Multi-dimensional data-driven military simulation training precise evaluation decision system

Abstract

This invention discloses a multi-dimensional data-driven precision assessment and decision-making system for military simulation training, relating to the field of military simulation training technology. The system achieves digital modeling of all training elements by constructing a multi-source state parameter set. Based on the temporal distribution characteristics of action and physiological data, combined with environmental factor analysis, it identifies individual fatigue / stress characteristics and generates a preliminary individual risk identifier R1 with type labels and causal attributes. Combining changes in environmental sensor values ​​and historical training data, R1 is updated to the R2 subtype. When R2 indicates a high-risk state, the system analyzes the rate of change in team interaction data, combining temporal priority tests and Granger causality tests to form a team collaboration early warning feature ST. By constructing an individual-team collaboration risk judgment model and calculating the risk coupling factor CF, it determines a joint high-risk state and generates decision instructions. The system dynamically adjusts the difficulty / pacing of training scenarios, achieving closed-loop control and intelligent decision-making throughout the training process.

Description

Technical Field

[0001] This invention belongs to the field of military simulation training technology, specifically involving a multi-dimensional data-driven precision evaluation and decision-making system for military simulation training. Background Technology

[0002] In the field of military simulation training, traditional assessment methods often focus on single-dimensional data or post-event debriefing, making it difficult to achieve real-time status perception and dynamic risk warning during the training process. Existing technologies often lack the ability to comprehensively collect and analyze data on trainees' physiological states, movement patterns, environmental factors, and team interactions, resulting in a lag in the early identification of individual performance and team collaboration risks, making it difficult to support the generation of precise intervention decisions.

[0003] As training scenarios become more complex and the intensity of adversarial competition increases, traditional systems exhibit significant shortcomings in dynamically adjusting training pace, integrating multi-source sensor data, and identifying environment-induced risks. For example, the impact of environmental factors on individual performance is often masked by global parameter analysis, and the causal attribution of team collaboration failures lacks temporal priority testing and quantitative analysis methods, resulting in insufficiently targeted intervention measures and difficulty in effectively blocking risk transmission paths.

[0004] To overcome the aforementioned limitations, this invention proposes a multi-dimensional data-driven precision assessment and decision-making system for military simulation training. This system constructs a multi-source state parameter set by integrating action, physiological, environmental, and interaction data. Combined with individualized environmental sensitivity analysis and team collaboration link monitoring, it achieves dynamic assessment throughout the entire process, from individual risk identification to joint response decision-making. Finally, through differentiated intervention strategies and auxiliary support modules, it improves the accuracy of training assessment and the timeliness of decision-making, supporting real-time risk management and capability enhancement in complex training scenarios. Summary of the Invention

[0005] To overcome the shortcomings and deficiencies of the existing technology, the present invention adopts the following technical solution: A multi-dimensional data-driven precision evaluation and decision-making system for military simulation training. The system workflow includes the following steps: S1. Collect multi-source sensing data from the training environment, trainees, and simulation equipment. After spatiotemporal alignment and standardization, construct a multi-source state parameter set. The multi-source sensing data includes motion sensor data, physiological sensor data, environmental sensor data, and interaction record data. The multi-source state parameter set is organized in time series form, containing corresponding combinations of state parameters, motion parameters, environmental parameters, and interaction parameters of the training unit at the same time. S2. Based on the action data and temporal distribution characteristics of the multi-source state parameter set, identify the action execution mode of the trainees, combine the correlation changes of physiological data and action signals, determine the individual fatigue or stress characteristics, and generate a preliminary individual risk label R1. S3. Based on the changes in environmental parameters of key areas of the training ground, combined with historical training data and mission scenarios, determine whether there are conditions that could induce tactical errors or coordination obstacles, and update the individual risk label to R2. S4. If the individual risk identifier R2 is in a high-risk state, analyze the change characteristics of the interactive data at the team communication node, compare it with the preset collaborative response threshold, identify the team collaborative failure trend, and form a team collaborative early warning feature. S5. Logically correlate individual risk identifier R2 with team collaborative early warning features to construct an individual-team collaborative risk judgment model. If the risk is determined to be a joint high-risk state, a joint response decision instruction will be generated. S6. Based on the joint response decision instructions, dynamically adjust the difficulty or pace of the training scenario, coordinate with the guidance system to implement targeted interventions, and activate the corresponding auxiliary training support modules.

[0006] Preferably, step S2 specifically includes: Filtering and feature extraction are performed on motion data and physiological data to reconstruct motion smoothness curves and physiological load curves; The completion time, accuracy deviation, and stability index of key action nodes are calculated, and the physiological load curve is differentially processed to obtain the physiological change rate. The physiological rate of change is compared with the preset normal fluctuation threshold, and continuous abnormal periods are marked. Combined with the amount of degeneration of the movement during the period, three risk types are classified: physiological-motor coordination abnormality, potential physiological overload, and skill or psychological error. Corresponding risk labels are assigned to form a preliminary individual risk label R1. Based on the correlation analysis between local environmental parameters and individual degradation, and combined with the comparison of the team's average performance under the same environment, it was determined that the individual risk causation attribute was either environmentally induced or due to individual factors.

[0007] Preferably, the process of updating the individual risk identifier in step S3 includes: Sliding window analysis was performed on the time series curves of environmental parameters to calculate the growth rate of adverse environmental factors. Combined with the unit's environmental adaptability threshold and the mission-environment matching coefficient, potential high-environment risk areas or time periods were marked. The spatiotemporal range corresponding to the initial individual risk identifier R1 is matched with potential high environmental risk areas / time periods. Combining individual environmental sensitivity with differences in the overall performance of the team, R2 is divided into environmentally driven individual risk R2a and individual intrinsic risk R2b. Both R2a and R2b retain the original risk type label and trigger attributes of R1 for subsequent differentiated collaborative analysis and intervention.

[0008] Preferably, the team collaborative early warning features formed in step S4 include: Extract key communication node interaction data from the team, and sample and calculate indicators such as instruction clarity, response latency, and information completeness; For the interaction links of high-risk individuals, calculate the rate of change in response latency and the rate of decline in information integrity to determine the trend of declining collaboration efficiency; By using temporal priority tests and Granger causality tests, the order and causal orientation of individual anomalies and team synergy decay were determined. Combining R2a or R2b subtypes, the early warning characteristics of team collaboration are divided into environment-team-dominated, environment-individual-dominated, team-induced, individual-drag, and bidirectional reinforcement early warning types, clarifying the causes of collaboration failure and the key points of intervention.

[0009] Preferably, the construction of the individual-team collaborative risk assessment model in step S5 includes: Individual risk identifiers R2, team collaborative early warning characteristics, and causal orientation are mapped into multi-dimensional state vectors; The degree of risk coupling is quantified from three dimensions: spatiotemporal correlation, causal strength, and impact diffusion, and the risk coupling factor is calculated. By comparing the risk coupling factors with preset thresholds and combining R2 subtypes with collaborative early warning feature types, the joint high-risk types are determined, including environment-team-dominated, environment-individual-dominated, team-individual-dominated, individual-team-dominated, and bidirectional collaborative joint high-risk. Based on the joint high-risk type, risk subject, and degree of coupling, a joint response decision instruction is generated, which includes real-time enhanced assistance, targeted intervention, activation of auxiliary support modules, and early warning records.

[0010] Preferably, the intervention logic for environment-dominated individual risk R2a is as follows: In response to the team-led collaborative early warning environment, keep the environmental interference unchanged, repair the team communication links, optimize the collaborative rhythm, and simultaneously implement AR tactical guidance and differentiated status adjustment for the affected groups; When dealing with environment-individual-driven collaborative early warning, keep the environmental interference unchanged, provide precise technical assistance to high-risk individuals, and simultaneously adjust the team's collaborative rhythm to adapt to the individual's recovery. When responding to enhanced two-way early warning, the system simultaneously provides team collaboration link repair and precise individual assistance to block the two-way transmission of risks without reducing the difficulty of the task.

[0011] Preferably, the intervention logic for individual intrinsic risk R2b is as follows: When a team-induced collaborative warning is issued, the focus should be on fixing the team collaboration process, maintaining the overall training difficulty, and providing individuals with AR guidance and status recovery prompts. When responding to individual drag-type collaborative warnings, action calibration, physiological regulation and psychological counseling are implemented for high-risk individuals to optimize team collaboration in order to reduce individual drag without reducing environmental pressure and task difficulty. When responding to enhanced two-way early warning, we simultaneously implement precise individual assistance and team collaborative adaptation to establish a two-way risk prevention mechanism.

[0012] Preferably, the execution process of the auxiliary training support module in step S6 includes: The AR display system overlays standard movement trajectories, tactical guidance, and collaborative situational information onto the trainees. Biofeedback devices are used to regulate breathing and guide calming of individuals under stress. During the intervention process, real-time data of individuals and teams are continuously collected to assess the recovery trend. Once the recovery threshold is met, the assisted mode is gradually withdrawn. Generate a debriefing report that includes risk type, causal test results, intervention measures and recovery effects, and push it to the commander's terminal simultaneously to update the trainees' competency profiles.

[0013] Preferably, the system includes: The data acquisition module is used to deploy motion sensors, physiological sensors, environmental sensors, and interactive recording devices to collect training data and construct a multi-source state parameter set; The individual risk identification module identifies individual risk types based on the temporal distribution characteristics of action data and physiological data, and generates a preliminary individual risk identifier R1 with attached type labels and trigger attributes; The environmental factors analysis module updates individual risk labels to the R2 subtype based on changes in environmental sensor values ​​and historical training data. The team collaboration early warning module analyzes the comparison results between the rate of change of interactive data and the collaborative response threshold to form the team collaboration early warning feature ST; The joint risk assessment module constructs an individual-team collaborative risk assessment model through logical correlation analysis and generates a joint response decision instruction D. The decision execution module adjusts the difficulty or pace of the training scenario based on decision instructions, and implements targeted interventions. The auxiliary support module provides differentiated auxiliary support. The dynamic monitoring and evaluation module continuously monitors the effectiveness of risk mitigation and the trend of status recovery.

[0014] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention avoids misjudging environmental factors and achieves a precise risk attribution system by comparing individualized environmental sensitivity analysis with historical tolerance thresholds. It does not simply correlate environmental changes with individual performance decline globally, but introduces intra-individual correlation coefficients and individual historical tolerance thresholds. By comparing the amount of performance degradation of an individual under environmental changes with the average performance of the team under the same environment, it can accurately distinguish whether the risk stems from the specific impact of the environment on the individual, the overall impact of the environment on the team, or a problem with the individual's own condition. This hierarchical attribution mechanism avoids misjudging individual capabilities due to environmental changes or missing intervention opportunities by ignoring individual specific sensitivities.

[0015] 2. This invention clarifies the transmission direction of individual anomalies and team collaboration failures through temporal priority tests and Granger causality tests, achieving a precise system for selecting intervention points. When linking individual risk with team collaboration warnings, a temporal priority test is implemented. By comparing the chronological order of the onset of individual high risk and the onset of team collaboration decline, combined with Granger causality tests, the transmission path of risk can be clearly determined. This enables subsequent decision-making instructions to accurately pinpoint the source and propagation chain of risk.

[0016] 3. This invention achieves a balance between training intensity and personnel safety by maintaining the task difficulty and environmental pressure unchanged and implementing differentiated assistance. After identifying joint high risks and generating decision instructions, the system's intervention logic does not simply reduce the training difficulty or suspend training. Instead, based on the R2 subtype and ST type, it adopts a strategy of maintaining the original task difficulty and environmental pressure unchanged and provides precise assistance by activating augmented reality prompts, biofeedback devices, and collaborative link optimization technology. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 The flowchart of the multi-dimensional data-driven military simulation training precision evaluation and decision-making system of the present invention is shown. Figure 2 A block diagram of the system of the present invention is shown. Detailed Implementation

[0019] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more exemplary embodiments. Numerous specific details are provided in the following description to give a full understanding of exemplary embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more of the specific details omitted, or other methods, components, steps, etc., can be employed. In other instances, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0021] Example 1: See Figure 1 The multi-dimensional data-driven military simulation training precision evaluation and decision-making system of this embodiment includes the following steps in its workflow: Data from motion sensors, physiological sensors, environmental sensors, and interactive recording devices deployed in the training environment, on trainees, and on simulation equipment are collected to construct a multi-source state parameter set. Based on the distribution characteristics of multi-source state parameter set action data along the training time sequence, the action execution mode of trainees is identified. Combining the corresponding change relationship between physiological data and action signals, it is determined whether there are individual fatigue or stress characteristics in the current training, and a preliminary individual risk label R1 is generated. Based on the changes in environmental sensor values ​​in key areas of the training field, combined with historical training data and mission scenario records, it is determined whether there are conditions that could induce tactical errors or coordination obstacles, and the individual risk label is updated to R2. If the individual risk identifier R2 is in a high-risk state, further analysis is performed on the rate of change of the parameter set interaction data at the team's preset communication nodes, and compared with the preset collaborative response threshold to identify the team's collaborative failure momentum and form a team collaborative early warning feature. By logically correlating individual risk identifier R2 with team collaborative early warning features, an individual-team collaborative risk judgment model is constructed. If the model output determines that the state is a joint high-risk state, a joint response decision instruction is generated. Based on the decision-making instructions, the difficulty or pace of the training scenario is adjusted, and the guidance system is coordinated to implement targeted interventions, while the auxiliary training support module is activated.

[0022] In this invention, it is necessary to first comprehensively perceive the operational status during military simulation training to provide data support for subsequent assessments of individual performance and team collaboration risks. This step specifically includes the following: Various types of sensors are deployed along the critical path and task nodes in the training environment, and monitoring devices are installed on trainees and simulation equipment, including but not limited to the following: Motion sensors: Deployed on key parts of the trainees' bodies and operating components of simulated equipment to collect real-time data on personnel movements, equipment operation sequences, and accuracy parameters, with a focus on monitoring abnormal movement patterns and operational delays. Physiological sensors: integrated into the wearable devices of trainees to monitor changes in their physiological state during training and to determine whether there is excessive fatigue, stress response or attention deficit. Environmental sensors: installed at key locations in the training ground to detect light intensity, temperature and humidity, terrain complexity, intensity and distribution characteristics of simulated battlefield effects, and to help assess the impact of environmental factors on training execution; Interactive recording device: used to capture interactive data such as voice communication, gesture signals, and data link information exchange between trainees, and generate a complete training interaction log.

[0023] The aforementioned devices continuously collect training process parameters at preset time frequencies. The system performs unified aggregation, spatiotemporal alignment, and standardization on all types of data to construct a multi-source state parameter set T, expressed as follows: ; T: Multi-source state parameter set, organized in time series form, is a summary of all training data collected by the system, including a four-dimensional standardized data combination of all collection times and all training units. A single element of parameter set T corresponds to a combination of four-dimensional data from the same training unit at the same time. The four data are completely synchronized in time and space and serve as the basic data unit for subsequent risk analysis.

[0024] Physiological or state parameter values ​​(such as personal heart rate or equipment operating status data) of a certain training unit (which can be an individual or simulated equipment) at the i-th data collection time.

[0025] : Corresponds to the action or operation data (such as aiming accuracy, equipment operation sequence) of the training unit at the i-th acquisition time.

[0026] : Environmental parameter values ​​(such as light intensity, electromagnetic interference, temperature and humidity) at the i-th acquisition time.

[0027] : Interaction record data at the i-th acquisition time (such as voice communication and data link information exchange records between trainees).

[0028] n: The total number of data collection points, which is positively correlated with the system's preset collection frequency and training duration.

[0029] In this invention, in order to achieve early identification of individual performance risks during training, the action execution patterns and physiological response characteristics of trainees are first identified based on action and physiological data from a multi-source state parameter set. Combined with environmental factors, it is determined whether there are individual ability bottlenecks or abnormal states, and then a preliminary individual risk label R1 is generated.

[0030] The technical process for this step includes the following: The real-time data sequences of motion sensors and physiological sensors of designated trainees during training time are extracted from the multi-source state parameter set, filtered and feature extracted, and the motion smoothness curve and physiological load curve are restored.

[0031] The processed motion curves are subjected to feature calculations, and key motion node indicators are analyzed, including but not limited to the completion time of aiming and firing, tactical movement, and equipment operation. Accuracy deviation δ and stability index σ; among which, completion time The effective time taken by trainees from the preset start point to the completion point of a certain type of action / operation represents the execution efficiency of the action / operation. Only the effective tactical operation time is counted, and the ineffective time such as non-tactical pauses and operation corrections is excluded. The accuracy deviation δ refers to the deviation between the actual execution result of a certain type of action / operation and the preset standard result / target result, representing the execution accuracy of the action / operation. In this embodiment, the absolute deviation is used as the core calculation benchmark. The smaller the deviation value, the higher the accuracy. The stability index σ is the standard deviation in statistics, which represents the completion time when trainees complete the same action / operation multiple times. The degree of dispersion of the accuracy deviation δ is a core quantitative indicator for evaluating the stability of action / operation execution. The smaller the σ value, the more stable the multiple executions. The more concentrated δ is, the higher the stability of the action / operation, and vice versa.

[0032] To determine whether trainees are in a state of fatigue or stress, the present invention further performs the following processing: Perform first-order difference operation on the physiological load curve to obtain the physiological change rate ΔP at each time point, and compare the change rate with the preset normal fluctuation threshold range.

[0033] Physiological change rate ΔP at each time point: refers to the instantaneous change rate of the physiological load parameter value of a participant at a certain data collection time point i relative to the previous adjacent data collection time point i-1 during military simulation training. It characterizes the magnitude and trend of physiological load change within a unit data collection time. The positive or negative value of ΔP reflects the direction of physiological load increase or decrease, and the absolute value reflects the severity of physiological state change. It is a key quantitative basis for determining whether participants have physiological overload or stress response. In this embodiment, ΔP is divided into the single physiological indicator change rate ΔPx (x is a specific physiological monitoring indicator) and the comprehensive physiological change rate. The former is used for state analysis of a single physiological dimension, and the latter is used for comprehensive determination of multi-dimensional physiological states. Both are instantaneous change rates at each time point.

[0034] The preset normal fluctuation threshold is derived statistically from the maximum ΔP fluctuation range under normal performance conditions in historical training data, and is usually set to a range of 2 standard deviations. If ΔP deviates continuously within a certain period of time, such as a persistently high heart rate or severe fluctuations in skin conductance, and the duration is not less than the preset critical duration, the threshold is set accordingly. If so, then the time period is marked as a suspected abnormal time period T1.

[0035] Further analysis of motion performance data within the T1 timeframe, including motion metrics such as completion time. The deterioration trend of accuracy deviation δ is calculated, and the amount of degradation of the action index relative to its own baseline level during this period is calculated.

[0036] Individual motor degeneration This refers to the decline in an individual's action / operation performance relative to their baseline level within a given time window. Let the average completion time of key actions within this window be [value missing]. Given that the mean accuracy deviation is δ, the stability index is σ, and the corresponding baseline window indices are Ato, 60, and σo, we can define: ;in, , , ≥0, and + + =1.

[0037] Team average degradation Within the same local environment and the same task phase, for team members The arithmetic or weighted average obtained after removing outliers.

[0038] Degradation of actions during time period T1 Physiological change rate ΔP is used for pattern recognition to classify the following three individual risk types, with different judgments and subsequent processing logics corresponding to different types: Type I risk (physiological-motor coordination disorder): when Furthermore, if ΔP remains abnormal, it is considered a high-risk situation, and intervention measures should be initiated as a priority. Type II risk (potential physiological overload): When ΔP remains abnormal but At this time, it is marked as a physiological compensatory state, judged as a potentially high-risk state, and high-frequency continuous monitoring is initiated. If subsequent motor degeneration occurs ( If it is, then it is upgraded to Type I risk; Type III Risk (Skill / Psychological Error): When However, when ΔP is normal, it is marked as a skill deviation or psychological fluctuation, and is judged as a targeted risk. Priority is given to analyzing the match between action patterns and task difficulty, which can be corrected through embedded guidance.

[0039] All three risk types generate a preliminary individual risk identifier R1, but must be accompanied by a corresponding risk type label (Type I, Type II, Type III) for subsequent modules to perform differentiated processing based on the label.

[0040] For the T1 time period, which is determined to be a high-risk state, retrieve the synchronous data of the corresponding environmental sensors in the training area (with the time window t set to the T1 time period) and analyze the potential impact of environmental factors.

[0041] Individualized environmental sensitivity analysis was employed, combining longitudinal comparisons of environmental change rates and individual performance change rates to avoid spurious correlations caused by global parameter analysis. The specific processing is as follows: For each trainee, extract the time series of environmental parameters for their local area. A differential correlation analysis within a sliding window was performed on the individual's environmental change rate by comparing it with the degradation amount ΔM(t) within the individual's own performance. With individual performance change rate Intra-individual correlation coefficient .

[0042] Set intra-individual correlation coefficient threshold At the same time, retrieve the individual's historical training data and calculate their environmental change tolerance threshold. This refers to the maximum rate of environmental change that allows the individual to maintain normal performance. If and The level remains consistently above the individual's historical tolerance threshold. If so, then environmental stress is determined to have a significant impact on the individual.

[0043] To further verify the dominant role of environmental factors, the average performance of teams under the same local environment was compared: If the team's average performance remains stable (the rate of decline in team average performance) The individual's performance declined significantly. This further confirms that environmental factors are the main trigger for the individual's risk. If the team's average performance declines simultaneously ( If so, priority should be given to the impact of the environment on the team as a whole, rather than individual-specific environmental sensitivity.

[0044] like or The individual's historical tolerance threshold was not exceeded. If the individual risk is determined to be mainly due to insufficient matching between the individual's own condition and the difficulty of the task, then it is determined that the individual risk is mainly due to insufficient matching between the individual's own condition and the difficulty of the task.

[0045] The identified periods of abnormal individual behavior, abnormal physiological responses, and individualized environmental sensitivity analysis results are comprehensively matched. Combined with the three risk types previously identified, if multiple abnormal characteristics highly overlap in time and space, such as when the overlap exceeds a preset threshold, [further action is taken]. If individualized environmental sensitivity analysis confirms that the environment has a significant impact on the individual, then the validity of the corresponding risk type and the environmental trigger attribute are confirmed; if the environment has no significant impact on the individual, then the risk is confirmed to originate from the individual's own factors. Finally, a preliminary individual risk label R1 with type label and trigger attribute (environmentally triggered / individually triggered) is generated, providing a classification basis for subsequent environmental factor impact analysis and risk label updates.

[0046] Preset threshold Set experience values ​​according to the complexity of the training scenario; for complex scenarios, the values ​​can be increased accordingly.

[0047] In this invention, based on the numerical changes of environmental sensors in key areas of the training field, combined with historical training data and the characteristics of the task scenario, it is determined whether there are conditions that could induce tactical errors or coordination obstacles, and the individual risk label is updated to R2.

[0048] This step includes the following technical aspects: Real-time environmental data is collected using environmental sensors deployed at key tactical positions and in complex environments within the training ground, and a time-series curve E(t) of environmental parameters is constructed. A sliding window analysis is then performed on the E(t) curve to calculate the growth rate of adverse environmental factors per unit time. Adverse environmental factors include, but are not limited to, decreased visibility, increased electromagnetic interference, and abrupt changes in terrain.

[0049] Access the historical training performance parameter table stored in the database, including but not limited to the following key metrics: Environmental adaptability threshold θE: refers to the upper limit of various environmental parameters that a specific training unit can maintain normal tactical effectiveness when performing a certain type of tactical mission under a given training level and equipment configuration. If the measured environmental parameter values ​​in the training field exceed this threshold, it indicates that the current environmental conditions have exceeded the unit's normal adaptability range, which is likely to cause operational errors by the trainees and team coordination obstacles. It is the core quantitative benchmark for initially judging the risk of tactical execution induced by the environment.

[0050] Task-environment matching coefficient β: This is a dimensionless quantitative index that characterizes the sensitivity of a specific type of tactical task to the influence of a certain type of environmental factor. The value ranges from 0 to 1. The closer the β value is to 1, the more sensitive the tactical task is to changes in the corresponding environmental factor. Even small fluctuations in the environmental factor can easily lead to a decline in tactical effectiveness and operational errors. The closer the β value is to 0, the less sensitive the tactical task is to changes in the corresponding environmental factor. The impact of environmental fluctuations on tactical execution can be ignored.

[0051] Based on the measured environmental parameter value E and the mission type, determine whether the current environment exceeds the adaptability threshold θE for the unit to perform the current mission. If E > θE, it is preliminarily determined that there is an environmentally induced tactical execution risk.

[0052] If both of the following conditions are met simultaneously: the growth rate of adverse environmental factors Continuously exceeding the preset rate threshold (Empirical threshold); If the current environmental parameter E exceeds the average environmental level Eavg of errors in similar tasks during the same period in history, then the area or time period is marked as a potentially high-risk area / time period ZE. Due to sudden environmental changes or excessively high preset difficulty, this area / time period is prone to cognitive overload, operational errors, or disjointed coordination among trainees.

[0053] The spatiotemporal range corresponding to the preliminary individual risk identifier R1 with attached type labels and causal attributes is matched with the current potential high-environment risk area / time period ZE. Combined with the results of individualized environmental sensitivity analysis and the performance of team members in the same environment, the risk attribution is further refined to achieve a differentiated upgrade from R1 to R2. The specific rules are as follows: 1. If the spatiotemporal overlap between the two exceeds a preset threshold, and the causative attribute of R1 is environmentally induced, then two subtypes of R2 are distinguished based on the prevalence of the environmental impact: R2a (Environment-Dominated Individual Risk): Changes in environmental parameters are strongly correlated with the individual's performance degradation. Furthermore, in the same local environment, the performance of most (≥50%) team members declined synchronously. At this point, individual performance abnormalities and potential team collaboration problems are identified, primarily attributed to environmental interference, and the subsequent environment-team collaboration impact analysis path is initiated. R2b (individual intrinsic risk): Changes in environmental parameters are associated with the individual's performance decline. However, other team members performed steadily in the same local environment. At this point, the assessment of individual risk primarily stems from the individual's own condition; subsequent collaborative analysis will focus on examining the individual's negative impact on team collaboration. 2. If the spatiotemporal overlap between the two exceeds the preset ratio threshold, but the causative attribute of R1 is the individual itself, or the universality of environmental influence cannot be clearly distinguished (such as insufficient team sample size), then R1 will not be upgraded for the time being, and it will be marked as requiring manual judgment, and its status and changes in team collaboration will be continuously monitored. 3. The upgraded R2a and R2b retain the original R1 risk type labels (Type I, Type II, Type III) and causative attributes, providing a classification basis for subsequent differential analysis and intervention measure formulation.

[0054] In this invention, if the individual risk identifier R2 is in a high-risk state, the rate of change of the parameter set interaction data at the team's preset communication node is further analyzed and compared with the preset collaborative response threshold to identify the team's collaborative failure momentum and form a team collaborative early warning feature.

[0055] This step mainly includes the following technical processes: extracting the interaction record data between key communication nodes within the team during the training process, sampling and analyzing it at fixed time intervals T, and calculating key indicators: instruction clarity Qc, response latency Tr, and information integrity Ic.

[0056] Command clarity Qc: Characterizes the degree to which team tactical commands are identifiable and understandable when transmitted between communication nodes. It is a dimensionless index with a value range of [0,1]. The closer the value is to 1, the higher the command clarity. The closer the value is to 0, the more ambiguous the command is and the less effectively it can be identified and understood. ;in, For speech recognition confidence; For semantic integrity (number of tactical command slot hits / total number of slots); , The weights are fixed; the slots are categorized into four types: target position, execution action, time constraint, and collaborating object.

[0057] Response latency Tr: Represents the time difference between the completion of command reception and the first valid response; it is a core time indicator for evaluating collaborative response efficiency, measured in seconds (s), and the result is rounded to one decimal place. A smaller Tr value indicates a more timely response, while a larger value indicates a more severe collaborative response latency. Valid response: triggered by any of the following: voice reply, action execution, or data link acknowledgment.

[0058] Information Completeness Ic: Characterizes the degree of completeness of core tactical information transmitted between communication nodes in a tactical command. It is a dimensionless index with a value range of [0,1]. The closer the value is to 1, the more complete the core information transmission; the closer it is to 0, the more severe the core information loss. Core tactical information consists of necessary information for completing tactical actions (such as target location, execution time, and action type), and no redundant information is included. It characterizes the degree of completeness of core tactical information transmitted between communication nodes in a tactical command. It is a dimensionless index with a value range of [0,1]. Let the total number of core information items contained in a certain command be... The number of core information entries successfully parsed by the receiving end is ,but When the core information includes four types of slots: location, time, action, and object, This represents the actual number of slots included in the instruction; if the number of missing items is greater than or equal to 1, it is reduced proportionally; for complete missing items, Ic=0. For the team to which high-risk individual R2 (R2a or R2b) belongs, and considering the risk type label (Type I, Type II, Type III) and subtypes associated with R2, the focus is on analyzing the interaction links related to this high-risk individual, and calculating the rate of change of instruction response latency Tr involving this high-risk individual within the current time period. and the rate of decline in information transmission integrity Ic The key points of the specific analysis are as follows: For R2a (environment-driven individual risk): the focus is on monitoring the impact of environmental disturbances on the overall team collaboration chain, rather than the drag on a single individual. Simultaneously, the changes in collaboration indicators of other members in the same environment are analyzed to determine whether the decline in collaboration is a group phenomenon caused by the environment. For R2b (individual intrinsic risk): the focus is on monitoring the drag on the collaborative link by this individual. For individuals with type I and type III risk, the focus is on monitoring the attenuation of the collaborative link. For individuals with type II risk, the focus is on the impact of changes in physiological compensatory state on collaborative efficiency during the collaborative process.

[0059] Set a continuous monitoring period window N. If within this window... The value remains positive (i.e., the response latency continues to increase) and A consistently negative value (i.e., a continuous increase in information omissions) indicates a continuous downward trend in the individual's and team's collaborative efficiency.

[0060] Retrieve baseline data on overall team collaboration effectiveness, obtain the expected values ​​Tr0 and Ic0 and their fluctuation range under normal collaboration conditions, and set collaboration effectiveness decay thresholds ηT and ηI.

[0061] If the interaction link indicators involving high-risk individuals are consistently below the collaborative effectiveness decay thresholds ηT and ηI, and the downward trend is stable, it indicates that the individual has become a weak or blocking node in the team's collaborative network.

[0062] Parameter threshold settings: For each trainee, the system selects a stable baseline window of no less than L0 seconds before / at the beginning of training, calculates the mean μM and standard deviation σM of the motion degradation sequence, and sets... Where k ranges from 1.5 to 3; critical duration The duration should be 10-120 seconds, adjusted according to the pace of the subject. Individual correlation coefficient values. The threshold values ​​can be 0.3 to 0.7; the spatiotemporal overlap threshold K1 can be 0.5 to 0.8; the synergistic decay thresholds ηT and ηI are determined based on the quantiles (e.g., 5th percentile) or mean + 2σ of Tr and Ic during normal synergistic periods. The risk coupling threshold CF0 is selected based on the ROC curve of historical joint risk events, ensuring that the recall rate is not lower than a preset value (e.g., 0.8).

[0063] Before association matching, a temporal priority check is introduced to clarify the causal direction between individual anomalies and team synergy attenuation, and then differential judgment is made in conjunction with R2 subtypes. The specific steps are as follows: First, extract the starting time of an individual's high-risk status. (The initial determination time of R2a / R2b) and the start time of team collaboration link decay (At the moment when the synergy index first falls below the decay threshold), compare the order of the two to initially determine the causal orientation; if the time difference between the two is less than a preset time threshold, then the Granger causality test is used to further confirm the dominant direction. The Granger causality test procedure is as follows: within a sliding window of length L, the individual performance sequence X(t) and the team synergy sequence Y(t) are resampled and aligned at a fixed sampling interval T; when a trend term exists in the sequence, differencing or detrending is used. A VAR(p) model is used, where p is selected by AIC / BIC (p ranges from 1 to 10), and the significance level α ranges from 0.01 to 0.1. If the lag term of Y predicts X statistically significantly, then the causal orientation of Y→X is determined to be valid, and vice versa; if both are significant, it is marked as a two-way reinforcement warning.

[0064] Based on the above time-series priority test results, the individual's high-risk state R2 (R2a or R2b) is correlated and matched with the attenuation characteristics of the collaborative links involving that individual. The timestamps of the interaction events and the participants are used as comparison criteria, combined with R2 subtypes and causal orientation, to make a differential judgment: If it is R2a (environment-driven individual risk): 1. If If the team attenuation index changes ahead of individual anomalies (time difference ≥ preset threshold), it is determined that environmental interference first causes team collaboration failure, which in turn induces individual performance abnormalities. A team collaboration early warning feature identifier ST is generated and marked as environment-team dominant. The intervention focuses on repairing team communication and cooperation, and offsets the collaboration attenuation caused by environmental interference by real-time enhancement of assistance and collaboration rhythm optimization, without changing the training environment interference intensity parameter.

[0065] 2. If If the team's indicators continue to deteriorate after an individual's abnormality occurs, it is determined that the environmental interference first induces the individual's abnormality, which in turn exacerbates the team's collaborative failure, generating ST and marking it as an environment-individual dominant type. For the environment-individual dominant type, the intervention focus is to provide precise technical assistance to high-risk individuals while keeping the environmental interference unchanged, and to simultaneously adjust the team's collaborative rhythm to adapt to the individual's recovery.

[0066] 3. If both occur almost simultaneously, and Granger causality tests confirm that the environment has a stronger causal influence on team collaboration, then the ST judgment should be based on the environment-team-dominant type; if the environment has a stronger causal influence on individuals, then the ST judgment should be based on the environment-individual-dominant type, or a two-way reinforcement warning should be triggered.

[0067] If it is R2b (individual intrinsic risk): 1. If If the team attenuation index changes ahead of individual anomalies (time difference ≥ preset threshold), it is determined that the individual anomaly is induced by team collaboration failure, ST is generated, and it is marked as a team-induced type. The intervention focus is to repair team communication and cooperation, while monitoring the recovery of individual status. 2. If If the team's metrics continue to deteriorate after an individual's abnormality occurs, it is determined that the individual's own abnormal state caused the team's collaboration problem, generating ST and marking it as an individual drag type. The intervention focus is to accurately support the individual and reduce its drag on the collaboration link. 3. If both occur almost simultaneously, and Granger causality tests confirm that team collaboration has a stronger causal impact on individuals, then the ST (Strain Induction) should be determined as team-induced; if it is confirmed that individuals have a stronger causal impact on team collaboration, then the ST should be determined as individual-dependent, or a bidirectional reinforcement warning should be triggered.

[0068] All of the above STs indicate that the team is at risk of potential collaborative failure. The core difference lies in the different causal orientations and intervention focuses, which provide a precise basis for subsequent joint risk assessment and decision-making instruction generation. At the same time, the results of time priority test and Granger causality test are recorded and incorporated into the training review data.

[0069] In this invention, the individual risk identifier R2 and the team collaborative early warning feature ST are logically correlated to construct an individual-team collaborative risk judgment model. If the model output result determines that the joint high-risk state, a joint response decision instruction is generated.

[0070] The core information and time-series priority test results of the current individual risk identifier R2 (R2a or R2b) and team collaborative early warning feature ST are extracted, including the risk subject (individual ID), risk level, R2 subtype, ST type, causal orientation, scope of influence, and spatiotemporal information. This information is uniformly mapped into a multi-dimensional state vector V, whose structure includes: subject information Vsub (individual ID, team affiliation); risk level Vlev (numerical representation, e.g., high, medium, and low corresponding to 3, 2, and 1 respectively); R2 subtype Vr2 (R2a / R2b); ST type Vst (environment-team dominant / environment-individual dominant / team-induced / individual dragging); causal orientation Vcd (team → individual / individual → team / bidirectional); scope of influence Vscope (affected team roles or functions); and time series Vtime. This state vector provides support for the quantitative analysis of risk propagation and coupling in subsequent models.

[0071] To comprehensively assess the inherent correlation and coupling degree between individual risk and team collaborative risk, this invention constructs a risk coupling determination rule Rc, which includes the following three dimensions: Spatiotemporal correlation Gt: Analyzes the degree of overlap between the risk periods and subjects (individuals and their related teams) corresponding to R2 and ST; Causal strength Gc: Combining time priority test results, R2 subtyping, and ST typology, the significant causal relationship and its strength between individual performance data sequences and team collaboration data sequences are quantified and analyzed using Granger causality tests or information theory methods (such as transitive entropy). For R2a+ environment-team-led ST: the focus is on analyzing the causal impact of environmental data sequences on team collaboration, and the transductive causal impact of team collaboration on individual performance; For R2a+ environment-individual-dominated ST: the focus is on analyzing the causal impact of environmental data sequences on individual performance, and the transductive causal impact of individual performance on team collaboration; For ST induced by R2b+ teams: the focus is on analyzing the causal guiding effect of team collaboration data sequences on individual performance data sequences; For R2b+ individual drag ST: focus on analyzing the causal guiding effect of individual performance data sequences on team collaboration data sequences; For two-way reinforcement early warning scenarios: simultaneously analyze the two-way causal strength of individuals to teams and teams to individuals, and take the maximum value of the two as the core quantitative basis of Gc.

[0072] Impact Diffusion Gd: Quantifying the scope of risk propagation based on the team's communication / interaction network. Assume a directed weighted graph G=(V,E,W) is constructed within a sliding window, where nodes are training participants or communication nodes, and edge weights... Let be the effective interaction strength between i and j per unit time (which can be obtained by weighting message frequency, acknowledgment receipt, and semantic command slot hit rate). Let the set of high-risk subjects be . Then the diffusivity can be defined as Where Reach(u,v) is the reachability strength from u to v, which can be taken as the maximum path product weight or the exponential decay form of the shortest path, for example... ; Gd represents the weighted shortest path distance, and λ is the attenuation coefficient. A larger Gd indicates that the risk is more likely to spread to more members / subtasks.

[0073] Based on the above three indicators, the individual-team risk coupling factor CF is calculated using a weighted formula, which is: CF=wt×Gt+wc×Gc+wd×Gd, where the weights can be set according to the training subjects and the characteristics of the troops.

[0074] Set the threshold for the risk coupling factor CF. A combined high-risk assessment is performed by combining the R2 subtype and the ST type: For R2a+ environments - team-led ST: if CF exceeds Furthermore, both R2 and ST risk levels are at high or very high levels, indicating that the current state is a combined high-risk environment-team-dominated type, representing a complex risk that the training process faces environmental interference that causes team collaboration failure and subsequently induces abnormal individual performance. For R2a+ environments - individual-dominated ST: if CF exceeds Furthermore, both R2 and ST risk levels are at high or very high levels, indicating that the current state is a combined high risk of environment-individual dominance, representing a complex risk that the training process faces environmental interference that induces individual abnormalities and further exacerbates team collaboration failure. For R2b+ teams, the induced ST is: if CF exceeds Furthermore, both R2 and ST are at high or very high risk levels, indicating that the current state is a team-individual dominated joint high risk, representing a complex risk of individual abnormalities induced by team collaboration failure during the training process; For R2b+ individuals with drag-type ST: if CF exceeds Furthermore, both R2 and ST are at high or very high risk levels, indicating that the current state is a high-risk individual-team-dominated joint risk, representing a complex risk of team collaboration failure caused by individual abnormalities during the training process. For scenarios requiring enhanced two-way early warning: if CF exceeds Furthermore, both R2 and ST are at high or very high risk levels, indicating that the current state is a two-way collaborative high-risk situation, representing a complex risk of mutual influence between individuals and teams and two-way risk transmission.

[0075] After identifying a joint high-risk state, a joint response decision instruction D is generated based on the risk subjects, types, and degree of coupling, specifically including: Dynamically provide real-time enhanced assistance: While maintaining the original task complexity, environmental pressure, and intensity of confrontation, provide precise assistance to trainees through technical means to help them complete tasks and improve their capabilities under pressure, and set up assistance windows; Initiate targeted interventions: Implement differentiated interventions based on R2 subtypes (R2a / R2b), risk type labels, and trigger attributes, and send categorized prompts to on-site instructors or the embedded intelligent coaching system. Regarding R2a (environment-driven individual risk): 1. Corresponding Environment - Team-led ST: Keep the local environmental interference unchanged, focus on repairing the team communication links and optimizing the team's coordination rhythm, provide tactical guidance to all affected members through AR prompts, and implement synchronous intervention for all affected members; among them, Type I risk individuals are given real-time voice guidance + physiological state adjustment prompts, Type II risk individuals are given physiological relaxation guidance, and Type III risk individuals are given action standardization guidance. 2. Corresponding Environment - Individual-Driven ST: Keep the environmental disturbances unchanged, implement precise technical assistance for the individual simultaneously, carry out intervention in combination with risk type labels, help the individual adapt to the current environmental pressure through AR tactical guidance, and adjust the team's collaborative rhythm to adapt to the individual's recovery. 3. Enhanced two-way early warning: While maintaining the same environmental interference, provide the team with prompts to repair the collaborative link and provide precise assistance to individuals. Implement two-way technical support between the team and individuals to help trainees block the two-way transmission of risks under pressure. Regarding R2b (individual intrinsic risk): 1. For team-induced ST: Focus on repairing team communication and coordination, optimizing team collaboration processes, maintaining the overall training difficulty of the team, and providing the individual with AR tactical guidance and status recovery prompts to help them keep up with the team's collaboration rhythm; 2. For individuals with ST-type drag: The focus is on providing precise technical assistance to the individual, combining risk type labels to carry out action guidance, physiological adjustment or psychological counseling, and improving their operational accuracy and coordination efficiency through AR prompts. At the same time, targeted on-the-spot adjustments are made to team coordination to reduce the individual's drag on the coordination link and keep the overall training difficulty of the team unchanged. 3. Corresponding two-way enhanced early warning: Simultaneously provide precise assistance to individuals and optimize collaborative processes for teams, establish a collaborative adaptation mechanism between individuals and teams, maintain the training difficulty unchanged, and prevent the two-way transmission of risks; Activate the support module: provide augmented reality prompts for high-risk individuals, or provide teams with collaborative situational awareness overlays; Recording and Early Warning: Record joint risk events and decision-making instructions in detail, and push them to the training review system and commander's terminal to provide data support for training control and subsequent debriefing.

[0076] In this invention, the difficulty or pace of the training scenario is adjusted according to decision-making instructions, and the guidance system is linked to implement targeted intervention. At the same time, the auxiliary training support module is activated. This process includes the following steps: Upon receiving the joint response decision command D, the control system immediately enters dynamic adjustment mode, initiates the training scenario engine control subroutine, and dynamically modifies the training process by adjusting the scenario engine parameters according to the preset adjustment strategy in the decision command. Common adjustment methods include: Pace control: Insert brief pauses or slowdowns in the mission to provide the team with internal adjustment time, such as controlling the intensity of virtual enemy attacks to enter a stable phase; Assisted adjustment: Combining R2 subtypes and trigger attributes, while maintaining the original task difficulty and environmental pressure, implement differentiated technical assistance adjustments: Regarding R2a (environment-driven individual risk): 1. Corresponding Environment - Team-led ST: Keep the local environmental interference unchanged, simultaneously optimize the team's task allocation, reduce internal friction in team collaboration, and provide clearer AR tactical prompts and collaborative guidance to all members affected by the environment; 2. Corresponding Environment - Individual-led ST: Keep the environmental interference unchanged, while providing the individual with precise AR tactical guidance and operational assistance, optimizing its task adaptability, and helping it improve action efficiency under the current environmental pressure without reducing the task difficulty; 3. Two-way enhanced early warning: While maintaining the same level of environmental interference and task difficulty, the system simultaneously provides guidance on the team's collaborative situation and operational assistance to individuals, comprehensively improving trainees' coping abilities and alleviating training pressure without reducing difficulty; Regarding R2b (individual intrinsic risk): 1. For team-induced ST: focus on optimizing team task processes and repairing collaboration links, maintaining the overall training difficulty of the team, and providing the individual with AR tactical guidance and adaptive operation assistance to help them keep up with the team's pace; 2. For individual-related STs: The focus is on providing the individual with precise AR motion guidance and collaborative assistance to optimize their task adaptability, rather than adjusting environmental parameters or reducing task difficulty, while maintaining the overall training difficulty of the team unchanged; 3. Corresponding two-way enhanced early warning: While providing precise operational assistance to the individual, the team's coordination rhythm is fine-tuned to achieve coordination and adaptation between the individual and the team, while maintaining the same training difficulty; Resource Assistance: In a simulated environment, this provides virtual reinforcement ammunition or restores some simulated health points to training units, alleviating current training pressure. Resource assistance is only used to maintain training continuity and safety; it does not change the adversarial intensity parameters or scoring rules. Resource compensation is included in the debriefing scoring using an equivalent conversion factor to ensure that the training difficulty assessment is not reduced.

[0077] The control system, according to preset intervention logic, coordinates the intervention of the control panel, on-site instructors, or intelligent auxiliary systems. Intervention methods include: Instructor voice intervention: Through a dedicated communication channel, instructors send concise instructions or reminders to specific trainees or teams; Intelligent prompt injection: Prompt information such as text, icons, or AR arrows is injected through the trainees' terminal devices; Training process fine-tuning: Temporarily adjust the task execution order, switch the current high-risk link to auxiliary training mode, without reducing the operation requirements and task difficulty, only add AR tactical guidance, and switch back to assessment mode after the trainees' condition recovers. In the training environment and the equipment of trainees, auxiliary support modules are preset. By combining R2 subtypes (R2a / R2b), risk type labels, and identified high-risk types, the corresponding auxiliary support modules are activated to achieve differentiated assistance. Regarding R2a (environment-driven individual risk): 1. Corresponding Environment - Team-led ST: Simultaneously activates auxiliary modules for all affected members in the same environment, focusing on providing the team with collaborative situational awareness overlay information, communication link repair prompts, and overlaying AR standard action guidance and environment-adaptive operation prompts for individuals; 2. Corresponding Environment - Individual-led ST: Simultaneously activates the auxiliary module for all affected members in the same environment, focusing on providing the individual with personalized AR action guidance and physiological adjustment prompts, and providing collaborative adaptation prompts for the team; 3. Corresponding two-way enhanced early warning: Simultaneously activate team and individual auxiliary modules, taking into account both team collaborative repair and individual status support, and preventing the two-way transmission of risks.

[0078] Regarding R2b (individual intrinsic risk): 1. For team-induced ST: The focus is on visualizing the team's collaborative situation and providing communication link optimization prompts, while also activating AR action guidance and status recovery prompts for the individual to help them keep up with the team's collaborative rhythm; 2. ST for individuals with a drag on the team: The dedicated auxiliary module is activated only for the high-risk individual. The AR system is used to overlay standard action diagrams or trajectory guidance to address the individual's action risk. The biofeedback device is used to guide and adjust the breathing state to address the stress physiological response. At the same time, the team is provided with collaborative obstacle avoidance prompts to reduce the individual's drag on the team. 3. Corresponding two-way enhanced early warning: While activating the exclusive auxiliary module for the individual, it provides collaborative adaptation prompts for the team, realizing two-way support between the individual and the team.

[0079] To address individual movement risks, a standard movement diagram or trajectory guide is overlaid in the trainee's field of vision using an AR system; To address team collaboration risks, the location of key friendly units or their expected routes of action that may have been overlooked is highlighted on the situation display screen of the commander or relevant members. In response to stress physiological responses, a biofeedback device guides trainees to adjust their breathing patterns.

[0080] During the dynamic adjustment and support process, real-time performance data, physiological data and interaction data of individuals and teams are continuously collected to construct time series P'(t), A'(t), and C'(t) curves to assess the risk mitigation effect and the recovery trend.

[0081] Set recovery thresholds, such as individual physiological indicators P' returning to the normal range, action efficiency A' recovering to more than 80% of the baseline level, and team coordination indicators C' recovering to within the decay threshold η. If all key indicators meet the recovery conditions within multiple consecutive sampling periods, it indicates that the auxiliary measures are effective and the condition of the trainees and the team tends to stabilize; if the condition continues to deteriorate, the event is recorded and the system switches to safety mode to ensure personnel safety without reducing the difficulty of the task itself.

[0082] If the above recovery criteria are met, the system will automatically and gradually exit the auxiliary mode, smoothly transition to the normal training rhythm, and restart the regular monitoring and evaluation logic. Simultaneously, a complete report will be generated for this risk event (clearly labeled with R2 subtype, ST type, causal orientation, and risk attribution), time-series priority test results, Granger causality test results, and auxiliary measures. This report will be used for in-depth post-training review and updating of trainees' competency profiles. For R1 states marked as requiring manual assessment, relevant data and analysis results will be pushed to the instructor's terminal for manual assessment and handling.

[0083] Example 2: See Figure 2As shown, the multi-dimensional data-driven military simulation training precision evaluation and decision-making system of this embodiment includes: The data acquisition module is used to deploy motion sensors, physiological sensors, environmental sensors, and interactive recording devices to collect training data and construct a multi-source state parameter set; The individual risk identification module identifies individual risk types based on the temporal distribution characteristics of action data and physiological data, and generates a preliminary individual risk identifier R1 with attached type labels and trigger attributes; The environmental factors analysis module updates individual risk labels to the R2 subtype based on changes in environmental sensor values ​​and historical training data. The team collaboration early warning module analyzes the comparison results between the rate of change of interactive data and the collaborative response threshold to form the team collaboration early warning feature ST; The joint risk assessment module constructs an individual-team collaborative risk assessment model through logical correlation analysis and generates a joint response decision instruction D. The decision execution module adjusts the difficulty or pace of the training scenario based on decision instructions, and implements targeted interventions. The auxiliary support module provides differentiated auxiliary support. The dynamic monitoring and evaluation module continuously monitors the effectiveness of risk mitigation and the trend of status recovery.

[0084] The beneficial effects of this embodiment are: achieving precise control over the entire military training process through multi-module collaboration, data fusion driving risk quantification assessment, intelligent identification of individual-team composite risks, dynamic adjustment of training pace and implementation of differentiated intervention, and improvement of training efficiency and safety by combining AR assistance and biofeedback technology.

[0085] All formulas in this invention are dimensionless and calculated numerically. The preset parameters in the formulas can be set by those skilled in the art according to the actual situation.

[0086] The weighting coefficients of this invention are used to measure the degree of influence of different factors or variables on a certain outcome or decision. The weighting coefficient is defined as the numerical value assigned to each factor when comparing and evaluating multiple factors, reflecting their importance or priority. These weighting coefficients can be determined according to specific circumstances and needs, and are usually jointly formulated and confirmed by professionals or relevant stakeholders. By reasonably setting the weighting coefficients, programs or systems can be helped to make decisions or predictions more accurately.

[0087] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0088] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described below. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A multi-dimensional data-driven precision evaluation and decision-making system for military simulation training, characterized in that: The system workflow includes the following steps: S1. Collect multi-source sensing data from the training environment, trainees, and simulation equipment, and construct a multi-source state parameter set after spatiotemporal alignment and standardization processing; the multi-source sensing data includes motion sensor data, physiological sensor data, environmental sensor data, and interaction record data; the multi-source state parameter set is organized in time series form, containing corresponding combinations of state parameters, motion parameters, environmental parameters, and interaction parameters of the training unit at the same time. S2. Based on the action data and temporal distribution characteristics of the multi-source state parameter set, identify the action execution mode of the trainees, combine the correlation changes of physiological data and action signals, determine the individual fatigue or stress characteristics, and generate a preliminary individual risk label R1. S3. Based on the changes in environmental parameters of key areas of the training ground, combined with historical training data and mission scenarios, determine whether there are conditions that could induce tactical errors or coordination obstacles, and update the individual risk label to R2. S4. If the individual risk identifier R2 is in a high-risk state, analyze the change characteristics of the interactive data at the team communication node, compare it with the preset collaborative response threshold, identify the team collaborative failure trend, and form a team collaborative early warning feature. S5. Logically correlate individual risk identifier R2 with team collaborative early warning features to construct an individual-team collaborative risk judgment model. If the risk is determined to be a joint high-risk state, a joint response decision instruction will be generated. S6. Based on the joint response decision instructions, dynamically adjust the difficulty or pace of the training scenario, coordinate with the guidance system to implement targeted interventions, and activate the corresponding auxiliary training support modules.

2. The multi-dimensional data-driven military simulation training precision evaluation and decision-making system according to claim 1, characterized in that, Step S2 specifically includes: The motion and physiological data are filtered and feature extracted to reconstruct motion fluency and physiological load curves. The completion time, accuracy deviation, and stability index of key motion nodes are calculated. Simultaneously, the physiological load curve is differentially processed to obtain the physiological change rate. The physiological change rate is compared with a preset normal fluctuation threshold to mark continuous abnormal periods. Based on the amount of motion degradation within these periods, three risk types are classified: physiological-motion coordination abnormality, potential physiological overload, and skill or psychological error. Corresponding risk labels are assigned to form a preliminary individual risk identifier R1. Based on the correlation analysis between local environmental parameters and the amount of degradation in individual performance, combined with the comparison with the average performance of the team in the same environment, the individual risk causation attribute is determined to be either environmentally induced or individual-related.

3. The multi-dimensional data-driven military simulation training precision evaluation and decision-making system according to claim 2, characterized in that, The process of updating the individual risk label in step S3 includes: A sliding window analysis was performed on the time series curves of environmental parameters to calculate the growth rate of adverse environmental factors. Combined with the unit's environmental adaptability threshold and the mission-environment matching coefficient, potential high-environment risk areas or time periods were marked. The spatiotemporal range corresponding to the initial individual risk identifier R1 was matched with the potential high-environment risk areas / time periods. Combining individual environmental sensitivity with differences in the team's overall performance, R2 was divided into environment-dominated individual risk R2a and individual endogenous risk R2b. Both R2a and R2b retained the risk type label and causal attributes of the original R1 for subsequent differentiated collaborative analysis and intervention.

4. The multi-dimensional data-driven military simulation training precision evaluation and decision-making system according to claim 3, characterized in that, The team collaboration early warning features formed in step S4 include: Extract interaction data from key communication nodes within the team, and sample and calculate indicators of instruction clarity, response latency, and information completeness. For interaction links involving high-risk individuals, calculate the rate of change in response latency and the rate of decline in information completeness to determine the trend of declining collaborative efficiency. Through time-series priority tests and Granger causality tests, determine the order and causal orientation of individual anomalies and team collaborative decline. Combining R2a or R2b subtypes, classify team collaborative early warning characteristics into environment-team-dominated, environment-individual-dominated, team-induced, individual-drag, and bidirectional reinforcement early warning types to clarify the causes of collaborative failure and key intervention areas.

5. The multi-dimensional data-driven military simulation training precision evaluation and decision-making system according to claim 4, characterized in that, Step S5 involves constructing an individual-team collaborative risk assessment model, including: Individual risk identifiers (R2), team collaborative early warning features, and causal orientation are mapped into multi-dimensional state vectors. The degree of risk coupling is quantified from three dimensions: spatiotemporal correlation, causal strength, and impact diffusion, and the risk coupling factor is calculated. The risk coupling factor is compared with a preset threshold, and combined with R2 subtypes and collaborative early warning feature types, the joint high-risk type is determined, including environment-team-dominated, environment-individual-dominated, team-individual-dominated, individual-team-dominated, and bidirectional collaborative joint high-risk. Based on the joint high-risk type, risk subject, and degree of coupling, a joint response decision instruction is generated, which includes real-time enhanced assistance, targeted intervention, activation of auxiliary support modules, and early warning records.

6. The multi-dimensional data-driven military simulation training precision evaluation and decision-making system according to claim 5, characterized in that, The intervention logic for environment-dominated individual risk R2a is as follows: When dealing with a team-led collaborative early warning system, maintain the environmental interference unchanged, repair team communication links, optimize the collaborative rhythm, and simultaneously implement AR tactical guidance and differentiated status adjustment for affected groups. When dealing with an individual-led collaborative early warning system, maintain the environmental interference unchanged, provide precise technical assistance to high-risk individuals, and simultaneously adjust the team collaborative rhythm to adapt to individual status recovery. When dealing with a two-way enhanced early warning system, simultaneously provide team collaborative link repair and precise individual assistance to block the two-way transmission of risk without reducing the difficulty of the task.

7. The multi-dimensional data-driven military simulation training precision evaluation and decision-making system according to claim 5, characterized in that, The intervention logic for individual intrinsic risk R2b is as follows: When a team-induced collaborative warning is issued, the focus is on repairing the team's collaborative process, maintaining the overall training difficulty, and providing individuals with AR guidance and state recovery prompts. When an individual-dependent collaborative warning is issued, action calibration, physiological adjustment, and psychological counseling are implemented for high-risk individuals to optimize team collaboration in order to reduce individual drag without reducing environmental pressure and task difficulty. When responding to enhanced two-way early warning, we simultaneously implement precise individual assistance and team collaborative adaptation to establish a two-way risk prevention mechanism.

8. The multi-dimensional data-driven precision evaluation and decision-making system for military simulation training according to claim 1, characterized in that, The execution process of the auxiliary training support module in step S6 includes: The AR display system overlays standard movement trajectories, tactical guidance, and collaborative situational information onto trainees; the biofeedback device guides individuals under stress to regulate their breathing and calm their condition; during the intervention process, real-time data of individuals and teams is continuously collected to assess the recovery trend, and the auxiliary mode is gradually withdrawn after the recovery threshold is met; a debriefing report containing risk type, causal test results, intervention measures, and recovery effects is generated and simultaneously pushed to the commander's terminal and the trainees' competency profiles are updated.

9. The multi-dimensional data-driven military simulation training precision evaluation and decision-making system according to claim 1, characterized in that, The system includes: The system comprises the following modules: a data acquisition module for deploying motion sensors, physiological sensors, environmental sensors, and interactive recording devices to collect training data and construct a multi-source state parameter set; an individual risk identification module for identifying individual risk types based on the temporal distribution characteristics of motion and physiological data, generating a preliminary individual risk identifier R1 with type labels and causal attributes; an environmental factor analysis module for updating the individual risk identifier to the R2 subtype based on changes in environmental sensor values ​​and historical training data; a team collaboration early warning module for analyzing the comparison results between the rate of change of interactive data and the collaborative response threshold to form a team collaboration early warning feature ST; a joint risk judgment module for constructing an individual-team collaborative risk judgment model through logical association analysis and generating a joint response decision instruction D; a decision execution module for adjusting the difficulty or pace of the training scenario based on the decision instruction and implementing targeted interventions; an auxiliary support module for providing differentiated auxiliary support; and a dynamic monitoring and evaluation module for continuously monitoring the risk mitigation effect and state recovery trend.

Images