Multi-element large-scale system confrontation simulation method, system, equipment and medium

Abstract

The present invention relates to system confrontation simulation technology, and provides a multi-factor large-scale system confrontation simulation method, system, device and medium, the method comprising: constructing an initialization simulation model including multiple functional-level entities and deploying it to multiple computing nodes; obtaining the current state of each functional-level entity in the simulation process in real time, and evaluating the influence weight of each functional-level entity on the simulation result; according to the weight threshold and the influence weight of all functional-level entities, using a dynamic simplification strategy to simplify the current simulation model to obtain the next simulation model; according to the current simulation result, using a hybrid propulsion mechanism to dynamically obtain the simulation propulsion ratio of the next simulation model; updating the next simulation model according to the simulation propulsion ratio and performing simulation calculation. The method of the present invention can significantly improve the simulation efficiency, accuracy and adaptability by introducing a machine learning algorithm, a dynamic simplification strategy and a hybrid propulsion mechanism.

Description

Technical Field

[0001] The present invention belongs to the military and strategic fields, relates to system confrontation simulation technology, and specifically relates to a multi-element large-scale system confrontation simulation method, system, equipment and medium. Background Art

[0002] In modern military and other fields, large-scale system confrontation simulation has become an important research and analysis tool. By building complex simulation models, the behavior and interaction of multiple functional-level entities in confrontation scenarios can be simulated to evaluate the effects of different strategies and decisions. However, as the scale and complexity of simulations increase, traditional simulation methods face problems such as high consumption of computing resources, low simulation efficiency, and untimely model updates.

[0003] Existing confrontation simulation methods usually use static models and fixed simulation advancement ratios, which are difficult to adapt to dynamically changing confrontation scenarios. For example, in military simulations, the battlefield environment and the status of both sides may change at any time, and static models cannot reflect these changes in real time, resulting in a large deviation between the simulation results and the actual scenario. In addition, traditional simulation methods often lack dynamic evaluation of the weight of the influence of each functional level entity on the simulation results, resulting in the inability to effectively identify and simplify entities with less influence on the results during the simulation process, thereby wasting a lot of computing resources.

[0004] In order to solve the above problems, researchers have begun to explore dynamic simulation methods based on machine learning in recent years. These methods obtain the status of each functional level entity in the simulation process in real time, use machine learning algorithms to evaluate the weight of its influence on the simulation results, and dynamically simplify the simulation model according to the weight threshold. This method can not only improve the simulation efficiency, but also dynamically adjust the simulation advancement ratio according to the simulation results, so as to better adapt to complex confrontation scenarios.

[0005] However, existing dynamic simulation methods still have some shortcomings. First, existing methods usually only focus on the state evaluation of a single functional-level entity and lack a global perspective on the entire simulation system. Second, existing dynamic simplification strategies are often too simple and cannot effectively balance simulation accuracy and computational efficiency. Finally, existing simulation advancement mechanisms usually use a fixed advancement ratio, which is difficult to adapt to the needs of dynamic changes during the simulation process.

[0006] Therefore, there is an urgent need for a new multi-factor large-scale system confrontation simulation method that can evaluate the influence weights of entities at each functional level in real time, and dynamically adjust the simulation model and advancement ratio according to the weight threshold and simulation results, so as to improve the simulation efficiency and accuracy and better adapt to complex confrontation scenarios. Summary of the invention

[0007] In order to achieve the purpose of evaluating the influence weight of each functional level entity in real time, and dynamically adjusting the simulation model and advancement ratio according to the weight threshold and the simulation result, thereby improving the simulation efficiency and accuracy, and better adapting to the complex confrontation scene, in order to solve the shortcomings of the existing simulation method, the present invention discloses a multi-factor large-scale system confrontation simulation method, which comprises the following steps: S1. Construct an initialization simulation model including multiple functional-level entities according to the requirements of the confrontation scenario, and deploy the initialization simulation model to multiple computing nodes, wherein each computing node performs simulation calculation of at least one of the functional-level entities; S2. Acquire the current state of each functional-level entity in the simulation process in real time, and use a machine learning algorithm to evaluate the influence weight of each functional-level entity on the simulation result according to the current state; S4, according to the weight threshold and the influence weights of all the functional-level entities, adopt a dynamic simplification strategy to simplify the current simulation model to obtain the next simulation model; S5. According to the simulation result output by the current simulation model, a hybrid propulsion mechanism is used to dynamically obtain a simulation propulsion ratio of the next simulation model; S6. Update the next simulation model according to the simulation propulsion ratio and perform simulation calculation.

[0008] Furthermore, in step S1, an initialization simulation model including multiple functional level entities is constructed according to the requirements of the confrontation scenario, including: S11. Acquire the types and quantities of functional-level entities according to the requirements of the confrontation scenario, wherein the types include attacking entities, defending entities, commanding entities, sensing entities, and supporting entities; S12, defining the attributes of each of the functional level entities, and initializing each of the functional level entities, wherein the attributes include identity identification, state attributes, capability attributes, resource attributes, and location attributes; S13, configuring behavior rules for each of the initialized functional level entities, establishing the interaction logic between all the functional level entities to obtain a simulation model, wherein the behavior rules include state transition rules, interaction rules, decision rules and resource management rules; S14, initializing simulation parameters of the simulation model to obtain an initialized simulation model, wherein the simulation parameters include a simulation step size, a propulsion ratio, and an initial state.

[0009] Furthermore, in step S2, the current state of each functional level entity during the simulation process is obtained in real time, and the influence weight of each functional level entity on the simulation result is evaluated using a machine learning algorithm according to the current state, including: S21. After each simulation is completed, current state data of the functional-level entity is collected, where the current state data includes position, speed, resource state, and historical tasks; S22, extracting features from the current state data to construct a feature vector, wherein the features include the health state, attack capability, defense capability, and historical contribution of the functional-level entity; S23. Use historical simulation data to train a neural network model, input the feature vector into the neural network model, and obtain the influence weight of the functional level entity on the simulation result.

[0010] Furthermore, in an improved example of the above multi-factor large-scale system confrontation simulation method, the method further includes: S3. Build a dynamic simplification strategy, including: S31, given a weight threshold, defining the functional-level entity whose impact weight is less than the weight threshold as a failed entity; S32: Use a simplified model to replace the failed entities in the current simulation model to obtain the next simulation model, or aggregate all the failed entities with the same attributes in the current simulation model into an equivalent entity to obtain the next simulation model, and construct a dynamic simplification strategy.

[0011] Furthermore, step S3 further includes: S33, introducing an adversarial mechanism in the simulation process, respectively obtaining the simulation results of the current simulation model and the next simulation model, calculating the output difference of the two simulation results to evaluate and optimize the dynamic simplification strategy.

[0012] Furthermore, in step S5, according to the simulation result output by the current simulation model, a hybrid propulsion mechanism is used to dynamically obtain the simulation propulsion ratio of the next simulation model, including: S51, obtaining a current simulation result output by the current simulation model, wherein the simulation result includes an occurrence event and a computing node utilization rate; S52, obtaining the proportion of time events and status events in the occurred events; S53: According to the proportion and the computing node utilization rate, a hybrid advancement mechanism is adopted to obtain a simulation advancement ratio of the next simulation model.

[0013] Furthermore, in step S53, according to the proportion and the computing node utilization rate, a hybrid advancement mechanism is adopted to obtain the simulation advancement ratio of the next simulation model, including: S531: According to the proportion and the computing node utilization rate, the formula Calculate the acceleration factor, where the proportion is the ratio of the time event or the state event in all events, and the weight is the influence of the proportion of different time events or state events on the acceleration factor, and the value range is [0,1], 0 means ignoring the proportion of the state event, and 1 means fully considering the proportion of the state event; S532, setting a reference propulsion ratio, and calculating the product of the reference propulsion ratio and the acceleration factor to obtain the simulation propulsion ratio of the next simulation model.

[0014] An embodiment of the present invention also provides a multi-factor large-scale system confrontation simulation system, including a simulation model construction module, an impact weight evaluation module, a model simplification module, a simulation propulsion ratio calculation module and a simulation engine module.

[0015] The simulation model construction module is used to construct an initialization simulation model including multiple functional-level entities according to the requirements of the confrontation scenario, and deploy the initialization simulation model to multiple computing nodes, wherein each computing node performs simulation calculations of at least one of the functional-level entities; The impact weight evaluation module is used to obtain the current state of each functional level entity in the simulation process in real time, and use a machine learning algorithm to evaluate the impact weight of each functional level entity on the simulation result according to the current state; The model simplification module is used to simplify the current simulation model to obtain the next simulation model by adopting a dynamic simplification strategy according to the weight threshold and the influence weights of all the functional-level entities; The simulation propulsion ratio calculation module is used to dynamically obtain the simulation propulsion ratio of the next simulation model by adopting a hybrid propulsion mechanism according to the simulation result output by the current simulation model; The simulation engine module is used to update the next simulation model according to the simulation propulsion ratio and perform simulation calculations.

[0016] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned multi-factor large-scale system confrontation simulation methods, and can realize real-time evaluation of the influence weights of each functional-level entity, and dynamically adjust the simulation model and advancement ratio according to the weight threshold and the simulation results, thereby improving the simulation efficiency and accuracy, better adapting to the purpose of complex confrontation scenarios, and solving the shortcomings of existing simulation methods.

[0017] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program for executing any of the above-mentioned multi-factor large-scale system confrontation simulation methods, and can realize real-time evaluation of the influence weights of each functional-level entity, and dynamically adjust the simulation model and advancement ratio according to the weight threshold and simulation results, thereby improving the simulation efficiency and accuracy, better adapting to the purpose of complex confrontation scenarios, and solving the shortcomings of existing simulation methods.

[0018] Compared with the prior art, the at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects: The present invention provides a multi-factor large-scale system confrontation simulation method, which can significantly improve the simulation efficiency, accuracy and adaptability by introducing machine learning algorithms, dynamic simplification strategies and hybrid advancement mechanisms. The specific technical effects are as follows: 1. It can improve simulation efficiency: By deploying the initial simulation model to multiple computing nodes and optimizing the simulation model using dynamic simplification strategies, unnecessary computing resource consumption is reduced. According to the influence weight of the functional-level entity, the entity with less influence on the simulation results is dynamically eliminated, thereby reducing the simulation complexity and significantly improving the simulation efficiency.

[0019] 2. Enhanced simulation accuracy: Using machine learning algorithms to evaluate the impact weights of functional-level entities in real time can more accurately reflect the role of each entity in the confrontation scenario. By dynamically adjusting the simulation model, key entities can be fully simulated, thereby improving the accuracy and reliability of the simulation results.

[0020] 3. Dynamically adapt to complex scenarios: The simulation propulsion ratio is dynamically adjusted through the hybrid propulsion mechanism, and the simulation step size can be optimized in real time according to the current simulation results and scenario changes. This dynamic adjustment mechanism makes the simulation process more flexible and can adapt to sudden changes and complex interactions in confrontation scenarios.

[0021] 4. Optimize computing resource allocation: Dynamically simplify the simulation model according to the impact weight and weight threshold of the functional-level entity to avoid wasting computing resources. At the same time, through parallel computing of multiple computing nodes, the resource utilization efficiency is further improved, which is particularly suitable for large-scale system confrontation simulation.

[0022] 5. Improve the scalability of the simulation system: The method of the present invention supports the distributed deployment of multiple computing nodes, and can flexibly expand the simulation scale to adapt to larger and more complex confrontation scenarios. At the same time, the combination of dynamic simplification strategy and hybrid advancement mechanism enables the simulation system to maintain efficient operation during expansion.

[0023] 6. Support real-time decision-making and feedback: By acquiring the status of functional-level entities in real time and dynamically adjusting the simulation model, the present invention can provide timely and accurate simulation results to support real-time decision-making in confrontation scenarios. This is of great significance for rapid response in military and other fields.

[0024] 7. Reduce simulation development and maintenance costs: The use of dynamic simplification strategies and hybrid advancement mechanisms reduces the reliance on fixed models and high-performance computing resources, and reduces the development and maintenance costs of simulation systems. At the same time, the introduction of machine learning algorithms reduces manual intervention and further improves the automation level of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0026] Figure 1 A flowchart of a multi-factor large-scale system confrontation simulation method disclosed in an embodiment of the present invention; Figure 2 It is an architecture diagram of a multi-factor large-scale system confrontation simulation system disclosed in an embodiment of the present invention; Figure 3 A schematic diagram of a computer device disclosed in an embodiment of the present invention; Among them, 201, simulation model construction module; 202, impact weight evaluation module; 203, model simplification module; 204, simulation propulsion ratio calculation module; 205, simulation engine module; 301, memory; 302, processor. DETAILED DESCRIPTION

[0027] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0028] The following describes the implementation methods of the present application through specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that the following embodiments and the features of the embodiments can be combined with each other in the absence of conflict. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work belong to the scope of protection of the present application.

[0029] The present invention discloses a multi-factor large-scale system confrontation simulation method, see Figure 1 As shown, the method comprises the following steps: S1. Construct an initialization simulation model including multiple functional-level entities according to the requirements of the confrontation scenario, and deploy the initialization simulation model to multiple computing nodes, wherein each computing node performs simulation calculation of at least one of the functional-level entities; S2. Acquire the current state of each functional-level entity in the simulation process in real time, and use a machine learning algorithm to evaluate the influence weight of each functional-level entity on the simulation result according to the current state; S4, according to the weight threshold and the influence weights of all the functional-level entities, adopt a dynamic simplification strategy to simplify the current simulation model to obtain the next simulation model; S5. According to the simulation result output by the current simulation model, a hybrid propulsion mechanism is used to dynamically obtain a simulation propulsion ratio of the next simulation model; S6. Update the next simulation model according to the simulation propulsion ratio and perform simulation calculation.

[0030] Furthermore, in step S1, an initialization simulation model including multiple functional level entities is constructed according to the requirements of the confrontation scenario, including: S11. Acquire the types and quantities of functional-level entities according to the requirements of the confrontation scenario, wherein the types include attack entities, defense entities, command entities, perception entities, and support entities.

[0031] S12. Define the attributes of each of the functional level entities and initialize each of the functional level entities, wherein the attributes include identity identification, state attributes, capability attributes, resource attributes and location attributes.

[0032] S13, configure behavior rules for each initialized functional entity, establish the interaction logic between all functional entities to obtain a simulation model, and the behavior rules include state transition rules, interaction rules, decision rules and resource management rules. Taking a tank as an example, its state transition rule can be designed so that the tank may change from an "active" state to a "damaged" state after being attacked. Taking a fighter and an air defense missile system as an example, its interaction rule can be designed so that when a fighter enters the air defense missile range, the air defense missile system will intercept. Taking the Red Army Command Center as an example, its decision rule can be designed to decide whether to send a tank unit to attack based on the intelligence of a reconnaissance drone. Taking logistics supply as an example, its resource management rule can be to automatically go to supply when the tank's fuel is less than 50.

[0033] S14, initializing the simulation parameters of the simulation model to obtain an initialized simulation model, wherein the simulation parameters include a simulation step, a propulsion ratio, and an initial state. For example, each simulation step can be set to represent 1 minute in reality. The propulsion ratio can be set to a ratio of simulation time to real time of 1:1. The initial state can be set to the initial state of all functional-level entities as "standby" or "active", and the location attribute is initialized according to the map.

[0034] Furthermore, in step S2, the current state of each functional level entity during the simulation process is obtained in real time, and the influence weight of each functional level entity on the simulation result is evaluated using a machine learning algorithm according to the current state, including: S21. After each simulation, the current status data of the functional-level entity is collected, and the current status data includes position, speed, resource status and historical tasks, wherein the position is the coordinate position of the functional-level entity in the simulation environment, and the environmental conditions can also be obtained according to its position; the speed represents the moving speed of the functional-level entity; the resource status represents the current resource reserve of the functional-level entity, such as ammunition, energy, repair tools, etc.; the historical tasks are the records and situations of the tasks performed by the functional-level entity during the simulation process. These data can be obtained through sensors or log systems in the simulation environment after each simulation, and stored in the database for subsequent analysis.

[0035] S22, extract features from the current state data to construct a feature vector, wherein the features include the health status, attack capability, defense capability and historical contribution of the functional entity, wherein the health status of the functional entity can be evaluated based on the current resource status and historical tasks of the functional entity, for example, when the resources are exhausted or the entity is seriously damaged, the health status is low; the subsequent attack capability of the functional entity can be evaluated based on the weapon status, ammunition reserves and historical attack records of the functional entity; the subsequent defense capability of the functional entity can be evaluated based on the defense equipment, repair capability and historical defense records of the functional entity; the historical contribution of the functional entity can be evaluated based on the number of tasks completed by the functional entity during the simulation process, the difficulty of the tasks and the contribution to the overall battle situation. These features can be combined into a feature vector in a weighted or normalized manner for subsequent model input.

[0036] S23. Use historical simulation data to train a neural network model, input the feature vector into the neural network model, and obtain the influence weight of the functional-level entity on the simulation result. Specifically, it can be achieved by the following steps: First, extract the feature vector and the corresponding simulation result (such as win or loss, task completion, etc.) from the historical simulation data. Secondly, select a suitable neural network model, such as a multi-layer perceptron (MLP) or a convolutional neural network (CNN). The specific choice depends on the complexity of the data and the relationship between the features. Then, use the feature vector as input and the simulation result as output to train the neural network model. Cross-validation and other techniques can be used to optimize the hyperparameters of the model. Finally, during the simulation process, the currently collected feature vector is input into the trained neural network model, and the model will output the influence weight of the functional-level entity on the simulation result.

[0037] Furthermore, in an improved example of the above multi-factor large-scale system confrontation simulation method, the method further includes: S3. Build a dynamic simplification strategy, including: S31, given a weight threshold, defining the functional-level entity whose impact weight is less than the weight threshold as a failed entity. Specifically, when designing a dynamic simplification strategy, a weight threshold can be set according to simulation requirements. , which can be calculated based on indicators such as the combat effectiveness of functional-level entities, the degree of impact on simulation results, and resource consumption; for each functional-level entity (such as combat units, equipment, etc.), its impact weight is calculated ,when When the functional entity is defined as a failed entity, the functional entity can be defined as a failed entity. For example, in a military confrontation simulation, assuming that the influence weight of a combat unit is 0.05, and the weight threshold is set to 0.1, since 0.05 < 0.1, the combat unit is defined as a failed entity.

[0038] S32: Use a simplified model to replace the failed entities in the current simulation model to obtain the next simulation model, or aggregate all the failed entities with the same attributes in the current simulation model into an equivalent entity to obtain the next simulation model, and construct a dynamic simplification strategy.

[0039] For failed entities, a simplified model can be used to replace the complex entity in the current simulation model. The simplified model can be a low-precision version of the entity, or an approximate representation based on a statistical model. If the attributes of multiple failed entities are the same (such as multiple combat units of the same type), they can be aggregated into an equivalent entity, and the attributes of the equivalent entity are the weighted average or sum of these failed entities. For example, in a simulation, assuming that 10 combat units of the same type are defined as failed entities, and the attributes of these combat units (such as firepower, mobility, etc.) are the same, they can be aggregated into an equivalent entity, and the firepower of the equivalent entity is the sum of the 10 combat units, and the mobility is their average value.

[0040] Furthermore, in order to ensure the effectiveness of the constructed dynamic simplification strategy, the above step S3 further includes: S33, introducing a confrontation mechanism in the simulation process, respectively obtaining the simulation results of the current simulation model and the next simulation model, calculating the output difference of the two simulation results to evaluate and optimize the dynamic simplification strategy, specifically including the following process: First, during the simulation process, the current simulation model and the simplified simulation model are run respectively to obtain the results of the two simulations. Secondly, the output difference Δ between the two simulation results is calculated. The output difference Δ can be calculated based on key indicators (such as combat results, resource consumption, time, etc.). Finally, the effectiveness of the dynamic simplification strategy is evaluated based on the output difference Δ. If Δ is within an acceptable range, the simplification strategy is considered effective; otherwise, the weight threshold is adjusted or the model is simplified and the simulation is performed again. For example, in a simulation, assume that the combat result of the current simulation model is a victory, while the combat result of the simplified simulation model is a draw. The output difference Δ is 1 (the difference between victory and draw). If the acceptable difference threshold is set to 0.5, Δ=1 exceeds the acceptable range, and the simplification strategy needs to be optimized. The optimization method can be achieved by adjusting the weight threshold. Or improve the simplified model implementation until the output difference Δ is within an acceptable range during simulation.

[0041] Furthermore, in step S5, according to the simulation result output by the current simulation model, a hybrid propulsion mechanism is used to dynamically obtain the simulation propulsion ratio of the next simulation model, including: S51, obtaining a current simulation result output by the current simulation model, wherein the simulation result includes an occurrence event and a computing node utilization rate; S52, obtaining the proportion of time events and status events in the occurred events; S53: According to the proportion and the computing node utilization rate, a hybrid advancement mechanism is adopted to obtain a simulation advancement ratio of the next simulation model.

[0042] Furthermore, in step S53, according to the proportion and the computing node utilization rate, a hybrid advancement mechanism is adopted to obtain the simulation advancement ratio of the next simulation model, including: S531: According to the proportion and the computing node utilization rate, the formula Calculate the acceleration factor, where the proportion is the ratio of the time event or the state event in all events, and the weight is the influence of the proportion of different time events or the state events on the acceleration factor, and the value range is [0,1]. 0 means ignoring the proportion of the state event, and 1 means fully considering the proportion of the state event. The weight value can be adjusted according to the specific simulation scenario. For example, if the state event has a greater impact on the simulation accuracy, a larger weight value (close to 1) can be set; if the state event has a smaller impact on the simulation accuracy, a smaller weight value (close to 0) can be set.

[0043] When calculating the acceleration factor, the value range of the computing node utilization rate can be set to [0,1], where 0 means that the computing node is idle and 1 means that the computing node is fully loaded. For example, when the computing node utilization rate (U) = 0.8, the status event proportion (S) = 0.6, and the weight (W) corresponding to the status event is selected as 0.7, the acceleration factor (α) = (1-0.8) / (1-0.6*0.7) = 0.2 / 0.58≈0.345.

[0044] S532, set a benchmark advancement ratio, calculate the product of the benchmark advancement ratio and the acceleration factor to obtain the simulation advancement ratio of the next simulation model. When calculating the simulation advancement ratio, when the computing node utilization rate is higher, the acceleration factor is smaller, and the calculated simulation advancement ratio is smaller, so as to avoid computing node overload. When the proportion of state events is higher, the acceleration factor is smaller, and the calculated simulation advancement ratio is smaller, so as to ensure the processing accuracy of state events. When the acceleration factor is 0.345, assuming that the benchmark advancement ratio is 5.0, the simulation advancement ratio of the next simulation model is 0.345*5=1.725. Preferably, it is also necessary to give an upper limit and a lower limit of the advancement ratio to ensure that the calculated simulation advancement ratio is between the upper limit and the lower limit, so as to avoid the situation of simulation distortion when the simulation advancement ratio of the next simulation model is too large or too small when performing simulation calculation.

[0045] Based on the same inventive concept, a multi-factor large-scale system confrontation simulation system is also provided in the embodiments of the present invention, as described in the following embodiments. Since the principle of solving the problem by the multi-factor large-scale system confrontation simulation system is similar to the multi-factor large-scale system confrontation simulation method provided in the above embodiments, the implementation of the confrontation simulation system can refer to the implementation of the above confrontation simulation method, and the repeated parts will not be repeated. As used below, the term "unit" or "module" can be a combination of software and / or hardware that implements predetermined functions. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0046] Figure 2 is a structural block diagram of a multi-factor large-scale system confrontation simulation system disclosed in an embodiment of the present invention, such as Figure 2 As shown, the confrontation simulation system includes a simulation model construction module 201, an impact weight evaluation module 202, a model simplification module 203, a simulation propulsion ratio calculation module 204 and a simulation engine module 205, and the structure is described below.

[0047] The simulation model construction module 201 is used to construct an initialization simulation model including multiple functional-level entities according to the requirements of the confrontation scenario, and deploy the initialization simulation model to multiple computing nodes, wherein each computing node performs simulation calculations of at least one of the functional-level entities; The influence weight evaluation module 202 is used to obtain the current state of each functional level entity in the simulation process in real time, and use a machine learning algorithm to evaluate the influence weight of each functional level entity on the simulation result according to the current state; The model simplification module 203 is used to simplify the current simulation model to obtain the next simulation model by adopting a dynamic simplification strategy according to the weight threshold and the influence weights of all the functional-level entities; The simulation propulsion ratio calculation module 204 is used to dynamically obtain the simulation propulsion ratio of the next simulation model by adopting a hybrid propulsion mechanism according to the simulation result output by the current simulation model; The simulation engine module 205 is used to update the next simulation model according to the simulation propulsion ratio and perform simulation calculations.

[0048] The embodiments of the present invention achieve the following technical effects: 1. It can improve simulation efficiency: By deploying the initial simulation model to multiple computing nodes and optimizing the simulation model using dynamic simplification strategies, unnecessary computing resource consumption is reduced. According to the influence weight of the functional-level entity, the entity with less influence on the simulation results is dynamically eliminated, thereby reducing the simulation complexity and significantly improving the simulation efficiency.

[0049] 2. Enhanced simulation accuracy: Using machine learning algorithms to evaluate the impact weights of functional-level entities in real time can more accurately reflect the role of each entity in the confrontation scenario. By dynamically adjusting the simulation model, key entities can be fully simulated, thereby improving the accuracy and reliability of the simulation results.

[0050] 3. Dynamically adapt to complex scenarios: The simulation propulsion ratio is dynamically adjusted through the hybrid propulsion mechanism, and the simulation step size can be optimized in real time according to the current simulation results and scenario changes. This dynamic adjustment mechanism makes the simulation process more flexible and can adapt to sudden changes and complex interactions in confrontation scenarios.

[0051] 4. Optimize computing resource allocation: Dynamically simplify the simulation model according to the impact weight and weight threshold of the functional-level entity to avoid wasting computing resources. At the same time, through parallel computing of multiple computing nodes, the resource utilization efficiency is further improved, which is particularly suitable for large-scale system confrontation simulation.

[0052] 5. Improve the scalability of the simulation system: The method of the present invention supports the distributed deployment of multiple computing nodes, and can flexibly expand the simulation scale to adapt to larger and more complex confrontation scenarios. At the same time, the combination of dynamic simplification strategy and hybrid advancement mechanism enables the simulation system to maintain efficient operation during expansion.

[0053] 6. Support real-time decision-making and feedback: By acquiring the status of functional-level entities in real time and dynamically adjusting the simulation model, the present invention can provide timely and accurate simulation results to support real-time decision-making in confrontation scenarios. This is of great significance for rapid response in military and other fields.

[0054] 7. Reduce simulation development and maintenance costs: The use of dynamic simplification strategies and hybrid advancement mechanisms reduces the reliance on fixed models and high-performance computing resources, and reduces the development and maintenance costs of simulation systems. At the same time, the introduction of machine learning algorithms reduces manual intervention and further improves the automation level of the system.

[0055] In this embodiment, a computer device is provided, such as Figure 3 As shown, it includes a memory 301, a processor 302 and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, any of the above-mentioned multi-factor large-scale system confrontation simulation methods is implemented.

[0056] Specifically, the computer device may be a computer terminal, a server or a similar computing device.

[0057] In this embodiment, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program for executing any of the above-mentioned multi-element large-scale system confrontation simulation methods.

[0058] Specifically, computer-readable storage media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer-readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable storage media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0059] Obviously, those skilled in the art should understand that the modules or steps of the above-mentioned embodiments of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. In this way, the embodiments of the present invention are not limited to any specific combination of hardware and software.

[0060] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the embodiments of the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-factor large-scale system confrontation simulation method, characterized in that: The method comprises: Constructing an initialization simulation model including multiple functional-level entities according to the requirements of the confrontation scenario, and deploying the initialization simulation model to multiple computing nodes, wherein each computing node performs simulation calculation of at least one of the functional-level entities; Acquire the current state of each functional-level entity in the simulation process in real time, and use a machine learning algorithm to evaluate the influence weight of each functional-level entity on the simulation result according to the current state; According to the weight threshold and the influence weights of all the functional-level entities, a dynamic simplification strategy is adopted to simplify the current simulation model to obtain the next simulation model; According to the simulation result output by the current simulation model, a hybrid propulsion mechanism is used to dynamically obtain a simulation propulsion ratio of the next simulation model; The next simulation model is updated according to the simulation propulsion ratio and simulation calculation is performed.

2. The multi-factor large-scale system confrontation simulation method according to claim 1, characterized in that: According to the requirements of the confrontation scenario, an initialization simulation model including multiple functional level entities is constructed, including: Acquire the types and quantities of functional-level entities according to the requirements of the confrontation scenario, wherein the types include offensive entities, defensive entities, command entities, perception entities, and support entities; Defining attributes of each of the functional level entities and initializing each of the functional level entities, wherein the attributes include identity, state attributes, capability attributes, resource attributes, and location attributes; Configure behavior rules for each of the initialized functional level entities, establish interaction logic between all the functional level entities to obtain a simulation model, wherein the behavior rules include state transition rules, interaction rules, decision rules and resource management rules; The simulation parameters of the simulation model are initialized to obtain an initialized simulation model, wherein the simulation parameters include a simulation step length, a propulsion ratio and an initial state.

3. The multi-factor large-scale system confrontation simulation method according to claim 1 is characterized in that: Acquire the current state of each functional level entity in the simulation process in real time, and use a machine learning algorithm to evaluate the influence weight of each functional level entity on the simulation result according to the current state, including: After each simulation is completed, current state data of the functional level entity is collected, and the current state data includes position, speed, resource state and historical tasks; Extracting features from the current state data to construct a feature vector, wherein the features include the health state, attack capability, defense capability, and historical contribution of the functional-level entity; A neural network model is trained using historical simulation data, and the feature vector is input into the neural network model to obtain the influence weight of the functional level entity on the simulation result.

4. The multi-factor large-scale system confrontation simulation method according to claim 1, characterized in that: The method further comprises: Given a weight threshold, defining the functional-level entity whose impact weight is less than the weight threshold as a failed entity; A simplified model is used to replace the failed entities of the current simulation model to obtain the next simulation model, or all the failed entities with the same attributes in the current simulation model are aggregated into an equivalent entity to obtain the next simulation model, so as to construct a dynamic simplification strategy.

5. The multi-factor large-scale system confrontation simulation method according to claim 4 is characterized in that: Also includes: An adversarial mechanism is introduced during the simulation process to obtain the simulation results of the current simulation model and the next simulation model respectively, and the output difference of the two simulation results is calculated to evaluate and optimize the dynamic simplification strategy.

6. The multi-factor large-scale system confrontation simulation method according to claim 1, characterized in that: According to the simulation result output by the current simulation model, a hybrid propulsion mechanism is used to dynamically obtain the simulation propulsion ratio of the next simulation model, including: Obtaining a current simulation result output by the current simulation model, wherein the simulation result includes an occurrence event and a computing node utilization rate; Obtaining the proportion of time events and status events in the occurred events; According to the proportion and the computing node utilization rate, a hybrid advancement mechanism is adopted to obtain the simulation advancement ratio of the next simulation model.

7. The multi-factor large-scale system confrontation simulation method according to claim 6, characterized in that: According to the proportion and the computing node utilization rate, a hybrid advancement mechanism is adopted to obtain a simulation advancement ratio of the next simulation model, including: According to the proportion and the computing node utilization rate, the formula Calculate the acceleration factor, where the proportion is the ratio of the time event or the state event in all events, and the weight is the influence of the proportion of different time events or state events on the acceleration factor, and the value range is [0, 1], 0 means ignoring the proportion of the state event, and 1 means fully considering the proportion of the state event; A reference propulsion ratio is set, and the product of the reference propulsion ratio and the acceleration factor is calculated to obtain the simulation propulsion ratio of the next simulation model.

8. A multi-element large-scale system confrontation simulation system, characterized in that: include: A simulation model construction module, wherein the simulation model construction module is used to construct an initialization simulation model including multiple functional-level entities according to the requirements of the confrontation scenario, and deploy the initialization simulation model to multiple computing nodes, wherein each computing node performs simulation calculations of at least one of the functional-level entities; An impact weight evaluation module, the impact weight evaluation module is used to obtain the current state of each functional level entity in the simulation process in real time, and use a machine learning algorithm to evaluate the impact weight of each functional level entity on the simulation result according to the current state; A model simplification module, wherein the model simplification module is used to simplify the current simulation model to obtain the next simulation model by adopting a dynamic simplification strategy according to a weight threshold and the influence weights of all the functional-level entities; A simulation propulsion ratio calculation module, the simulation propulsion ratio calculation module is used to dynamically obtain the simulation propulsion ratio of the next simulation model by adopting a hybrid propulsion mechanism according to the simulation result output by the current simulation model; A simulation engine module, wherein the simulation engine module is used to update the next simulation model according to the simulation propulsion ratio and perform simulation calculations.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the multi-factor large-scale system confrontation simulation method described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for executing the multi-element large-scale system confrontation simulation method according to any one of claims 1 to 7.

Images