An intelligent war game simulation system based on a digital twin fusion large model

Abstract

The present application relates to the technical field of intelligent war game, in particular to an intelligent war game simulation system based on digital twin fusion large model, which comprises a digital twin modeling unit, a large model decision unit, a deduction execution unit and a dynamic interaction unit. The digital twin modeling unit adopts a dynamic precision adaptive modeling mechanism, which can realize dynamic switching of model precision on demand, balance the demand for detailed simulation and the consumption of computing resources, and realize sub-second dynamic synchronization of virtual models and war game scenes through a timing calibration algorithm. Through the multi-modal feature analysis module, spatial adversarial, timing decision and rule trigger features are fused, and the entity performance boundary and scene rule constraint are embedded in the war game rule verification module, which can generate compliant adversarial decision parameters in line with the logic of war game, solving the problem of insufficient model precision adaptation and decision constraint.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent war game, in particular, to an intelligent war game simulation system based on digital twin fusion large model. BACKGROUND

[0002] War game is the core means of modern national defense strategy planning, tactical and operational method verification and command decision training. Its core is to pre-act the process and results of confrontation in virtual space through modeling and simulation technology. With the improvement of the complexity of military systems and the deepening of joint operation needs, traditional war game systems face serious challenges in terms of simulation fidelity, decision intelligence, and real-time continuity of the simulation process. How to build a simulation system that can simulate real battlefield with high fidelity, embed intelligent decision-making capabilities, and ensure efficient synchronization throughout the process has become a key direction for the development of the field.

[0003] In the prior art, related patents have explored digital twin applications, simulation process optimization, etc. For example, Chinese patent CN202311436752.8 discloses a digital twin war game weather data simulation and intelligent decision-making method and system. Its technical points are: combining digital twin technology with weather data simulation, building six sub-models of task target constraint, marine battlefield geographic information analysis, force and equipment deployment, enemy data analysis, weather comprehensive prediction, and action game, combining actual weather data with the model to simulate the influence of different weather on the simulation, and providing reference for military decision-making; and Chinese patent CN202211414513.8 discloses a war game system, which includes a simulation design module (receiving round setting instruction to generate round planning result), a simulation control module (dynamically managing simulation round information to generate round event information), a simulation simulation module (simulating to generate round result according to round event information), and a simulation command module (receiving command instruction to generate command information), which can solve the problems of fixed time and space resolution, weak rule adaptability, inflexible control, and difficulty in realizing multi-level linkage in traditional war game, and improve the efficiency and flexibility, scalability of the system.

[0004] Despite the design advantages of the aforementioned technical solutions, they also suffer from the following technical shortcomings: First, insufficient model accuracy adaptation and decision constraints: CN202311436752.8 fails to dynamically adjust model accuracy based on the interaction frequency and spatial distance of wargame entities, making it unable to balance detailed simulation and resource consumption, and lacks a high-precision temporal calibration mechanism; decision-making revolves solely around meteorological data, failing to integrate spatial, temporal, and other multimodal characteristics, and also lacking entity performance boundaries and scene rule constraints, easily leading to illogical decisions. Second, inadequate virtual fidelity and simulation continuity: CN202211414513.8 lacks a high-fidelity virtual model of wargame entities, making it difficult to reproduce entity attributes and real-time states; it lacks a low-latency data interaction channel and a dynamic correction mechanism for simulation deviations, making it impossible to adjust simulation paths deviating from expectations in a timely manner, affecting continuity. Therefore, we propose an intelligent wargame simulation system based on a digital twin fusion large model. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent war game simulation system based on a digital twin fusion model, so as to solve the problems of insufficient model accuracy and decision constraints and lack of virtual fidelity and simulation coherence mentioned in the background art.

[0006] To address the aforementioned technical problems, the present invention aims to provide an intelligent wargaming simulation system based on a digital twin fusion model, comprising:

[0007] The digital twin modeling unit adopts a dynamic precision adaptive modeling mechanism, which constructs a hierarchical virtual model of the wargame scenario through the fusion of multi-source heterogeneous data. The hierarchical virtual model automatically adjusts the model precision level according to the granularity requirements of the wargame simulation, and achieves sub-second dynamic synchronization between the virtual model and the wargame simulation scenario through a time-series calibration algorithm.

[0008] The large model decision unit receives wargame scenario state data output by the digital twin modeling unit. It employs a multimodal feature fusion and wargame rule constraint embedding mechanism to fuse the spatial adversarial features and temporal decision features of the wargame scenario. The performance boundaries of the wargame adversarial entities and scenario rules are embedded as constraints into the pre-trained model to generate adversarial decision parameters that conform to the wargame deduction logic. The large model decision unit includes a multimodal feature parsing module, a fusion inference module, and a wargame rule verification module.

[0009] The deduction execution unit adopts a wargame space-time coupling deduction mechanism, drives an adversarial process according to a wargame deduction time sequence based on a wargame virtual model of the digital twin modeling unit and adversarial decision parameters of the large model decision unit, simulates a wargame adversarial result through space-time correlation calculation, and is internally provided with a dynamic correction mechanism to adjust a deduction path according to a deviation of the virtual model from an expected state of the wargame deduction;

[0010] The dynamic interaction unit is used to establish a two-way channel of wargame data between the digital twin modeling unit, the large model decision unit and the deduction execution unit, realize low-delay interactive transmission of wargame scene feature parameters, adversarial decision instructions and deduction state data, and guarantee real-time and continuity of the wargame deduction.

[0011] As a further improvement of the technical solution, the digital twin modeling unit comprises a geometric feature mapping module, an attribute parameter binding module and a state label synchronization module, wherein:

[0012] The geometric feature mapping module generates and stores three-dimensional contour data and spatial coordinates of the wargame adversarial entity based on three-dimensional scanning data of the wargame adversarial entity and parameterized modeling rules, adopts multi-detail level model loading technology, and supports millimeter to meter level precision dynamic switching;

[0013] The attribute parameter binding module binds inherent performance parameters of the wargame adversarial entity to entity types based on a wargame deduction rule library and an entity type configuration table, adopts parameter dynamic correlation technology, and automatically updates performance parameters when the state of the wargame adversarial entity changes;

[0014] The state label synchronization module marks the current state of the wargame adversarial entity based on entity behavior result data output by the deduction execution unit, adopts state real-time triggering technology, and controls state switching delay to be less than 100 milliseconds.

[0015] As a further improvement of the technical solution, the digital twin modeling unit further comprises a precision level division module, a precision adjustment triggering module and a time sequence calibration execution module, wherein:

[0016] The precision level division module divides model precision into three levels based on interaction frequency and spatial scale of the wargame adversarial entity by adopting distance-precision mapping rules; first-level precision (millimeter level) is suitable for an interaction scenario in which distances between wargame adversarial entities are in a close range, second-level precision (centimeter level) is suitable for a scenario in which distances between wargame adversarial entities are in a medium range, and third-level precision (meter level) is suitable for a scenario in which distances between wargame adversarial entities are in a long range;

[0017] The precision adjustment trigger module triggers precision upgrade automatically when the relative motion state between the wargaming entities meets the preset trigger condition based on the relative motion data of the wargaming entities, and triggers precision downgrade automatically when there is no interaction between the wargaming entities for a preset duration.

[0018] The time sequence calibration execution module calculates the time deviation between the virtual model and the physical scene by a time sequence calibration algorithm based on the physical scene acquisition data with time stamp, and corrects the state parameters of the virtual model by linear interpolation if the time deviation exceeds the preset acceptable range, so as to ensure the synchronization of the virtual model and the wargaming scene.

[0019] As a further improvement of the technical solution, the multi-modal feature analysis module realizes the structured extraction of the wargaming scene features through feature quantization and standardization, specifically including:

[0020] For the spatial confrontation features in the wargaming scene state data, the absolute coordinates of the confrontation entities are converted into quantized parameters of relative distance, azimuth angle and terrain shielding coefficient by the coordinate system conversion method;

[0021] For the time sequence decision features, the historical action sequence of the confrontation entities is intercepted by a sliding window, and the time sequence parameters of action interval and strategy conversion frequency are extracted;

[0022] For the rule trigger features, the rule conditions of resource threshold and round progress are converted into binary feature codes (1 for meeting and 0 for not meeting);

[0023] And the spatial confrontation features, time sequence decision features and rule trigger features are normalized to a unified data dimension to form a multi-modal fusion feature vector.

[0024] As a further improvement of the technical solution, the fusion reasoning module generates an initial confrontation decision scheme through feature weighted fusion and knowledge graph reasoning, specifically including:

[0025] The fusion feature vector output by the multi-modal feature analysis module is dynamically weighted by using the attention mechanism, in which the spatial feature weight linearly increases with the entity interaction intensity, and the time sequence feature weight linearly increases with the action coherence index;

[0026] A wargaming confrontation knowledge graph is constructed based on a pre-trained model, the graph nodes include entity type, action strategy and rule clause, and the optimal decision logic of the historical confrontation case is matched by calculating the cosine similarity of the nodes;

[0027] The initial confrontation decision scheme including action priority, resource allocation ratio and strategy alternative sequence is output according to the reasoning result, and the scheme is stored in a structured data format, and the data fields include decision ID, target entity ID and execution time sequence label.

[0028] As a further improvement to this technical solution, the wargame rule verification module ensures decision compliance through constraint matching and gradient correction, specifically including:

[0029] Retrieve constraints that match the current scenario from the wargame rule base, including the performance boundaries of wargame adversaries and scenario rules;

[0030] The parameters and constraints of the initial adversarial decision-making scheme are compared item by item to determine whether there are any violations such as parameter out-of-bounds or rule conflicts.

[0031] The gradient adjustment method is used to correct violations. For performance boundary violations, the parameters are truncated according to the maximum allowable value. For scenario rule violations, resources are reallocated and the action sequence is adjusted according to the rule priority. After correction, the output adversarial decision parameters conform to the wargaming simulation logic.

[0032] As a further improvement to this technical solution, the simulation execution unit includes a time-series advancement module, a spatial interaction calculation module, and an adversarial result synthesis module, wherein:

[0033] The timing progression module is used to drive the adversarial process according to the wargame simulation sequence, and to mark round-based simulations with round markers. As a time series unit, in the first At the start of each round, the adversarial decision parameters output by the large model decision unit are loaded; real-time simulations are performed using time steps. As a time series unit, in The state data of the wargame adversaries is updated synchronously within the interval; among which... Indicates the start time point of the real-time simulation. Indicates starting time point As a baseline, plus a time step The final time point obtained later;

[0034] The spatial interaction computing module is used to process the spatial interaction of wargame adversaries, and obtains entity coordinates based on the wargame virtual model of the digital twin modeling unit. ,in Indicates the entity number and terrain parameters. That is, the terrain slope, used to calculate the attack effect value. and degree of movement obstruction Among them, attack effect value Related firepower parameters Values, entity spacing Terrain correction factor Degree of movement obstruction Correlation of terrain type coefficients Elevation difference ;

[0035] The confrontation result synthesis module is used for integrating the time sequence data and the space interaction result to generate a wargame confrontation result, wherein the entity life value is updated in association with the attack effect value , the resource change is associated with the interaction frequency, the rule triggering result is associated with the attack effect value Threshold determination, all results are stored with round marks As time sequence labels.

[0036] As a further improvement of the technical solution, the dynamic correction mechanism of the deduction execution unit further includes a deviation monitoring module, a correction strategy generation module and a path adjustment execution module, wherein:

[0037] The deviation monitoring module is used for calculating the deviation of the wargame virtual model from the expected state, comparing the entity position Output by the digital twin modeling unit with the expected position To obtain the position deviation amount , comparing the resource value With the expected resource Obtain the resource deviation amount ;

[0038] The correction strategy generation module is used to develop a deduction path adjustment scheme, , the threshold value corresponding to the resource deviation When And Maintain the current path; otherwise, generate a position compensation value , resource compensation value And time sequence compensation step ;

[0039] The path adjustment execution module is used to adjust the deduction path by applying the correction parameters to update the entity coordinates to , wherein , , The current space coordinates of the entity are , , The position compensation values in the axis, axis, axis direction, respectively; the resource value is updated to , and the subsequent deduction time sequence is adjusted to , and the adjustment result is fed back to the digital twin modeling unit to update the wargame virtual model.

[0040] As a further improvement of the technical solution, the dynamic interaction unit includes a data acquisition module, a low-delay transmission module and a state synchronization module, wherein:

[0041] The data acquisition module is configured to acquire the wargame scene feature parameters from the digital twin modeling unit, acquire the confrontation decision instruction from the large model decision unit, and acquire the deduction state data from the deduction execution unit;

[0042] The low-delay transmission module is configured to build a wargame data bidirectional transmission channel, adopt a lightweight encapsulation transmission technology based on a UDP protocol, and realize low-delay interactive transmission of the wargame scene feature parameters, the confrontation decision instruction and the deduction state data.

[0043] The state synchronization module is configured to synchronize the states of the wargame virtual scene of the digital twin modeling unit, the decision generation logic of the large model decision unit and the deduction process of the deduction execution unit according to the data transmitted by the low-delay transmission module, and guarantee the real-time and continuity of the wargame deduction.

[0044] As a further improvement of the technical solution, the technical solution further comprises a deduction result evaluation unit configured to evaluate the effectiveness of the wargame confrontation result output by the deduction execution unit, and the deduction result evaluation unit comprises a result and benchmark acquisition module, a multi-dimensional evaluation module and an evaluation result feedback module.

[0045] The result and benchmark acquisition module is configured to acquire the wargame confrontation result data from the deduction execution unit, acquire the wargame virtual model state data of the corresponding deduction stage from the digital twin modeling unit, and simultaneously acquire the data interaction delay record of the corresponding deduction stage from the dynamic interaction unit as the evaluation benchmark data.

[0046] The multi-dimensional evaluation module is configured to judge the effectiveness of the confrontation result against the existing standard, and specifically comprises the following: judging the deduction logic compliance according to the compliance standard output by the wargame rule verification module, judging the virtual scene synchronization accuracy according to the sub-second level synchronization requirement of the time sequence calibration execution module, and judging the data interaction real-time according to the low-delay requirement of the dynamic interaction unit. If all the three items meet the requirements, the result is determined to be effective; if any of the three items fails to meet the requirements, the result is marked as a to-be-optimized result and the item failing to meet the requirement is recorded.

[0047] The evaluation result feedback module is configured to store the effective evaluation result to a wargame deduction result library, and feed back the to-be-optimized result and the item failing to meet the requirement to the large model decision unit and the digital twin modeling unit. The information fed back to the large model decision unit is used to optimize the subsequent confrontation decision parameter generation logic, and the information fed back to the digital twin modeling unit is used to assist in adjusting the model accuracy level or the time sequence calibration frequency.

[0048] Compared with the prior art, the present application has the following advantages:

[0049] 1.The application adopts a dynamic precision adaptive modeling mechanism through a digital twin modeling unit, can realize dynamic switching of model precision on demand, balances the demand for detailed simulation and the consumption of computing resources, and realizes sub-second dynamic synchronization of the virtual model and the war game scenario through a timing calibration algorithm; and a large model decision unit fuses spatial confrontation, timing decision and rule triggering features through a multi-modal feature analysis module, embeds entity performance boundaries and scene rule constraints in combination with a war game rule verification module, can generate compliant confrontation decision parameters in line with the war game deduction logic, and solves the problems of model precision adaptation and insufficient decision constraints.

[0050] 2.The application generates three-dimensional profile data of war game confrontation entities through a geometric feature mapping module of a digital twin modeling unit, dynamically updates entity inherent performance parameters through an attribute parameter binding module, and marks real-time states of entities through a state label synchronization module, constructs a high-fidelity war game virtual model, accurately reproduces entity geometric features, attribute parameters and current states; and a dynamic interaction unit adopts a lightweight encapsulation transmission technology based on a UDP protocol to construct a low-latency bidirectional data channel, a deduction execution unit is built-in with a dynamic correction mechanism to timely adjust the deduction path deviating from the expectation, guarantees the real-time and continuity of the war game deduction, and solves the problems of virtual fidelity and deduction continuity. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 The figure is a schematic diagram of the system framework of the application;

[0052] The meanings of the various reference numerals in the figure are as follows:

[0053] 100, digital twin modeling unit; 110, geometric feature mapping module; 120, attribute parameter binding module; 130, state label synchronization module; 140, precision level division module; 150, precision adjustment trigger module; 160, timing calibration execution module;

[0054] 200, large model decision unit; 210, multi-modal feature analysis module; 220, fusion reasoning module; 230, war game rule verification module;

[0055] 300, deduction execution unit; 310, timing advance module; 320, spatial interaction calculation module; 330, confrontation result synthesis module; 340, deviation monitoring module; 350, correction strategy generation module; 360, path adjustment execution module;

[0056] 400, dynamic interaction unit; 410, data acquisition module; 420, low-latency transmission module; 430, state synchronization module;

[0057] 500, deduction result evaluation unit; 510, result and benchmark acquisition module; 520, multi-dimensional evaluation module; 530, evaluation result feedback module. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0059] like Figure 1 As shown, this embodiment provides an intelligent wargaming simulation system based on a digital twin fusion model, including:

[0060] The digital twin modeling unit 100 adopts a dynamic precision adaptive modeling mechanism. It constructs a hierarchical virtual model of the war game scenario through the fusion of multi-source heterogeneous data. The hierarchical virtual model automatically adjusts the model precision level according to the granularity requirements of the war game simulation, and achieves sub-second dynamic synchronization between the virtual model and the war game simulation scenario through a time-series calibration algorithm.

[0061] In this embodiment, the digital twin modeling unit 100 includes a geometric feature mapping module 110, an attribute parameter binding module 120, and a status label synchronization module 130, wherein:

[0062] The geometric feature mapping module 110 is based on the 3D scanning data and parametric modeling rules of the wargame adversarial entities. It adopts multi-detail level model loading technology to generate and store the 3D contour data and spatial coordinates of the wargame adversarial entities, and supports dynamic switching of precision from millimeter level to meter level.

[0063] Specifically, a 3D laser scanner is first used to perform a full-angle scan of the wargame entities (such as combat equipment models, terrain sand table components, etc.) to acquire point cloud data. Then, according to parametric modeling rules (such as CAD-based parametric modeling standards), the point cloud data is converted into 3D model files (such as FBX format files) that support multi-level detail technology. The geometric feature mapping module 110, with the help of multi-level detail model loading technology, can quickly switch between millimeter-level, centimeter-level, and meter-level precision model instances according to the rendering requirements and computing resource status of the wargame simulation. When the wargame entity is in the center of the simulation interface and needs to be displayed in detail, a millimeter-level precision model is loaded; when the entity is at the edge of the field of view or when multiple entities need to be rendered in batches, it automatically switches to a meter-level precision model to balance the rendering effect and system performance.

[0064] The attribute parameter binding module 120 binds the inherent performance parameters of the wargame entity with the entity type based on the wargame rule library and the entity type configuration table, and automatically updates the performance parameters when the state of the wargame entity changes, such as being damaged or resource consumption.

[0065] Specifically, the attribute parameter binding module 120 pre-establishes a wargame rule library (storing rules such as equipment range, moving speed, damage threshold) and an entity type configuration table (defining attribute templates of different types of entities). After the geometric feature mapping module 110 generates the three-dimensional model of the entity, the attribute parameter binding module 120 binds the inherent performance parameters (retrieved from the configuration table) of the corresponding entity type with the identification information (such as model ID) of the three-dimensional model through the parameter dynamic association technology, forms an associated data structure of “geometric feature-performance parameter” and stores it in the system database. During the deduction process, if the deduction execution unit 300 outputs the behavior result data of the entity being damaged (such as a signal that a certain equipment is hit), the attribute parameter binding module 120 will automatically update the performance parameters (such as reducing the moving speed by a certain percentage and shortening the range) bound to the three-dimensional model of the entity according to the corresponding rules of “damage and performance decay” in the wargame rule library, and synchronize the updated parameters to the state label synchronization module 130 and the precision adjustment trigger module 150.

[0066] Further, the wargame rule library data is derived from public military standards, equipment technical manuals and authoritative deduction data sets (such as 2024 Rand “Information Warfare Wargame: Rulebook (2nd Edition)”), classified according to entity interaction, environmental influence and time sequence, converted into a machine interpretable format, and verified and corrected through real deduction cases. The entity type configuration table is based on equipment classification standards, public parameters and general templates to build a “large class-medium class-subclass” system, bind entity inherent performance, state association and other parameters, support dynamic update, and both are based on public data.

[0067] The state label synchronization module 130 marks the current state of the wargame entity based on the entity behavior result data output by the deduction execution unit 300 using the state real-time triggering technology, and the state switching delay is controlled within sub-hundred milliseconds.

[0068] In the embodiment, the digital twin modeling unit 100 further comprises a precision level division module 140, a precision adjustment trigger module 150 and a time sequence calibration execution module 160, wherein:

[0069] The precision level division module 140 divides the model precision into three levels based on the interaction frequency and spatial scale of the wargame entities, using a distance-precision mapping rule; the first level precision (millimeter level) is suitable for interaction scenarios where the distance between wargame entities is in the near distance range, the second level precision (centimeter level) is suitable for scenarios where the distance between wargame entities is in the medium distance range, and the third level precision (meter level) is suitable for scenarios where the distance between wargame entities is in the long distance range;

[0070] Specifically, in the system initialization phase, based on the typical scenarios of wargame, the "distance-precision mapping rule" is defined in advance: the distance range between wargame entities is divided into near distance (such as 0-10 meters), medium distance (such as 10-50 meters), and long distance (such as more than 50 meters), and corresponding first level precision (millimeter level), second level precision (centimeter level), and third level precision (meter level) are set. During the deduction process, the precision level division module 140 continuously obtains the real-time position information of each entity from the state label synchronization module 130, calculates the relative distance between entities, determines the precision level to be used at present, and sends the precision level information to the precision adjustment trigger module 150 and the geometric feature mapping module 110.

[0071] The precision adjustment trigger module 150 uses threshold monitoring technology based on the relative motion data of the wargame entities, and automatically triggers precision upgrade when the relative motion state between the wargame entities meets the preset trigger condition; when there is no interaction between the wargame entities for a continuous preset time, it automatically triggers precision downgrade;

[0072] Specifically, the precision adjustment trigger module 150 obtains the relative motion data (such as relative speed, acceleration, motion direction change, etc.) of the wargame entities from the state label synchronization module 130, combines the current precision level information sent by the precision level division module 140, and determines whether to trigger precision adjustment through threshold monitoring technology;

[0073] For example, when two entities originally in "long distance (third level precision)" rapidly shorten the distance to the medium distance range due to relative motion and the relative speed exceeds the preset threshold (such as 5 meters / second), the precision upgrade (switching from third level precision to second level precision) is automatically triggered; if the wargame entities in a certain area have no interaction behavior for 5 minutes, the precision adjustment trigger module 150 automatically triggers the precision downgrade, reducing the system's investment in computing resources for the model in that area;

[0074] After each precision adjustment trigger, the precision adjustment trigger module 150 sends a precision adjustment instruction to the geometric feature mapping module 110 to drive it to switch the model precision.

[0075] The time sequence calibration execution module 160 is based on the physical scene acquisition data with timestamp, adopts the time deviation correction technology, calculates the time deviation of the virtual model and the physical scene through the time sequence calibration algorithm, and if the time deviation exceeds the preset acceptable range, the state parameters of the virtual model are corrected through linear interpolation, to ensure the synchronization of the virtual model and the war game scene.

[0076] Specifically, the time sequence calibration execution module 160 continuously acquires physical scene acquisition data with timestamp (such as real war game sand table time synchronization signals collected by external sensors, entity position change time markers, etc.) through a special data acquisition interface; at the same time, the time-related data of the virtual model are acquired from the internal modules of the digital twin modeling unit (such as the model state update time of the geometric feature mapping module 110, the state label time of the state label synchronization module 130). Subsequently, the time deviation correction technology is adopted, and the time deviation of the virtual model and the physical scene is calculated through the time sequence calibration algorithm (such as the time deviation fitting algorithm based on the least square method). If the time deviation (such as the virtual model time is 200 milliseconds slower than the physical scene) exceeds the preset acceptable range (such as ±100 milliseconds), the state parameters of the virtual model (such as entity position, velocity, state label timestamp, etc.) are corrected through linear interpolation, so that the time advancement of the virtual model is kept in synchronization with the physical scene, and the time sequence consistency of the virtual model and the real scene in the war game is ensured.

[0077] The large model decision unit 200 is used for receiving the war game scene state data output by the digital twin modeling unit 100, adopting a multi-modal feature fusion and war game rule constraint embedding mechanism, fusing and processing the spatial confrontation features and time sequence decision features of the war game scene, and embedding the performance boundary of the war game confrontation entity and the scene rules as constraint conditions into the pre-trained model to generate confrontation decision parameters conforming to the war game deduction logic; wherein the large model decision unit 200 includes a multi-modal feature analysis module 210, a fusion reasoning module 220 and a war game rule verification module 230.

[0078] In the present embodiment, the multi-modal feature analysis module 210 realizes the structured extraction of the war game scene features through feature quantization and standardization processing, specifically including:

[0079] For the spatial confrontation features in the war game scene state data, the absolute coordinates of the confrontation entity are converted into quantization parameters of relative distance, azimuth angle and terrain shielding coefficient through the coordinate system conversion method;

[0080] For the time sequence decision features, the historical action sequence of the confrontation entity is intercepted through a sliding window, and the time sequence parameters of action interval and strategy conversion frequency are extracted;

[0081] For rule trigger features, the resource threshold and the rule condition of round progress are converted into binary feature codes; (1 for meeting, 0 for not meeting);

[0082] And the space confrontation features, time sequence decision features and rule trigger features are normalized to a unified data dimension to form a multi-modal fusion feature vector.

[0083] Specifically, the multi-modal feature analysis module 210 first receives the wargame scenario state data (including entity coordinates, historical action records, resource reserves, round progress, etc.) output by the digital twin modeling unit 100, and then realizes the structured extraction of the wargame scenario features through feature quantization and standardization processing, which specifically includes:

[0084] The conversion method of "WGS84 geographic coordinate system to local Cartesian coordinate system" is adopted to convert the entity absolute coordinates (longitude, latitude, and elevation) into local coordinates (x, y, z) with the center point of the deduction scene as the origin. The relative distance between entities is calculated by the Euclidean distance formula, and the azimuth angle is solved by using the trigonometric function (such as arctan2). At the same time, based on the three-dimensional terrain model generated by the digital twin modeling unit 100, the terrain occlusion coefficient is calculated by ray tracing method to judge the terrain occlusion between two entities (no occlusion is 1, complete occlusion is 0, and partial occlusion is calculated according to the occlusion area ratio), and the space confrontation feature quantization is completed.

[0085] According to the deduction type, the size of the sliding window is determined (when the real-time deduction takes 1 second as the time step, a 10-second sliding window is used; when the round deduction takes 3 rounds as a window period, a 10-second sliding window is used). The historical action sequence of the entity in the window is intercepted, the time difference or round difference between the adjacent two actions is extracted as the action interval, and the ratio of the number of strategy changes in the window to the total time length of the window is calculated as the strategy conversion frequency to form a time sequence decision feature parameter matrix.

[0086] The resource threshold (such as the minimum reserve of tank fuel and the threshold of remaining ammunition) and the round progress threshold (such as the progress node of the 8th round in the round deduction to trigger the reinforcement rule) of the current scene are called from the wargame rule library. The conditions of "entity current resource value ≥ resource threshold" and "current round number ≥ round progress threshold" are determined as meeting the rule condition (coded as 1), and vice versa (coded as 0) to generate the binary feature code of the rule trigger feature.

[0087] Further, the min-max normalization method is used to map the spatial confrontation feature, the time sequence decision feature and the rule triggering feature parameter into the interval of 0 to 1 by subtracting the minimum value of the feature in all entities from the actual value of the feature parameter and then dividing the difference between the maximum value and the minimum value of the feature, and the multi-modal fusion feature vector (the vector dimension is the product of the entity number and the total feature dimension, for example, when the spatial confrontation feature contains 3 dimensions, the time sequence decision feature contains 2 dimensions, and the rule triggering feature contains 3 dimensions, the total feature dimension is 8) is formed by splicing and transmitted to the fusion reasoning module 220 in real time.

[0088] In the embodiment, the fusion reasoning module 220 generates an initial confrontation decision scheme through feature weighted fusion and knowledge graph reasoning, specifically including:

[0089] The attention mechanism is used to dynamically weight the fusion feature vector output by the multi-modal feature analysis module 210, wherein the spatial feature weight linearly increases with the entity interaction strength, and the time sequence feature weight exponentially increases with the action continuity index;

[0090] The pre-trained model is used to construct a war game confrontation knowledge graph, and the graph nodes include entity types, action strategies and rule clauses. The optimal decision logic of the historical confrontation case is matched by calculating the cosine similarity of the nodes.

[0091] According to the reasoning result, an initial confrontation decision scheme including action priority, resource allocation ratio and strategy alternative sequence is output, and the scheme is stored in a structured data format, and the data fields include decision ID, target entity ID and execution time sequence label.

[0092] Specifically, the fusion reasoning module 220 of the embodiment takes the multi-modal fusion feature vector output by the multi-modal feature analysis module 210 as input, generates an initial confrontation decision scheme through feature weighted fusion and knowledge graph reasoning, and specifically includes:

[0093] The entity interaction strength is quantified (represented by the reciprocal of the relative distance between entities, that is, the smaller the relative distance, the greater the interaction strength value), so that the weight corresponding to the spatial feature linearly increases with the increase of the interaction strength value. At the same time, the action continuity is obtained by calculating the ratio of the number of actions using the same strategy in the sliding window to the total number of actions in the window, so that the weight corresponding to the time sequence feature exponentially increases with the increase of the continuity value, and the dynamic weighting of the multi-modal fusion feature vector is completed.

[0094] Based on the "military field BERT pre-training model" (publicly disclosed wargame text data set fine-tuning optimization), a wargame confrontation knowledge graph is constructed. The nodes in the graph include "entity types" (such as main battle tanks, fighter jets, and infantry tanks), "action strategies" (such as frontal assault, flanking encirclement, and defense standby), and "rule provisions" (such as "armored entities prefer to attack enemy armored targets" and "airborne entities are prohibited from entering air defense fire coverage areas"). The edges between the nodes represent the association relationship (such as the "applicable strategy" edge connecting the "main battle tank" node and the "frontal assault" node). By calculating the cosine similarity of the current weighted feature vector and the historical confrontation case feature vector in the knowledge graph (i.e., calculating the dot product of the two vectors and then dividing by the product of the two vector lengths), the decision logic corresponding to the highest similarity historical case is selected as the optimal decision logic.

[0095] According to the optimal decision logic, an initial confrontation decision scheme is output. The scheme includes action priority (such as "tank No. 1 attacks enemy tank No. 2 first, and then performs a defense task"), resource allocation ratio (such as "allocate 30% of the ammunition reserve to tank No. 1 and 25% to tank No. 2"), and strategy alternative sequence (such as "prefer to use the flanking encirclement strategy, switch to the frontal assault strategy if the terrain is obstructed, and switch to the defense standby strategy if the frontal assault is blocked"). The scheme is stored in a JSON structured format, with data fields including decision ID, target entity ID, and execution timing label, and is transmitted synchronously to the wargame rule verification module 230.

[0096] It can be understood that the "military field BERT pre-training model" of the embodiment is based on the general BERT architecture and is trained in two stages. At the same time, the fine-tuning stage uses the public data set: the wargame training cases published by the National Defense University, containing tens of thousands of verified deduction scenarios, situation descriptions, and decision records. The fine-tuning uses cross-validation to optimize parameters, and the model decision logic is completely generated by data induction. At the same time, the fine-tuning of the military field BERT pre-training model is based on the public and 100,000-entry wargame deduction text data set, which covers typical wargame scenarios such as tactical commands, equipment interaction, and terrain influence, with a uniform character length of 512 per sample. During fine-tuning, the cosine annealing learning rate strategy is used, with an initial learning rate of , the learning rate is attenuated to of the current value every 20 iterations, the total number of iterations is 100, the batch size is set to 32, the optimizer is AdamW, and the weight decay coefficient is .

[0097] Further, for the entity interaction intensity, it is represented by the reciprocal of the relative distance between the two entities. To avoid the interaction intensity being unbounded when the relative distance approaches 0, it is constrained based on mathematical normalization rules: let the relative distance be (and m, the interaction intensity is considered as 0 when the distance exceeds the range), the calculation formula of the normalized interaction intensity is defined as m, if then the value is forced to be , at this time ), so that is constrained in the interval (0, 1], and the processing is based on the monotonicity and boundedness of the mathematical function, which guarantees the stability of the spatial feature weight calculation.

[0098] In the embodiment, the wargame rule verification module 230 realizes decision compliance guarantee through constraint matching and gradient correction, specifically including:

[0099] Retrieve the constraint conditions matched with the current scene from the wargame rule library, including the performance boundaries of the wargame confrontation entity and the scene rules;

[0100] Compare the parameters of the initial confrontation decision scheme with the constraint conditions item by item, and determine whether there are rule violation items such as parameter out-of-bound and rule conflict;

[0101] For the rule violation items, use the gradient adjustment method for correction, wherein for the performance boundary violation items, perform parameter truncation according to the maximum allowed value, and for the scene rule violation items, re-allocate resources and adjust the action time sequence according to the rule priority, and output the confrontation decision parameters after correction that conform to the wargame deduction logic.

[0102] Specifically, the wargame rule verification module 230 realizes decision compliance guarantee through constraint matching and gradient correction, specifically including:

[0103] According to the current deduction scene label (such as “urban street fighting scene”, “plateau mountain combat scene”, “sea cooperative confrontation scene”, etc.), retrieve the constraint conditions matched with the current scene from the wargame rule library, wherein the performance boundaries of the wargame confrontation entity include “maximum ammunition carrying capacity of main battle tank 12 rounds”, “maximum combat radius of fighter 800 kilometers”, “maximum driving speed of infantry fighting vehicle 60 kilometers / hour”, etc., and the scene rules include “air units are prohibited from low-altitude attack in urban street fighting scene”, “tank moving speed is calculated at 80% of the reference speed in plateau mountain combat scene”, “destroyer needs to keep 5-10 kilometers of cooperative distance with frigate in sea cooperative confrontation scene”, etc.;

[0104] ​​Compare each parameter in the initial countermeasure decision scheme with the retrieved constraints one by one, determine whether there are any rule violation items such as parameter out-of-bound (e.g., "the ammunition allocation of Tank No. 1 in the initial scheme is 4 rounds, which exceeds its maximum ammunition carrying capacity of 3 rounds") and rule conflict (e.g., "the fighter aircraft in the initial scheme performs a low-altitude attack task, which conflicts with the prohibition rule of urban street fighting scenario"), and record the specific type (performance boundary violation or scenario rule violation) and associated parameters of the rule violation item;

[0105] For the rule violation item, use the gradient adjustment method for correction. For the performance boundary violation item, perform truncation processing on the violation parameter according to the maximum allowed value of the entity performance boundary (e.g., adjust the ammunition allocation of Tank No. 1 from 4 rounds to its maximum allowed carrying capacity of 3 rounds). For the scenario rule violation item, sort the rules according to their priority (e.g., scenario-specific rules have higher priority than general countermeasure rules), re-allocate related resources and adjust the action timing (e.g., "due to the higher priority of the urban street fighting scenario rule than the air attack strategy, change the action strategy of the fighter aircraft from low-altitude attack to high-altitude support, and simultaneously adjust the 15% fuel resources originally planned for low-altitude attack to fuel allocation for high-altitude support task"). Output the countermeasure decision parameters that comply with the logic of the war game after correction, and transmit them to the deduction execution unit 300 in real time, while feeding back to the dynamic interaction unit 400 for state synchronization between units.

[0106] In addition, the war game rule library realizes automatic updating through the "scenario tag-rule subset" association mechanism: a unique tag is configured for each deduction scenario (e.g., urban street fighting, highland, and naval formation), which is bound to the scenario characteristics such as terrain, climate, and equipment type; the rule library is divided into corresponding subsets according to the scenario tag (e.g., the "urban street fighting" subset contains building shelter judgment and street fighting movement speed correction rules, and the "highland" subset contains oxygen content impact on personnel physical strength and mountain off-road mobility rules). When the digital twin modeling unit 100 detects changes in scenario characteristics through terrain data analysis, environmental parameter input, and other technical means, the digital twin modeling unit 100 transmits the scenario characteristic change information to the deduction execution unit 300, which extracts the core characteristics of the new scenario and matches the corresponding tag through its built-in rule calling function module, then closes the calling interface of the original scenario rule subset and opens the calling interface of the new scenario rule subset, and synchronously loads the new rule subset. This mechanism is based on data matching and signal transmission logic between modules, relies on computer technology to realize the dynamic adaptation of scenarios and rules, and ensures that the deduction behavior complies with the current scenario constraints.

[0107] It should be noted that in the large model decision unit 200 of the embodiment, the multi-modal feature analysis module 210, the fusion reasoning module 220, and the war game rule verification module 230 realize real-time interaction through a system internal high-speed data bus interface. The multi-modal feature analysis module 210 updates the multi-modal fusion feature vector once every 1 deduction time unit (1 second for real-time deduction, and 1 round for round-based deduction). The fusion reasoning module 220 synchronously outputs new decision logic based on the updated feature vector. The war game rule verification module 230 performs real-time compliance verification and correction on the new decision logic, ensuring that the generated countermeasures always conform to the war game scenario state and war game rule requirements, and fully exerting its support role as the "intelligent decision core" of war game deduction.

[0108] The deduction execution unit 300 adopts a war game space-time coupling deduction mechanism. Based on the war game virtual model of the digital twin modeling unit 100 and the countermeasures of the large model decision unit 200, it drives the countermeasure process according to the war game deduction time sequence, simulates the war game countermeasure result through space-time correlation calculation, and internally builds a dynamic correction mechanism to adjust the deduction path according to the deviation between the virtual model and the expected state of the war game deduction.

[0109] In the embodiment, the deduction execution unit 300 includes a time sequence advancing module 310, a space interaction calculation module 320, and a countermeasure result synthesis module 330.

[0110] The time sequence advancing module 310 is used to drive the countermeasure process according to the war game deduction time sequence. For round-based deduction, the round mark is taken as the time sequence unit. At the start time of the first round, the countermeasures output by the large model decision unit 200 are loaded. For real-time deduction, the time step is taken as the time sequence unit. The state data of the war game countermeasures entity is updated synchronously within the interval ; wherein, represents the starting time point of real-time deduction, represents the ending time point obtained by adding the time step to the starting time point .

[0111] Specifically, the time sequence advancing module 310 specifically includes operations for driving the countermeasure process according to the war game deduction time sequence. First, the type of deduction is determined according to the deduction requirements:

[0112] For round-based deduction, the round mark (such as , representing the first, second, and third rounds) is taken as the time sequence unit. At the start time of the At the beginning of the round, the countermeasures parameter (including action priority and resource allocation ratio) output by the large model decision unit 200 is called through the internal data interface of the system, loaded into the state control module of each countermeasure entity, and the entity action target and resource quota of this round are initialized;

[0113] For real-time deduction, the time step is taken as the time unit, and the real-time state data (including position coordinates, resource consumption, and action progress) of each entity is synchronously collected through a high-frequency data sampling interface and updated to the system state database within the time interval to ensure that the entity state matches the deduction time in real time, where is the starting time point of real-time deduction, is the end time point of the time step, and the synchronous update frequency is consistent with .

[0114] The space interaction calculation module 320 is used to process the space interaction of the countermeasure entity, and the entity coordinates are obtained based on the virtual model of the countermeasure entity in the digital twin modeling unit 100, where represents the entity number, and the terrain parameter is the terrain slope, the attack effect value and the movement resistance degree are calculated; wherein the attack effect value is associated with the value of the firepower parameter , the entity distance , and the terrain correction coefficient ; the movement resistance degree is associated with the terrain type coefficient and the elevation difference .

[0115] Specifically, the space interaction calculation module 320 processes the space interaction of the countermeasure entity, and first calls the entity coordinates ( is the unique number of the entity, corresponding to different countermeasure entities) and the terrain parameter (the terrain slope, the value range ) from the geometric feature mapping module 110 of the digital twin modeling unit 100, and then calculates the attack effect value and the movement resistance degree through the formula: the formula for calculating the attack effect value is , wherein is the attack effect value (the larger the value, the stronger the attack effect), Firepower strength parameter of the attacking entity (retrieved from the entity type configuration table, such as 5-10 for the main gun of a main battle tank and 1-3 for the firepower strength of infantry weapons), Defensive coefficient of the defending entity (retrieved from the entity type configuration table, such as 3-8 for an armored vehicle and 0.5-1.5 for infantry), Entity distance (calculated by the coordinates of the two entities and According to the Euclidean distance formula: ), Terrain correction coefficient (retrieved from the state tag synchronization module 130 of the digital twin modeling unit 100, such as 1.0 for open terrain, 0.8 for semi-sheltered terrain, and 0.3 for completely sheltered terrain), the calculation logic is that the attack effect is directly proportional to and , and inversely proportional to ,

[0116] Further, the formula for calculating the degree of movement obstruction is , where is the degree of movement obstruction (with a value of 0.2-5.0, the larger the value, the more difficult the movement), is the terrain type coefficient (unitless, retrieved from the terrain parameter library, such as 0.2 for plains, 0.5 for hills, 0.7 for mountains, and 1.5 for swamps), is the elevation difference between the current position and the target position (the calculation formula is: , is the current elevation of the entity, is the elevation of the target position, is the absolute value of the elevation difference), is the reference elevation difference (industry commonly used reference value is 10 meters), the calculation logic is that the movement obstruction is determined by the terrain basic obstruction and the elevation difference additional obstruction, the larger the elevation difference, the stronger the obstruction.

[0117] The confrontation result synthesis module 330 is used to integrate the time series data and spatial interaction results to generate the Kriegsspiel confrontation results, where the entity life value update is associated with the attack effect value , the resource change is associated with the interaction frequency, the rule trigger result is associated with the threshold determination of the attack effect value , and all results are stored with the round mark as the time series tag.

[0118] Specifically, for the entity life value update dimension, the formula is where HPinit is the initial HP value at the beginning of the time sequence (obtained from the entity type configuration table, such as an armored vehicle 100 or an infantry 50), , the attack effect value output by the space interaction calculation module 320, , the HP decay coefficient (unitless, armored entity 0.8, infantry entity 1.2), the calculation logic is that the HP decreases as the HP increases, and the decay rate is distinguished by ; ;

[0119] For the resource change dimension, the formula is , where is the updated resource value (such as "shots" for ammunition and "liters" for fuel), is the current resource value at the beginning of the time sequence, is the entity interaction frequency within the time sequence (such as the number of attacks and the number of movements, counted by the time sequence advancing module 310), is the unit interaction resource consumption (obtained from the entity type configuration table, such as 1 shot per attack ), the calculation logic is that the resource consumption is proportional to the interaction frequency;

[0120] For the rule trigger result dimension, the trigger threshold is obtained from the wargame rule library (such as triggering "entity serious injury", triggering "entity destruction"), and the current value is compared to determine whether the rule is triggered; all results are time-labeled with round markers (round-based) or time points (real-time), stored in JSON format.

[0121] In this embodiment, the dynamic correction mechanism of the deduction execution unit 300 further includes a deviation monitoring module 340, a correction strategy generation module 350, and a path adjustment execution module 360, wherein:

[0122] The deviation monitoring module 340 is used to calculate the deviation of the wargame virtual model from the expected state of the deduction, and the position deviation amount is obtained by comparing the entity position output by the digital twin modeling unit 100 with the expected position , and the resource deviation amount is obtained by comparing the resource value with the expected resource ;

[0123] Specifically, in implementation, the entity expected state parameters (including expected position , expected resource ), and the current position of the entity is obtained from the digital twin modeling unit 100 and the current resource value : the formula for calculating the position deviation amount is wherein is the spatial deviation amount, and the calculation logic is to quantify the spatial deviation degree by the Euclidean distance, The greater the deviation, the more serious the deviation; the formula for calculating the resource deviation amount is wherein is the absolute resource deviation amount (the unit is consistent with the resource type), and the calculation logic is to quantify the resource difference by the absolute value, The greater the difference, the more serious the difference.

[0124] The correction strategy generation module 350 is used to formulate the deduction path adjustment scheme, , the threshold value corresponding to the resource deviation , when and , the current path is maintained; otherwise, the position compensation value , the resource compensation value , and the timing compensation step size are generated.

[0125] Specifically, first, the deviation threshold value is set according to the entity type, and then the current deviation amount is compared with the threshold value: if and , the current path is maintained; if the threshold value is exceeded, the position compensation value (respectively , , axis compensation amount, positive value adjusts to the positive direction, negative value adjusts to the negative direction), resource compensation value (the unit is consistent with the resource type, positive value supplements the resource, negative value reduces the resource), and timing compensation step size are calculated. , is the corrected timing step size, the greater the deviation , the smaller the value, which improves the adjustment accuracy.

[0126] The path adjustment execution module 360 is used to adjust the deduction path by applying the correction parameters, and the entity coordinates are updated to wherein , , are the current spatial coordinates of the entity, , , are respectively axis, axis, ​​The position compensation value in the axial direction; the resource value is updated as , the subsequent deduction timing adjustment is , and the adjustment result is fed back to the digital twin modeling unit 100 to update the wargame virtual model.

[0127] Specifically, the entity current coordinate is updated as , the current resource value is updated as , the subsequent deduction timing unit is adjusted as , and the time interval is changed to ; the adjusted entity state data is fed back to the state tag synchronization module 130 of the digital twin modeling unit 100 in real time, used to update the wargame virtual model, ensuring that the virtual model is consistent with the deduction expected state.

[0128] It should be noted that in the deduction execution unit 300 of the embodiment, the timing advancing module 310, the space interaction calculation module 320, and the confrontation result synthesis module 330 are real-time linked according to the "timing driving-space interaction-result generation" process, and the deviation monitoring module 340, the correction strategy generation module 350, and the path adjustment execution module 360 monitor the deduction deviation and dynamically control throughout the process. All modules realize data interaction through a system high-speed data bus, ensuring the continuity, accuracy, and controllability of the wargame confrontation process, and fully play the role of the "process driving core".

[0129] The dynamic interaction unit 400 is used to establish a two-way channel of wargame data between the digital twin modeling unit 100, the large model decision unit 200, and the deduction execution unit 300, realize low-delay interactive transmission of wargame scene feature parameters, confrontation decision instructions, and deduction state data, and guarantee the real-time and continuity of wargame deduction.

[0130] In the embodiment, the dynamic interaction unit 400 includes a data acquisition module 410, a low-delay transmission module 420, and a state synchronization module 430, wherein:

[0131] The data acquisition module 410 is used to acquire wargame scene feature parameters from the digital twin modeling unit 100, acquire confrontation decision instructions from the large model decision unit 200, and acquire deduction state data from the deduction execution unit 300;

[0132] Specifically, the multi-source data acquisition is completed through the internal standardized interface of the system: the scene characteristic parameters such as entity coordinates, terrain data, and entity state labels are acquired from the digital twin modeling unit 100; the counter decision instructions such as action priority, resource allocation ratio, and compliance decision parameters are acquired from the large model decision unit 200; and the deduction state data such as current time sequence mark, attack effect value, entity life value, and resource value are acquired from the deduction execution unit 300. The acquisition adopts a trigger strategy, and each unit data is automatically triggered to collect when updated, the frequency is synchronized with the unit data update frequency, and the collected data is uniformly converted into a standardized format and temporarily stored in the local cache area.

[0133] The low-delay transmission module 420 is used to build a two-way transmission channel for the war game data, adopts a lightweight encapsulation transmission technology based on the UDP protocol, and realizes low-delay interactive transmission of the war game scene characteristic parameters, the counter decision instructions, and the deduction state data.

[0134] Specifically, the low-delay transmission module 420 designs a lightweight transmission mechanism based on the UDP protocol: the data encapsulation adopts a “protocol header (including data type, sending unit identifier, and time stamp) + compressed data body” structure to reduce the transmission amount; the transmission adopts a priority queue, the counter decision instructions are set as the highest priority to ensure that the key data is transmitted in priority; the packet loss processing is realized through “key data retransmission + non-key data redundancy”, the key data is retransmitted when missing, and the static data is attached with redundancy to reduce the influence of packet loss. The channel communicates through a preset port, and automatically switches to a backup path when the delay exceeds the threshold.

[0135] The state synchronization module 430 is used to synchronize the states of the war game virtual scene of the digital twin modeling unit 100, the decision generation logic of the large model decision unit 200, and the deduction process of the deduction execution unit 300 according to the data transmitted by the low-delay transmission module 420, so as to guarantee the real-time and continuity of the war game deduction.

[0136] Specifically, the state synchronization module 430 realizes the operation of multi-unit state cooperation according to the transmission data, and implements the process of “data analysis-conflict determination-synchronous execution”: the transmission data is analyzed and distributed to the corresponding buffer according to the unit type; the conflict determination is based on the data of the deduction execution unit 300 (calibrated dynamically) and the verification result of the war game rules; in the synchronous execution stage, the corresponding data is pushed to each unit, the virtual scene display, the decision logic, and the deduction process are updated, and the consistency of the states of each unit is ensured.

[0137] It should be noted that in the dynamic interaction unit 400 of the embodiment, the data acquisition module 410, the low-delay transmission module 420, and the state synchronization module 430 work cooperatively in a closed-loop process of “acquisition-transmission-synchronization”, and through a standardized data format and a priority transmission mechanism, real-time data interaction between the digital twin modeling unit 100, the large model decision unit 200, and the deduction execution unit 300 is realized, the consistency and coherence of the scene, decision, and execution state in the war game deduction process are ensured, and the role of the “data interaction hub” is fully played.

[0138] The deduction result evaluation unit 500 is further included, which is used for evaluating the effectiveness of the war game confrontation result output by the deduction execution unit 300, including a result and benchmark acquisition module 510, a multi-dimensional evaluation module 520, and an evaluation result feedback module 530.

[0139] The result and benchmark acquisition module 510 is used for acquiring war game confrontation result data from the deduction execution unit 300, acquiring virtual model state data of the corresponding deduction stage from the digital twin modeling unit 100, and simultaneously acquiring data interaction delay records of the corresponding deduction stage from the dynamic interaction unit 400, as evaluation benchmark data;

[0140] Specifically, the result and benchmark acquisition module 510 acquires data in a targeted manner by a deduction stage through a system internal data interface: the war game confrontation result data acquired from the deduction execution unit 300 includes entity life value update records, resource change details, rule trigger results, and corresponding time sequence labels; the virtual model state data of the corresponding deduction stage acquired from the digital twin modeling unit 100 includes entity coordinates, terrain parameters, and entity state labels (strictly corresponding to the time sequence labels of the confrontation result data); and the data interaction delay records acquired from the dynamic interaction unit 400 include the timestamp difference (the difference between the sending timestamp and the receiving timestamp) of data transmission between units and the time sequence node at which the delay occurs. The acquisition timing is synchronized with the result output of the deduction execution unit 300 (for round-based, at the end of each round, and for real-time, at the end of each time step), the acquired data is uniformly associated with time sequence labels, and is temporarily stored in a structured table form in an evaluation benchmark database.

[0141] The multi-dimensional evaluation module 520 is used for judging the effectiveness of the confrontation result against existing standards, specifically including: judging the compliance of the deduction logic according to the compliance standard output by the war game rule verification module 230, judging the synchronization accuracy of the virtual scene according to the sub-second level synchronization requirement of the time sequence calibration execution module 160, and judging the real-time performance of data interaction according to the low-delay requirement of the dynamic interaction unit 400. If all three items meet the standards, the result is determined to be valid, and if any item does not meet the standard, the result is marked as a to-be-optimized result and the not-meeting item is recorded;

[0142] Specifically, the multi-dimensional evaluation module 520 judges the effectiveness of the confrontation result against the standard dimension. When implemented, three core indicators are evaluated, specifically including:

[0143] Logical compliance evaluation. Based on the compliance standards output by the wargame rule verification module 230, the entity actions, resource consumption, and rule triggers in the confrontation result are compared one by one to determine whether they meet the preset rules (such as "whether the attack range of armored units is within the rule-limited distance" and "whether the resource consumption matches the interaction frequency"). The specific content and corresponding time sequence of non-compliance items are recorded.

[0144] Synchronization accuracy evaluation. Based on the sub-second synchronization requirement of the time sequence calibration execution module 160, the time deviation (difference between state update time and result generation time) of the virtual model state data of the digital twin modeling unit 100 and the confrontation result data of the deduction execution unit 300 under the same time sequence label is calculated to determine whether the deviation is within the sub-second range.

[0145] Interaction real-time evaluation. Based on the low-latency requirement of the dynamic interaction unit 400, the number of times and the proportion of data interaction delay records that exceed the preset delay threshold are counted to determine whether the data transmission meets the real-time requirement.

[0146] Further, the evaluation conclusion is generated according to "all three indicators meet the standard, and any one does not meet the standard is marked as a result to be optimized". The specific parameters of non-compliance items (such as non-compliant rule clauses, excessive time deviation values, and delay threshold transmission times) are recorded.

[0147] The evaluation result feedback module 530 is used to store the effective evaluation results in the wargame deduction result library, and to feed back the results to be optimized and non-compliance items to the large model decision unit 200 and the digital twin modeling unit 100. The information fed back to the large model decision unit 200 is used to optimize the subsequent confrontation decision parameter generation logic, and the information fed back to the digital twin modeling unit 100 is used to assist in adjusting the model accuracy level or time sequence calibration frequency.

[0148] Specifically, the evaluation result feedback module 530 is used to classify and process evaluation results and implement closed-loop feedback. When implemented, the results are processed according to their types, specifically including:

[0149] The effective evaluation result is stored in the war game result library in the format of "time sequence label + evaluation conclusion + key indicator data", serving as a reference case for subsequent deduction; the to-be-optimized result and the substandard item are fed back by unit type - the information fed back to the large model decision unit 200 includes the decision parameter corresponding to the non-compliance item (such as "the specific value of the resource allocation ratio exceeding the rule threshold") and the influence analysis of the interaction delay on decision generation, which is used to fuse the parameter weight in the optimization decision logic of the reasoning module 220; the information fed back to the digital twin modeling unit 100 includes the time sequence node with synchronization deviation exceeding the standard and the corresponding virtual model state data, which is used to assist the geometric feature mapping module 110 to adjust the model precision level (such as increasing the grid precision in the deviation concentrated area) or the time sequence calibration execution module 160 to optimize the calibration frequency. The feedback data is pushed through the low-delay transmission channel of the dynamic interaction unit 400, ensuring that the relevant units obtain the optimization basis in time.

[0150] It should be noted that in the deduction result evaluation unit 500 of the embodiment, the result and the reference collection module 510, the multi-dimensional evaluation module 520 and the evaluation result feedback module 530 work cooperatively according to the process of "data collection-multi-dimensional evaluation-result feedback", realize accurate correspondence between the evaluation and the deduction process through the associated time sequence label, and continuously optimize the large model decision unit 200 and the digital twin modeling unit 100 through the directional feedback mechanism, so as to fully play the role of "effectiveness verification core".

[0151] Those skilled in the art can understand that the processes for implementing all or part of the steps of the above embodiments can be completed by hardware or by programs instructing relevant hardware.

[0152] The basic principles, main features and advantages of the present application are shown and described above. Those skilled in the art should understand that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only preferred examples of the present application and are not intended to limit the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. An intelligent wargaming simulation system based on a digital twin fusion model, characterized in that, include: The digital twin modeling unit adopts a dynamic precision adaptive modeling mechanism, which constructs a hierarchical virtual model of the wargame scenario through the fusion of multi-source heterogeneous data. The hierarchical virtual model automatically adjusts the model precision level according to the granularity requirements of the wargame simulation, and achieves sub-second dynamic synchronization between the virtual model and the wargame simulation scenario through a time-series calibration algorithm. A large-scale model decision unit (LSM) receives wargame scenario state data output by the digital twin modeling unit. It employs a multimodal feature fusion and wargame rule constraint embedding mechanism to fuse the spatial adversarial features and temporal decision features of the wargame scenario. The MLM embeds the performance boundaries of the adversarial entities and scenario rules as constraints into a pre-trained model, generating adversarial decision parameters that conform to the wargame deduction logic. The MLM includes a multimodal feature parsing module, a fusion inference module, and a wargame rule verification module. The multimodal feature parsing module achieves structured extraction of wargame scenario features through feature quantization and standardization, specifically including: For the spatial adversarial features in the wargame scenario state data, a coordinate system transformation method is used to convert the absolute coordinates of adversarial entities into quantitative parameters such as relative distance, azimuth angle, and terrain occlusion coefficient; For time-series decision characteristics, the historical action sequence of adversarial entities is extracted by using a sliding window to extract time-series parameters such as action interval and strategy switching frequency; For rule triggering features, convert the rule conditions such as resource thresholds and round progress into binary feature codes; Furthermore, the spatial adversarial features, temporal decision features, and rule-triggered feature parameters are normalized to a unified data dimension to form a multimodal fusion feature vector; The simulation execution unit adopts a wargame spatiotemporal coupling simulation mechanism. Based on the wargame virtual model of the digital twin modeling unit and the adversarial decision parameters of the large model decision unit, it drives the adversarial process according to the wargame simulation sequence, simulates the wargame adversarial results through spatiotemporal correlation calculation, and has a built-in dynamic correction mechanism to adjust the simulation path according to the deviation between the virtual model and the expected state of the wargame simulation. The dynamic interaction unit is used to establish a two-way data channel for wargames between the digital twin modeling unit, the large model decision-making unit, and the simulation execution unit, so as to realize low-latency interactive transmission of wargame scene feature parameters, adversarial decision-making instructions, and simulation status data, and ensure the real-time performance and continuity of wargame simulation.

2. The intelligent war game simulation system based on a digital twin fusion model according to claim 1, characterized in that, The digital twin modeling unit includes a geometric feature mapping module, an attribute parameter binding module, and a state label synchronization module, wherein: The geometric feature mapping module is based on the 3D scanning data and parametric modeling rules of the wargame adversarial entities. It uses multi-detail level model loading technology to generate and store the 3D contour data and spatial coordinates of the wargame adversarial entities, and supports dynamic switching of precision from millimeter level to meter level. The attribute parameter binding module is based on the wargame simulation rule base and entity type configuration table. It uses dynamic parameter association technology to bind the inherent performance parameters of wargame adversaries to entity types and automatically updates the performance parameters when the state of the wargame adversaries changes. The state tag synchronization module uses real-time state triggering technology to mark the current state of the wargame adversaries based on the entity behavior result data output by the simulation execution unit.

3. The intelligent wargaming simulation system based on a digital twin fusion model according to claim 2, characterized in that, The digital twin modeling unit further includes a precision level division module, a precision adjustment trigger module, and a timing calibration execution module, wherein: The precision level division module divides the model precision into three levels based on the interaction frequency and spatial scale of the wargame adversarial entities and using a distance-precision mapping rule. The precision adjustment triggering module is based on the relative motion data of the wargame adversaries and uses threshold monitoring technology. When the relative motion state between the wargame adversaries meets the preset triggering conditions, the precision is automatically upgraded; when there is no interaction between the wargame adversaries for a preset duration, the precision is automatically downgraded. The timing calibration execution module collects data from the physical scene with timestamps and uses time deviation correction technology. It calculates the time deviation between the virtual model and the physical scene through a timing calibration algorithm. If the time deviation exceeds the preset acceptable range, the state parameters of the virtual model are corrected through linear interpolation to ensure the synchronization between the virtual model and the war game scenario.

4. The intelligent war game simulation system based on a digital twin fusion model according to claim 3, characterized in that, The fusion reasoning module generates an initial adversarial decision-making scheme through feature weighted fusion and knowledge graph reasoning, specifically including: An attention mechanism is used to dynamically weight the fused feature vector output by the multimodal feature parsing module, where the spatial feature weight increases linearly with the entity interaction intensity, and the temporal feature weight increases exponentially with the action coherence. A wargame adversarial knowledge graph is constructed based on a pre-trained model. The graph nodes contain entity types, action strategies, and rule clauses. The optimal decision logic for matching historical adversarial cases is calculated by the cosine similarity of the nodes. Based on the reasoning results, an initial adversarial decision-making scheme is output, which includes action priority, resource allocation ratio, and strategy alternative sequence. The scheme is stored in a structured data format, and the data fields include decision ID, target entity ID, and execution time sequence label.

5. The intelligent wargaming simulation system based on a digital twin fusion model according to claim 4, characterized in that, The wargame rule verification module ensures decision compliance through constraint matching and gradient correction, specifically including: Retrieve constraints that match the current scenario from the wargame rule base, including the performance boundaries of wargame adversaries and scenario rules; The parameters and constraints of the initial adversarial decision-making scheme are compared item by item to determine whether there are any violations such as parameter out-of-bounds or rule conflicts. The gradient adjustment method is used to correct violations. For performance boundary violations, the parameters are truncated according to the maximum allowable value. For scenario rule violations, resources are reallocated and the action sequence is adjusted according to the rule priority. After correction, the output adversarial decision parameters conform to the wargaming simulation logic.

6. The intelligent war game simulation system based on a digital twin fusion model according to claim 5, characterized in that, The simulation execution unit includes a time-series advancement module, a spatial interaction calculation module, and an adversarial result synthesis module, wherein: The timing progression module is used to drive the adversarial process according to the wargame simulation sequence, and to mark round-based simulations with round markers. As a time series unit, in the first At the start of each round, the adversarial decision parameters output by the large model decision unit are loaded; real-time simulations are performed using time steps. As a time series unit, in The state data of the wargame adversaries is updated synchronously within the interval; among which... Indicates the start time point of the real-time simulation. Indicates starting time point As a baseline, plus a time step The final time point obtained later; The spatial interaction computing module is used to process the spatial interaction of wargame adversaries, and obtains entity coordinates based on the wargame virtual model of the digital twin modeling unit. ,in Indicates the entity number and terrain parameters. That is, the terrain slope, used to calculate the attack effect value. and degree of movement obstruction Among them, attack effect value Related firepower parameters Values, entity spacing Terrain correction factor Degree of movement obstruction Correlation of terrain type coefficients Elevation difference ; The adversarial result synthesis module is used to integrate temporal data and spatial interaction results to generate wargame adversarial results, wherein entity life values ​​are updated in relation to attack effect values. Resource changes are correlated with interaction frequency, and rule triggering results are correlated with attack effectiveness values. Threshold determination, all results are marked with rounds. Stored as time-series tags.

7. The intelligent war game simulation system based on a digital twin fusion model according to claim 6, characterized in that, The dynamic correction mechanism of the simulation execution unit also includes a deviation monitoring module, a correction strategy generation module, and a path adjustment execution module, wherein: The deviation monitoring module is used to calculate the deviation between the wargame virtual model and the expected state in the simulation, and compare the entity positions output by the digital twin modeling unit. Compared to the expected location Receive position deviation Compare resource values With expected resources Obtain resource deviation ; The correction strategy generation module is used to formulate a simulation path adjustment plan and set thresholds corresponding to position deviations. Thresholds corresponding to resource deviation ,when and Maintain the current path if necessary; otherwise, generate a location compensation value. Resource compensation value and timing compensation step size ; The path adjustment execution module is used to apply correction parameters to adjust the inference path and update the entity coordinates. ,in , , The current spatial coordinates of the entity. , , They are respectively axis, axis, Position compensation value in the axial direction; resource value updated to The subsequent simulation sequence was adjusted to The adjustment results are fed back to the digital twin modeling unit to update the wargame virtual model.

8. The intelligent war game simulation system based on a digital twin fusion model according to claim 7, characterized in that, The dynamic interaction unit includes a data acquisition module, a low-latency transmission module, and a status synchronization module, wherein: The data acquisition module is used to collect wargame scenario feature parameters from the digital twin modeling unit, collect adversarial decision instructions from the large model decision unit, and collect simulation status data from the simulation execution unit. The low-latency transmission module is used to construct a two-way transmission channel for wargame data. It adopts a lightweight encapsulation transmission technology based on the UDP protocol to achieve low-latency interactive transmission of wargame scene feature parameters, adversarial decision instructions, and simulation status data. The state synchronization module is used to synchronize the states of the wargame virtual scene of the digital twin modeling unit, the decision generation logic of the large model decision unit, and the simulation process of the simulation execution unit based on the data transmitted by the low-latency transmission module, so as to ensure the real-time performance and continuity of the wargame simulation.

9. The intelligent war game simulation system based on a digital twin fusion model according to claim 8, characterized in that, It also includes a wargaming result evaluation unit, which is used to evaluate the effectiveness of the wargaming results output by the wargaming execution unit. This unit includes a result and benchmark acquisition module, a multi-dimensional evaluation module, and an evaluation result feedback module, wherein: The results and benchmark acquisition module is used to acquire wargame confrontation result data from the simulation execution unit, acquire wargame virtual model status data corresponding to the simulation stage from the digital twin modeling unit, and acquire data interaction delay records corresponding to the simulation stage from the dynamic interaction unit as evaluation benchmark data. The multi-dimensional evaluation module is used to judge the validity of the adversarial results by comparing them with existing standards. Specifically, it includes: judging the compliance of the deduction logic by the compliance standards output by the wargame rule verification module; judging the synchronization accuracy of the virtual scene by the sub-second synchronization requirements of the timing calibration execution module; and judging the real-time performance of data interaction by the low latency requirements of the dynamic interaction unit. If all three items meet the standards, the result is deemed valid. If any item does not meet the standards, it is marked as a result to be optimized and the non-compliant item is recorded. The evaluation result feedback module is used to store the effective evaluation results in the war game simulation result library, and to feed back the results to be optimized and the substandard items to the large model decision unit and the digital twin modeling unit. The information fed back to the large model decision unit is used to optimize the subsequent adversarial decision parameter generation logic, and the information fed back to the digital twin modeling unit is used to assist in adjusting the model accuracy level or the timing calibration frequency.

Images