning content focuses on two major contents: pattern recognition and dynamic environmental adaptation. Pattern recognition is used to identify and predict enemy behavior patterns, intention trends, etc. Dynamic environmental adaptation is used to adapt to environmental changes, such as terrain changes, weather conditions, etc. At the same time, two evaluation indicators, prediction accuracy and response time, are set to measure the degree of AI's proximity to the learning goal. Among them, the prediction accuracy measures the accuracy of AI's prediction of behavior or intention, and the response time evaluates the speed and efficiency of AI data analysis and pattern recognition.

[0141] (2.2) Choosing a learning method

[0142] In terms of algorithm selection, since learning AI is mainly used for local data processing, it is usually decision-making AI and reactive AI. It usually does not need to collect data autonomously, but only needs to process input data. Therefore, it mainly relies on machine learning algorithms for data processing. The following learning algorithms can be selected according to the type of data:

[0143] Supervised learning: In this method, the model is trained by using labeled data, i.e. each input data has a corresponding output target. Examples include neural networks, support vector machines (SVM).

[0144] Unsupervised learning: Pattern discovery on unlabeled data, the model needs to automatically discover patterns, structures or relationships from the input data, such as K-means clustering.

[0145] Semi-supervised Learning: Combining labeled and unlabeled data, using unlabeled data to assist model training, improving model performance and generalization ability, and is used when data is scarce and difficult to obtain.

[0146] In addition, an online learning mechanism is introduced: through the online learning mechanism, feedback from actual operation is collected to update and improve the algorithm in real time to adapt to new environments and new situations.

[0147] (2.3) Constructing the dataset

[0148] Perform data cleaning: First, fill in or delete missing parts in the data. Second, standardize or normalize the data to improve data compatibility.

[0149] Perform feature extraction: Use algorithms to automatically identify and extract important features or combine domain knowledge to manually select key features to reduce data processing dimensions.

[0150] Build datasets: Collect and build large datasets for training and validation from data input sources.

[0151] (2.4) Training and validating the model

[0152] Train the model: Use the constructed dataset to train the model and adjust the model parameters to optimize performance.

[0153] Verify the model: First, evaluate the model through indicators such as accuracy, recall rate, and F1 score. Second, evaluate the generalization ability of the model through cross-validation to prevent overfitting. Finally, continuously iterate and adjust the model parameters to improve its application effect in practical applications.

[0154] (2.5) Deployment and iterative optimization

[0155] Deploy the model and deploy the trained model in the actual game environment for real-time data analysis and decision support. Integrate learning AI into the overall decision-making system and work with other AI.

[0156] (3) How to set up reactive AI

[0157] Reactive AI is responsible for autonomously performing specific game tasks under specific game rules. Its intelligence level is mainly reflected in its ability to execute preset conditions and commands.

[0158] (3.1) Define reaction rules

[0159] Operation scope: defines the types of operations that the reactive AI can perform, such as attack, defend, move, trade, etc.

[0160] Trigger conditions: clearly define under what conditions which actions are triggered, such as automatically counterattacking when the enemy enters the attack range.

[0161] (3.2) Matching state-action pairs

[0162] Set state: Use state machines to define different states and their transition conditions, such as alert, attack, retreat, etc. At the same time, construct a behavior tree, using a hierarchical tree structure to represent complex behaviors and action combinations.

[0163] Execute actions: define executable actions and their specific effects, such as attack, move, dodge, etc. And set execution conditions so that reactive AI can select appropriate actions to execute based on the current state and environmental conditions.

[0164] (3.3) Adjustment and optimization

[0165] Dynamic adjustment: Adjust operating rules and priorities in real time based on actual conditions and feedback.

[0166] Test optimization: Regularly conduct various stress tests and scenario simulations to monitor the performance of reactive AI, and continuously optimize the response strategy through trial and error and reward mechanisms.

[0167] S102.2, Set decision preferences: Set a unique decision-making style for each machine agent role, such as adventurous machine agents tend to maximize future benefits, rational machine agents consider both benefits and risks to maximize comprehensive benefits, conservative machine agents consider optimizing benefits under current game conditions, etc. Or it may be consistent with special combat ideas.

[0168] Specifically:

[0169] (1) Method for setting up adventurous machine agents

[0170] Adventurous machine agents tend to pursue high-risk, high-return strategies and strive to maximize future benefits even in the face of greater risks.

[0171] Build a risk assessment model: Build a risk assessment model to quantify the risk level of different decisions, but at the same time ensure that the model has a high tolerance for high-risk strategies.

[0172] Establish a reward mechanism: Design a reward mechanism that favors future rewards. The reward mechanism needs to give high weight to the long-term rewards of high-risk actions.

[0173] Try and evaluate strategies: Enable AI to explore multiple strategy paths simultaneously, evaluate potential returns under different risk strategies, and optimize selection opportunities.

[0174] Controlling risks: Although high-risk strategies are preferred, certain risk management mechanisms need to be embedded to avoid disastrous consequences caused by excessive risk-taking.

[0175] (2) Method for setting up intelligent machine agents

[0176] Rational machine agents will comprehensively consider benefits and risks when making decisions, and pursue long-term and stable profit maximization.

[0177] Establish a risk assessment system: Introduce a multi-dimensional assessment system that considers both potential benefits and risks, using methods such as Bayesian decision theory that involve probability quantification and prediction.

[0178] Adjust preference balance parameters: adjust the balance point between algorithm benefits and risks, and set a rational preference parameter to maximize benefits while controlling risks.

[0179] Conduct scenario simulation: Intensively use historical data and current data analysis to improve the accuracy and rationality of decision-making. And through scenario simulation or scenario analysis, predict decision results and evaluate the possible consequences of various decision options in different scenarios.

[0180] Continuous feedback adjustment: Introducing a continuous feedback mechanism allows AI to learn from the results of each decision and dynamically adjust the decision-making strategy to enhance the benefits of future decisions.

[0181] (3) Conservative Machine Agent Setting Method

[0182] Conservative machine agents focus on optimizing benefits under current game conditions and tend to adopt low-risk strategies to ensure stable returns.

[0183] Establish a risk warning system: Establish a risk warning system to identify and evaluate possible risks at an early stage and take preventive measures in time to mitigate potential losses.

[0184] Establish a risk avoidance mechanism: Design an incentive mechanism to reward decisions with low risk and stable returns, and ensure that the strategy is biased towards avoiding large risks through risk weight adjustment. At the same time, select or develop algorithms that tend to act robustly, such as decision trees, specific implementations of genetic algorithms, etc., with the goal of stable growth.

[0185] Prepare buffer strategies: simulate the results of games under extremely unfavorable conditions. Prepare buffer strategies for possible losses based on the results of games to ensure that a certain strategic advantage can be maintained even in unfavorable situations.

[0186] S102.3, Set the depth of decision-making hierarchy: Determine the level of thinking of the machine agent role in the decision-making process. The depth and breadth of the machine agent role's consideration of issues in the decision-making process affect the strategic planning and tactical application of the machine agent role, and distinguish from simple reactions to complex strategic decision-making planning. Advanced machine agents rely on cognition, experience, and theory to make decisions. Low-level machine agents rely on experience and partial cognition and theory to make decisions.

[0187] In order to better understand and distinguish the different levels of machine intelligence and their corresponding decision-making criteria, the decision-making level can be divided into three levels: ordinary level I (low-level decision-making), standard level II (middle-level decision-making), and advanced level III (high-level decision-making).

[0188] (1) Ordinary Level I (low-level decision-making)

[0189] Ordinary Level I machine agents are mainly responsible for executing specific actions and operations, and responding in real time and executing tasks based on the force deployment, game conditions and resource allocation made by the middle-level decision-making. This level of agents needs to respond quickly and execute efficiently to ensure that the middle-level decisions are effectively implemented.

[0190] (1.1) Specific tasks

[0191] Force mobilization: Carry out reasonable troop movements according to orders and existing conditions to ensure tactical flexibility and responsiveness.

[0192] Specific combat actions: Execute specific combat actions, including selecting offensive targets, defensive strategies, etc., to ensure combat effectiveness and real-time response.

[0193] (1.2) Decision characteristics

[0194] Reliance on experience and rules: Ordinary Level I agents are mostly based on preset rules and simple empirical models, such as state-action pairs.

[0195] Fast response speed: Since the tasks are relatively specific and clear, ordinary Level I agents need to react and execute in a very short time.

[0196] Operational level decision-making: Tasks are focused on the operational level, executing predetermined strategies and specific instructions to ensure the accuracy and timeliness of underlying behaviors.

[0197] (2) Standard Level II (Middle-level decision-making)

[0198] Standard Level II machine agents are responsible for mid-term task planning and resource allocation. According to the strategic goals determined by high-level decisions, they break down large tasks into several subtasks and effectively allocate resources to achieve these tasks. They need to weigh multiple factors in the decision-making process, including time, resources, geographic location, etc., to ensure the smooth achievement of strategic goals.

[0199] (2.1) Specific tasks

[0200] Force deployment: Develop a specific force deployment plan to ensure that each tactical unit is in the best position to achieve maximum effectiveness.

[0201] Resource Collection: Manage and optimize the collection and allocation of resources to ensure their effective use and support the implementation of tactical operations.

[0202] (2.2) Decision characteristics

[0203] Comprehensive analysis capability: Standard Level II agents are able to comprehensively analyze multiple sources of information and develop forward-looking medium-term plans.

[0204] Multi-tasking: Effectively manage and schedule multiple tasks to ensure coherence and consistency of tactical objectives.

[0205] Tactical-level decision-making: Tasks are concentrated at the tactical level, involving more complex decision-making scenarios and multi-dimensional resource scheduling.

[0206] (3) Senior Level III (high-level decision-making)

[0207] Advanced Level III machine agents are responsible for formulating overall strategies and long-term plans, including determining mission objectives, regional expansion directions, and subtask deadlines. Agents at this level need to have comprehensive cognitive and in-depth analytical capabilities to make global strategic decisions.

[0208] (3.1) Specific tasks

[0209] Action decision determination: formulate overall mission objectives and strategic directions, such as determining the main attack areas and overall action strategies.

[0210] Regional expansion direction: Analyze the current and future environment, formulate regional expansion strategies, and ensure the maximization of global interests.

[0211] Subtask deadline setting: Set the completion time node for each subtask to ensure task time management and coordinated operations.

[0212] (3.2) Decision characteristics

[0213] Global cognition: Advanced Level III agents have deep environmental perception and cognitive capabilities, and can understand complex situations and predict future developments.

[0214] Strategic Planning: Ability to conduct complex strategic planning, including multi-objective optimization and the development and adjustment of long-term plans.

[0215] High-level decision-making: The task involves global strategic decision-making and macro-task planning, covering the optimization and coordination of the entire system.

[0216] S103, allocating gaming roles and gaming seats:

[0217] S103.1, Clarify role responsibilities: Assign clear responsibilities to each role, such as staff officer, commander, etc., to ensure that they have clear goals and directions of action during the game. Assign different tasks to different roles, such as reconnaissance, strike, support, etc., to form a joint force for coordinated operations.

[0218] S103.2, Clarify role relationships: Clarify the relationships between roles, including command relationships, collaboration relationships, affiliation relationships, business guidance relationships, and enemy-friend relationships. By building reasonable role relationships, simulating complex interpersonal relationships in the real world, and building rich game dynamics, the game process is closer to reality. For example, we need to clarify the command relationship between commanders at all levels to ensure that orders can be smoothly conveyed and executed; we need to consider the collaboration methods and tacit understanding between different roles to achieve common combat goals.

[0219] S103.3, clarify role authority: set the authority and restrictions of roles in the game process. Authority includes information access, decision execution, resource allocation, etc., as well as force mobilization restrictions, time limits, etc. For example: the decision authority of the superior machine agent is superior to the decision authority of the subordinate machine agent (the same applies to human agents); the decision authority of human agents is always higher than the decision authority of machine agents.

[0220] Step 3: Environment Setting

[0221] The environmental setting phase is a key step in building a human-machine intelligent game scenario. It describes the background of the game in detail, which is equivalent to characterizing the game and determining the upper limit of the development of the game results. Through this phase, we have clarified the intensity of the game and the specific time and space background in which it occurs, and then determined the strategy of force deployment, the upper limit of resource use, and the decision-making will of the leader in the game. The environmental setting basically dominates the setting of the game rules and the final development direction of the game results, and constructs a specific and realistic scenario for the entire game process, enabling participants and agents to conduct subsequent game phases in a more realistic situation, and also helps to test the adaptability and response strategies of agents in different environments.

[0222] S201, set the game intensity:

[0223] The game intensity represents the tension and conflict level of the entire combat environment, and is set to different levels such as invisible game level, specific tension level, local or total war level, etc. according to different needs. For example: the game environment at the invisible game level tends to be an implicit game of deterrence in the cognitive domain or information domain, leading to changes in the rules of enemy and friendly organizational relations; the game environment at the specific tension level tends to be a containment strategy of blockade nature in the information domain and physical domain, leading to the concentration of troops and resources in specific areas; and the game environment at the war level focuses more on firepower confrontation and conflict in the physical domain, leading to changes in the rules of the game.

[0224] S202, setting the game dimension:

[0225] The environment setting also needs to consider the game dimension. The determination of the game dimension determines the subsequent game rules that may be applicable to these combat fields. The game dimension includes the physical domain, cognitive domain, and information domain, and each domain will have an impact on the game process and results. By dividing and setting these combat domains, the scope and boundaries of the game can be more clearly defined, providing participants and agents with a clear action framework and decision-making basis. A diversified and comprehensive combat environment can be constructed, making the game process more complex and rich.

[0226] S202.1, Setting the physical domain: Consider the specific space and environmental conditions where the game takes place, including physical factors such as terrain, weather, and facilities. The physical domain setting directly affects the deployment of forces, routes of action, and tactical choices.

[0227] S202.2, Setting cognitive domain: Focus on the thinking, judgment and decision-making process of participants and agents. It involves multiple aspects such as the psychological state, cultural knowledge background, and experience accumulation of participants. The setting of cognitive domain will directly affect the decision-making quality and game strategy of participants.

[0228] S202.3, Setting up information domain: This refers to the field of information transmission, processing and utilization in the game. The setting up of information domain will directly affect the organization and command, combat planning and resource scheduling.

[0229] S202.4, Setting time and space constraints: This involves the specific time and space scope of the game, including combat time, battlefield scope, action time limit, etc. The setting of time and space constraints will directly affect the rhythm and progress of the game, and provide participants and agents with a clear action framework and decision-making basis.

[0230] Step 4: Rule Setting

[0231] In the scenario design of the human-machine intelligent game system, the rule setting link provides participants and intelligent agents with clear action guidelines and decision-making basis. By simulating the rules and constraints of the real world, the real-world situation can be simulated for the intelligent game system. Different combat fields require different game rules. After scientific abstraction of various military confrontation rules, the intelligent game system can simulate the characteristics of military struggles in different combat fields in the real world. In addition, it is supplemented by additional rules that are more in line with real scenarios, such as resource scheduling rules and enemy-friend organizational relationship rules, to enrich the details of the intelligent game system and restore the constraints of the real game scenario to the maximum extent. Finally, through the application of system engineering rules, a comprehensive analysis of the components, organizational structure, information flow, control mechanism, etc. of the game system is carried out, so that the relationship between the various parts of the system is coordinated and coordinated with each other, and finally the system error is eliminated, the real scenario is matched, and the overall optimal operation of the game system is achieved.

[0232] Military confrontation rules, resource scheduling rules, enemy and friendly organizational relationship rules, and system engineering rules constitute the core content of rule setting, which provides the intelligent game system with a clear decision-making framework as well as the explainability and transparency of the decision-making process, ensures the fairness and consistency of the game, and jointly shapes the operating mechanism of the game system.

[0233] S301, setting military confrontation rules: formulate military action norms so that participants and agents can play games in accordance with military laws, including activity space restrictions, fire damage effects, weapon equipment performance, etc. These rules cover various combat types such as attack, defense, reconnaissance, counterattack, and action rules in various combat fields such as land, sea, air, space, network, and electricity, to ensure the authenticity and rationality of the game process. For example: physical domain land march rules, line of sight rules, unit gathering and dispersion rules, fire confrontation rules, etc.; information domain electromagnetic interference rules, intelligence theft confusion rules, etc.; cognitive domain public opinion war rules, psychological intervention rules, morale assessment rules, cognitive attack judgment rules, etc.

[0234] S302, set resource scheduling rules: clarify the acquisition methods, allocation principles, dynamic adjustments and usage restrictions of game resources such as manpower, materials, equipment, including intelligence, data, computing power, and algorithms, to simulate the principle of limited resources in the game process. For example, after the intelligent agent is captured, the intelligence data and computing power resources it possesses can be used by the capturer to promote the evolution of the intelligent agent, thereby simulating the impact of resource changes on the outcome of the game.

[0235] S303, Setting rules for friendly and enemy organizational relationships: Define the alliance, hostility or neutral relationship and organizational structure between game characters, provide a basis for teamwork and confrontation during the game, and affect the dynamics of the game. For example, the rules may stipulate the joint defense relationship between the parties, the mode of cooperation, the confrontation strategy, and the conditions for the capture and surrender of game characters.

[0236] S304, set system engineering rules: specify the comprehensive effects between roles and between roles and the environment. Analyze the game system as a whole, comprehensively consider the mutual connection and mutual restriction between each part, analyze their impact on the system and optimize them to eliminate deviations, and achieve the authenticity and accuracy of the overall game performance. The setting of system engineering rules helps to improve the overall coordination of the game, making the game results more in line with the actual situation.

[0237] Step 5: Situation Setting

[0238] In the intelligent game system, situation setting is a crucial link, which involves the allocation of game resources, the quantitative setting of game rules, and the initial adaptation of game scenarios. This is not only the starting point of the game, but also a key factor affecting the subsequent game and the final result. Situation setting mainly focuses on command relations, force organization, and resource allocation, which together constitute the skeleton of the combat situation and determine the formation and development direction of the situation. By allocating intelligence resources to directly form a credible decision-making basis to replace the two capabilities of observation (Observe) and judgment (Orient) in the OODA (Observe Orient Decide Act) loop, the organization and command resources and force deployment resources give the system decision-making (Decide) and action (Act) capabilities respectively.

[0239] S401, Allocate intelligence resources: Allocate current intelligence resources and set intelligence collection, analysis and processing capabilities according to role requirements. Understand the current enemy and our situation, form a credible basis for decision-making, and give the game role the ability to perceive, so as to support the subsequent independent decision-making basis.

[0240] S402, Organization and Command Resources: Set the organizational structure, chain of command and communication methods between roles. The command relationship represents the power and responsibility relationship between roles, which determines the information transmission path, the execution method of decision-making and the effect of coordinated operations. According to the characteristics and needs of the game system, set a reasonable organizational structure, clarify the responsibilities and powers of commanders at all levels, establish a chain of command and communication mechanism, enable various roles to work together, and enable game roles to have independent decision-making capabilities.

[0241] S403, deploying military resources: deploying military forces, equipment and material resources according to the game strategy and combat domain. Determine the scale and structure of the military force, clarify the configuration and use rules of the equipment and the corresponding material support rules. Involve multiple aspects such as the type, number and combination of participants. According to different combat needs and strategies, match human-machine intelligent bodies, weapons and equipment, personnel and other resources to give the game characters the ability to act independently.

[0242] Step 6: Verification and Testing

[0243] Verification and testing are key steps to ensure the quality of scene design, improve functions, and enhance user experience. The first purpose is to test whether there are parameter distortions and rule loopholes within the scene. The second is to test the impact of the scene on the game system. Through simulation tests, user feedback, performance evaluation and other methods, the verification and testing phase conducts a comprehensive and in-depth examination of the rationality and effectiveness of the scene design, providing a valuable basis for subsequent optimization and improvement.

[0244] S501, simulation test: Use the simulation platform to simulate the scenario and verify the rationality and effectiveness of the scenario design. Set the corresponding test scenario on the simulation platform. By simulating the actual game process, observe the operation of the system in different scenarios and detect whether there are potential problems and loopholes.

[0245] S502, User Feedback: Invite experts and users to participate in the test and collect feedback. Select representative experts and potential users to form a test user group. Organize experts and users to conduct actual operations and experience, and collect their opinions and suggestions on scene design through questionnaires, interviews, etc.

[0246] S503, Performance Evaluation: Evaluate the performance of the gaming system in different scenarios. Determine key performance evaluation indicators, such as operating efficiency, response time, and processing speed. Test the operating performance of the system in the scenario and collect data. Analyze the collected performance data to determine whether the scenario design meets the performance requirements, and whether there are performance bottlenecks or resource waste.

[0247] S504, scenario optimization: Optimize and improve the scenario. Combine simulation test results, user feedback and performance evaluation reports to find out the problems and deficiencies in the scenario design. Develop a specific optimization plan, including adjusting the scenario settings, optimizing game rules, etc. Repeat S501, S502, and S503 again to verify whether the optimized scenario design achieves the expected effect, until the scenario design meets the requirements.

[0248] In the specific implementation process, various aspects of the scene design can be flexibly adjusted and expanded in combination with actual conditions and needs. For example, more tactical actions and strategic backgrounds can be introduced to increase the complexity and diversity of the scene; advanced computer technology and artificial intelligence technology can also be used to achieve automatic generation and intelligent optimization of the scene. These extensions and applications will further enhance the practicality and innovation of the present invention.

[0249] The above is only a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation, characterized in that: include: Step 1: Establish the scenario theme based on the specific needs and goals of the war game, and refine the scenario theme based on the needs; Step 2: On the basis of the scene theme, the game roles are set, and the game roles include human agents, machine agents, and human-machine intelligence hybrids; the human agent is an agent played by humans, the machine agent is an AI agent or multiple AI composite agents, and the human-machine intelligence hybrid is a composite agent composed of humans and one or more AI agents; Step 3: Set the intensity of the game and the specific time and space background according to the needs of the game, complete the game environment setting, and provide an action framework for the game characters; Step 4: According to the simulation objectives and scenario characteristics, formulate detailed combat rules, victory and defeat criteria, and constraints in the game process to provide decision-making basis for game players; Step 5: Determine the initial situation in the simulation scenario based on the scenario theme, game roles, and simulation environment settings. The initial situation includes intelligence data, command relationships, and force deployment, and the initial situation is used as the starting point for the simulation. Step 6: Verify and test the designed human-machine intelligent game confrontation scenario. Through simulation system simulation or actual human-machine confrontation experiments, verify the rationality and effectiveness of the scenario design. During the verification and testing process, collect performance data and feedback from the human-machine intelligent agent, optimize and improve the scenario, and ensure that it meets the needs and goals of the war game simulation.

2. According to claim 1, a method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation, characterized in that: Scenario themes include validating operational concepts, supporting operational training, evaluating equipment effectiveness, and testing innovative operational methods.

3. The method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation according to claim 1, characterized in that: The step 2 also includes: setting different levels of intelligence according to the importance of the machine agent in the game role to formulate strategies from basic to advanced.

4. The method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation according to claim 3 is characterized in that: The machine intelligent agent includes decision-making AI, learning AI and reactive AI. Decision-making AI is responsible for global strategic planning and decision-making, learning AI is responsible for local collection and analysis of data to optimize decisions, and reactive AI is responsible for autonomously performing corresponding game tasks according to its own specific game rules.

5. The method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation according to claim 4, characterized in that: Set up the decision-making AI as follows: 1) Define goals and tasks Set strategic goals and break them down into a series of executable subtasks; 2) Design Algorithm Design decision-making algorithms: Combine short-term tactical goals with long-term strategic goals to make comprehensive decisions; the decision-making algorithm operates in a dynamic battlefield environment, traverses the decision tree under the current situation, and finds the optimal strategy to achieve short-term tactical goals through trial and error; on this basis, conduct situation assessment, evaluate the changes in the battlefield situation after executing this optimal strategy for tactical goals, and comprehensively consider this decision-making strategy based on the contribution of its execution results to long-term strategic goals, so as to achieve the organic coordination of short-term tactical goals and long-term strategic goals; Design a decision-making algorithm evolution mechanism: Use the state-action-reward model to define the state, action, and corresponding reward in the decision-making process to guide the optimization of decision-making AI strategies; adjust strategies in real time based on instant feedback from battle results and resource changes; at the same time, optimize and upgrade long-term strategies based on previous decision-making results and long-term feedback from game results; modify the decision algorithm by modifying the decision tree and iterative optimization to achieve decision algorithm evolution; 3) Establish a knowledge base and rule system Use past game records to build a knowledge base and set decision rules and priorities based on experience and domain knowledge; 4) Make decisions based on subtasks High-level decision making: developing and modifying overall strategies and long-term plans in real time; Mid-level decision-making: Based on the strategic goals determined by high-level decisions, sub-tasks are divided, resources are allocated, and specific mid-term tasks are formulated; Low-level decision-making: Specific actions and actual use of resources for executing subtasks based on mid-term tasks determined by mid-level decisions.

6. The method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation according to claim 4, characterized in that: Set up the learning AI as follows: 1) Define learning objectives Define learning content, including pattern recognition and dynamic adaptation to the environment; set prediction accuracy and response time to measure how close the learning AI is to the learning goal; 2) Choose a learning method Choose a learning algorithm based on the type of data: supervised learning, unsupervised learning, or semi-supervised learning; At the same time, an online learning mechanism is introduced to collect feedback from actual operation and update and improve the algorithm in real time; 3) Build a dataset Perform data cleaning, automatically identify and extract important features, reduce data processing dimensions, collect and construct datasets for training and validation from data input sources; 4) Training and validating the model Use the constructed data set to train the model and adjust the model parameters to optimize performance; verify the model and continuously iterate and adjust the model parameters; 5) Deployment and iterative optimization Deploy trained models in actual game environments, conduct real-time data analysis and decision support, and integrate learning AI into the overall decision-making system.

7. The method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation according to claim 4, characterized in that: Set up reactive AI as follows: 1) Define reaction rules Define the types of actions that reactive AI will perform and clarify the triggering conditions for the actions; 2) Matching state-action pairs Use state machines to define different states and their transition conditions, and construct behavior trees to represent behaviors and action combinations using a hierarchical tree structure. Define the actions to be executed and their specific effects, and set the execution conditions so that the reactive AI can select the appropriate actions to execute based on the current state and environmental conditions; 3) Adjustment and optimization Adjust operating rules and priorities in real time based on actual conditions and feedback, conduct various stress tests and scenario simulations regularly, monitor the performance of reactive AI, and continuously optimize response strategies through trial and error and reward mechanisms.

8. The method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation according to claim 1, characterized in that: The step 2 further includes: setting a unique decision-making style for each machine agent; according to different decision-making styles, the machine agents include adventurous machine agents, rational machine agents and conservative machine agents; Set up the adventurous machine agent as follows: Build risk assessment models to quantify the risk level of different decisions, but at the same time ensure that the risk assessment model has a high tolerance for high-risk strategies; design a reward mechanism that prefers future returns and gives high weight to the long-term returns of high-risk actions; enable AI agents to explore multiple strategy paths simultaneously, evaluate potential returns under different risk strategies, and optimize selection opportunities; and embed risk management mechanisms; Set up the Intelligent Machine Agent as follows: Establish a risk assessment system that considers both potential benefits and risks; adjust the balance between algorithm benefits and risks and set preference parameters to maximize benefits while controlling risks; use historical and current data for analysis; and predict decision outcomes through scenario simulation or scenario analysis, and evaluate the possible consequences of various decision options in different scenarios; introduce a continuous feedback mechanism so that AI agents can learn from each decision outcome and dynamically adjust decision strategies to enhance the benefits of future decisions; Set up the conservative machine agent as follows: Establish a risk warning system to identify and evaluate possible risks at an early stage; set up a risk avoidance mechanism to reward decisions with low risks and stable returns, and ensure that strategies tend to avoid major risks through risk weight adjustments; at the same time, select or develop algorithms that tend to act robustly with the goal of stable growth; simulate the results of games under extremely unfavorable conditions, and prepare buffer strategies for possible losses based on the results of games.

9. The method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation according to claim 1, characterized in that: The step 3 includes: setting the game intensity and setting the game dimension; the setting of the game dimension includes setting the physical domain, cognitive domain, information domain and time and space constraints; the physical domain includes terrain, weather and facilities; the cognitive domain includes the psychological state, cultural knowledge background and experience accumulation of the participants; the information domain includes the areas of information transmission, processing and utilization in the game; the time and space constraints include combat time, battlefield scope and action time limit.

10. The method for designing human-machine intelligent game confrontation scenarios in the field of war game simulation according to claim 1, characterized in that: The step 4 comprises: Setting the rules of military confrontation: Formulate military action norms so that the game players can play the game in accordance with military laws, including activity space restrictions, fire damage effects, and weapon equipment performance; Set resource scheduling rules: clarify the acquisition methods, allocation principles, dynamic adjustments and usage restrictions of manpower, materials and equipment, including game resources, to simulate the principle of limited resources in the game process; Setting rules for friendly and enemy organizational relationships: defining alliances, hostility or neutral relationships between game characters and their organizational structures, providing a basis for teamwork and confrontation during the game, and influencing the game dynamics; Set system engineering rules: stipulate the comprehensive effects between game roles and between game roles and the game environment, analyze the game system as a whole, and comprehensively consider the interrelationships and mutual constraints between various parts.

Citation Information

Cited By

  • Near-earth space game situation analysis method based on improved intelligent pig game theory

    CN119227975A

  • Multistage deduction interaction data processing method and system based on intelligent game

    CN119514697A

  • Intelligent strategy deduction method and device for machine self-game

    CN119886856A

  • Intelligent Strategy Deduction Method and Device for Machine Self-Game Theory

    CN119886856B

  • Wargame confrontation ability evaluation method and system

    CN120105675A