Fire-fighting training multi-dimensional index comprehensive score management method and system

Abstract

The invention discloses a fire-fighting training multi-dimensional index comprehensive score management method and system, particularly relates to the technical field of fire-fighting training management, and is used for solving the problem of score attribution distortion caused by neglecting a causal relationship between cooperative actions in an existing score method. The method comprises the following steps: constructing a knowledge graph representing causal conduction association between actions by acquiring individual action behavior data and team task results of firefighters in real fire smoke thermal simulation training, and identifying a key causal chain in the knowledge graph based on a mutual exclusion constraint of resource occupation and a consistency constraint of a tactical target; monte Carlo disturbance simulation is carried out on the key causal chain execution process to generate a process sensitivity index, the contribution weight of each action on the key causal chain to a team result is evaluated based on the index, and finally, the multidimensional comprehensive score of the individual firefighter and the team is comprehensively calculated according to the contribution weight. The method achieves the more accurate evaluation of the cooperative training effect, and provides data support for the optimization of a training scheme.

Description

Technical Field

[0001] This invention relates to the field of fire training management technology, and more specifically, to a comprehensive scoring management method and system for multi-dimensional indicators of fire training. Background Technology

[0002] CFBT (Flame, Smoke, and Heat) simulation training is one of the core forms of practical training in the fire protection field. Multi-person collaborative response is a crucial scenario in this type of training, and its core objective is to improve the emergency rescue coordination capabilities of fire teams by simulating division of labor and cooperation in real disaster environments. In the fire protection training management system, the scoring management of CFBT simulation training is a key link supporting training effectiveness evaluation, capability gap identification, and training program optimization, falling under the important application scope of personnel training and team workflow management. Existing scoring management methods mostly focus on independently quantifying explicit indicators such as individual operational standards and single task completion, then using simple summation or averaging to form individual and team scores, thus meeting the basic need for data-driven evaluation of training effectiveness in fire protection training management.

[0003] Existing fire training scoring and management methods, when applied to multi-person collaborative training scenarios in CFBT (Fire-Free Training), fail to fully consider the causal transmission relationships between individual firefighter actions during the collaborative process. They rely solely on individual independent indicator scores and the overall task results of the team for scoring and calculation. This results in scoring results that cannot accurately define the actual impact weight of different firefighter actions on training effectiveness, leading to problems of distorted scoring attribution. Consequently, they are unable to objectively and comprehensively reflect the true level of team collaborative training, nor can they provide accurate and effective data support for subsequent targeted optimization of collaborative training programs and improvement of the team's overall emergency rescue capabilities. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a multi-dimensional index comprehensive scoring management method and system for fire training to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A multi-dimensional indicator-based comprehensive scoring management method for fire training includes: S1. Obtain individual action and behavior data of firefighters and team task results during real fire smoke and heat simulation training; S2. Based on the individual firefighter action data, determine whether there is a causal transmission relationship between individual firefighter actions. If so, construct a knowledge graph with individual firefighter actions as nodes and causal transmission relationships between individual firefighter actions as edges. S3. Based on the mutual exclusion constraint of resource occupation and the consistency constraint of tactical objectives in the knowledge graph, identify the key causal chain from individual firefighter actions to team task results; S4. Perform Monte Carlo perturbation simulation on the execution process of the key causal chain to quantify the sensitivity of the key causal chain to execution perturbations and generate process sensitivity indicators. S5. Evaluate the contribution weight of individual firefighter actions on key causal chains to team task results based on process sensitivity indicators. S6. Based on the contribution weight, calculate the individual multidimensional comprehensive score of the firefighter and the multidimensional comprehensive score of the team.

[0006] Furthermore, S1 includes: Data on individual firefighter actions and behaviors are collected using sensors and audio / video recording devices in training facilities. Record team task results through the training management terminal; The individual actions and behaviors of firefighters are time-stamped and linked with the team's task results for storage.

[0007] Furthermore, S2 includes: Analyze the dependencies between actions in individual firefighter behavior data; By considering the chronological order, determine whether there is a causal relationship where the subsequent action depends on the conditions created by the preceding action. If a causal relationship exists, a knowledge graph is constructed using individual firefighter actions as nodes. In the knowledge graph, the corresponding individual firefighter action nodes are connected by edges based on the causal transmission relationship that is determined to exist.

[0008] Furthermore, S3 includes: Traverse the entire causal transmission path in the knowledge graph from individual firefighter actions to team task results; We selected causal transmission paths that satisfy the condition that there is no conflict in resource occupation between individual firefighter actions in the temporal and spatial dimensions; From the causal transmission paths that satisfy the mutual exclusion constraint of resource occupation, we select the causal transmission paths in which the tactical intentions of individual firefighter actions all point to the same team task outcome; The causal transmission paths that satisfy the tactical objective consistency constraint are identified as key causal chains.

[0009] Furthermore, the causal transmission path in which the tactical intentions of individual firefighter actions all point to the same team task outcome is screened, including: parsing the top-level tactical objective based on the team task outcome set in the training; mapping each individual firefighter action in the causal transmission path to a preset standard operating procedure library to obtain its corresponding standard tactical intention; determining whether the standard tactical intentions corresponding to all individual firefighter actions jointly support the top-level tactical objective; and retaining the causal transmission path in which the standard tactical intentions jointly support the top-level tactical objective.

[0010] Furthermore, S4 includes: Define time perturbation parameters and success rate perturbation parameters for individual firefighter actions included in the key causal chain; Based on the defined time perturbation parameters and success rate perturbation parameters, the execution process of the key causal chain is simulated by multiple random sampling. The percentage of team task results that were not achieved in multiple random sampling simulations was statistically analyzed. A process sensitivity index is generated based on the proportion of team task results that were not achieved.

[0011] Furthermore, based on the defined time perturbation parameters and success rate perturbation parameters, the execution process of the critical causal chain is simulated multiple times by random sampling. This includes: for each individual firefighter's action in the critical causal chain, a random time delay following a specific distribution is superimposed on its baseline execution time to simulate time perturbation; for each individual firefighter's action, a random number is generated based on its baseline success rate to determine whether the action was successful in this simulation to simulate success rate perturbation; based on the randomly generated time delay and the success status of the action, the entire execution process of the critical causal chain and the final team task result are sequentially deduced.

[0012] Furthermore, S5 includes: Analyze process sensitivity metrics to determine the degree of sensitivity to disturbances in key causal chains; The weighting ratio of individual firefighter actions on the critical causal chain is set according to the perturbation sensitivity value of the critical causal chain; The contribution weight of each firefighter's individual actions to the team's task outcome is calculated according to the weight allocation ratio.

[0013] Furthermore, S6 includes: For each firefighter, the contribution weight of each firefighter's individual actions and behaviors on the key causal chain is summarized to obtain the sum of individual contribution weights; Calculate the firefighter's individual multidimensional comprehensive score based on the sum of individual contribution weights; The team's multidimensional comprehensive score is obtained by combining the individual multidimensional comprehensive scores of all firefighters.

[0014] On the other hand, the present invention provides a multi-dimensional comprehensive scoring and management system for fire training, comprising: The data acquisition module is used to acquire individual action and behavior data of firefighters and team task results during real fire and smoke heat simulation training; The association judgment module is used to determine whether there is a causal transmission relationship between individual firefighter actions based on individual firefighter action data. If so, a knowledge graph is constructed with individual firefighter actions as nodes and causal transmission relationships between individual firefighter actions as edges. The causal chain identification module is used to identify key causal chains from individual firefighter actions to team mission outcomes in a knowledge graph based on mutual exclusion constraints of resource usage and consistency constraints of tactical objectives. The disturbance simulation module is used to perform Monte Carlo disturbance simulation on the execution process of key causal chains to quantify the sensitivity of key causal chains to execution disturbances and generate process sensitivity indicators. The weighting evaluation module is used to evaluate the contribution weight of individual firefighter actions on key causal chains to the team's task results based on process sensitivity indicators. The scoring calculation module is used to comprehensively calculate the individual multidimensional comprehensive score of firefighters and the multidimensional comprehensive score of teams based on their contribution weights.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention achieves a more accurate and objective evaluation of team collaborative training effectiveness by deeply mining and quantitatively analyzing the causal transmission relationships between individual firefighter actions. It formally represents the causal relationships between actions by constructing a knowledge graph, and introduces resource exclusivity constraints and tactical goal consistency constraints to filter out truly critical action logic chains. This deepens the evaluation perspective from discrete individual operational indicators to a coherent collaborative process. Based on this, it quantifies the stability of the execution process by performing Monte Carlo perturbation simulation on key causal chains, and quantifies this stability as a process sensitivity index. This provides a scientific basis for subsequent contribution weight evaluation reflecting the robustness of the collaborative link, enabling the scoring mechanism to penetrate surface behavior and capture the actual impact path and intensity of different actions on the final task outcome within the collaborative network.

[0016] 2. By dynamically allocating the contribution weight of individual actions based on process sensitivity indicators, and comprehensively calculating multi-dimensional scores for individuals and teams, this method effectively solves the problem of attribution distortion caused by neglecting causal dependencies between actions in traditional methods. The final score results not only more realistically reflect the actual value contribution of each firefighter in collaborative tasks and accurately identify their strengths and weaknesses, but also provide a stable and reliable data profile for the overall collaborative effectiveness of the team. The resulting evaluation data can be directly used to guide the targeted optimization of subsequent collaborative training programs, such as strengthening training in weak causal links or adjusting resource allocation strategies. Thus, within the specific application scope of personnel training and team workflow management, it provides efficient and accurate data decision support for continuously improving the overall emergency rescue capabilities of fire teams. Attached Figure Description

[0017] Figure 1 This is a flowchart of a multi-dimensional index comprehensive scoring management method for fire training according to the present invention; Figure 2 This is a schematic diagram of the structure of a fire training multi-dimensional index comprehensive scoring management system according to the present invention. Detailed Implementation

[0018] 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.

[0019] Example 1: Figure 1 This invention presents a multi-dimensional index-based comprehensive scoring management method for fire training, comprising: S1. Obtain individual action and behavior data of firefighters and team task results during real fire smoke and heat simulation training; S2. Based on the individual firefighter action data, determine whether there is a causal transmission relationship between individual firefighter actions. If so, construct a knowledge graph with individual firefighter actions as nodes and causal transmission relationships between individual firefighter actions as edges. S3. Based on the mutual exclusion constraint of resource occupation and the consistency constraint of tactical objectives in the knowledge graph, identify the key causal chain from individual firefighter actions to team task results; S4. Perform Monte Carlo perturbation simulation on the execution process of the key causal chain to quantify the sensitivity of the key causal chain to execution perturbations and generate process sensitivity indicators. S5. Evaluate the contribution weight of individual firefighter actions on key causal chains to team task results based on process sensitivity indicators. S6. Based on the contribution weight, calculate the individual multidimensional comprehensive score of the firefighter and the multidimensional comprehensive score of the team.

[0020] To obtain individual firefighter action and behavior data and team task results during simulated fire, smoke, and heat training, the following methods are employed: Data on individual firefighter movements and behaviors is collected through sensors and audio / video recording devices in the training facility. These sensors include various sensors deployed on firefighters' personal protective equipment (PPE) and fixed sensors deployed within the simulated training environment. Position sensors collect the real-time coordinates and movement trajectories of firefighters in three-dimensional space; these sensors are combinations of ultra-wideband positioning base stations and tags or inertial measurement units (IMUs). Motion recognition sensors collect physical signals indicating the firefighters' specific tactical actions. For example, pressure sensors mounted on breaching tools detect the force and duration of breaching actions; hose pressure sensors detect water flow parameters during water jetting; and IMUs worn by firefighters detect changes in running, crawling, or climbing postures. Environmental sensors collect environmental data within the training facility, such as temperature measurements from thermocouples, smoke concentration readings from smoke generator controllers, and oxygen or toxic gas concentrations from gas sensors. The audio and video recording device includes a high-definition camera and microphone array deployed in key areas of the training facility. The high-definition camera and microphone array continuously record video and audio streams of the training process, and analyze the video stream through computer vision motion recognition algorithms to identify the visual actions of firefighters, and analyze the audio stream through speech recognition technology to identify the voice communication content and instructions between firefighters.

[0021] Team task results are recorded through a training management terminal. The training management terminal is a computer system running training management software, operated by training observers or commanders. Team task results refer to the achievement of the overall objectives set for this simulated training. The results include task completion status, total task time, and the sequence and time points of key sub-task completion. For example, in a training exercise focused on indoor search and rescue and firefighting, the specific results include whether the simulated casualty was successfully located and rescued, whether the set fire source was successfully controlled or extinguished, and the total time elapsed from the start of the training to the successful rescue of the simulated casualty and control of the set fire source. Based on predefined scoring rules and success criteria, observers input various data points of the team task results through the human-computer interface of the training management terminal at the end of the training or during the training, and record the system timestamp at the moment the team task results are confirmed.

[0022] The system aligns and stores the individual firefighter action data with the team task results using timestamps. Timestamp alignment maps individual firefighter action data from training facility sensors, audio / video recording devices, and training management terminals to a unified global timeline with the team task results. Specifically, before training begins, a unified time server sends a network time protocol synchronization signal to all data acquisition devices and recording terminals to ensure initial synchronization of their local clocks. For devices unable to synchronize via the network, time alignment is performed after data acquisition based on a global synchronization event triggered at the start of training. This global synchronization event could be a specific whistle sound or a light signal. This event is captured by all audio / video recording devices, and the time point of occurrence of the global synchronization event in each data stream is used as the reference time point to calculate and compensate for the time offset of each device's data. Association storage stores the timestamp-aligned individual firefighter action data and team task results in a relational database or time-series database according to a pre-defined data structure. Each individual firefighter action record in the database table contains the following fields: a unique identifier for the data record, the corresponding firefighter's identification identifier, the action type code, the action start timestamp, the action end timestamp, the raw sensor data or the feature value obtained after feature extraction, and the identifier of the team task result associated with the action. The timestamp field is expressed in milliseconds or microseconds from the standard time epoch. The team task result record contains the following fields: task result identifier, task type, completion status, total time elapsed, and the timestamp of the recorded team task result. By linking the task result identifier with the associated fields in the individual firefighter action record, a traceable data link is established between the individual firefighter action and the team task result.

[0023] Based on individual firefighter action data, determine whether there is a causal relationship between individual firefighter actions. If so, construct a knowledge graph with individual firefighter actions as nodes and causal relationships between individual firefighter actions as edges. This is achieved in the following way: This analysis examines the dependencies between actions in individual firefighter action data. The process reads stored individual firefighter action data records. Each record contains an action type code, an action start timestamp, an action end timestamp, and an identifier for the team task outcome associated with the action. Dependency analysis is based on the temporal sequence of the individual firefighter action data records and the semantic logic of the action type codes. Records belonging to the same team task outcome are grouped. Within each group, all individual firefighter actions are sorted according to their start timestamps, forming a chronological action sequence. A predefined action type dependency rule base is used to determine if logical dependencies exist between actions. This rule base is a pre-created data structure that stores the action prerequisite relationships defined in the firefighting tactical standard operating procedures. Each rule in the rule base consists of two parts: a prerequisite action type code and a subsequent action type code. For example, a rule in the rule base might state that action type code A is a prerequisite for action type code B. During analysis, the action type code of each firefighter's individual action in the action sequence is matched with the action type dependency rule base. If there is a rule in the action type dependency rule base that specifies a prerequisite action type code that matches the action type code of a firefighter's individual action in the action sequence, and that specifies a subsequent action type code that matches the action type code of another firefighter's individual action in the action sequence that is later in time, then it is preliminarily determined that the firefighter's individual action corresponding to the subsequent action type code has a logical dependency on the firefighter's individual action corresponding to the prerequisite action type code.

[0024] By considering the temporal sequence, it is determined whether a causal relationship exists where the subsequent action depends on the conditions created by the preceding action. This determination process, based on identifying logical dependencies, further verifies whether the subsequent action actually utilizes the physical or informational conditions created by the preceding action. Specific verification methods rely on the analysis of additional sensor data and audio / video recordings in the individual firefighter action data. For physical condition dependencies, verification is performed by examining changes in spatial location data and environmental sensor data. For example, suppose a firefighter's individual action is breaking down a door or window, and their subsequent individual action is entering a room. To determine whether entering the room depends on the conditions created by breaking down the door or window, it is necessary to verify that, after breaking down the door or window and before entering the room, data from location sensors shows that a firefighter moved through the broken door or window, or that video streams from audio / video recordings show a firefighter passing through the broken door or window structure. For informational condition dependencies, verification is performed by examining communication records or command transmissions. For example, suppose a firefighter's individual action is fire scene reconnaissance, and their subsequent individual action is targeted water spraying. To determine whether targeted water spraying depends on conditions created by fire reconnaissance actions, it is necessary to verify that, during or after the fire reconnaissance, audio streams from audio-visual recordings contain communication content reporting the location of the fire source, identified by speech recognition technology, and that this communication content occurs before the targeted water spraying action begins. If this verification passes, a causal link is determined between the two individual firefighter actions. To reduce false positives, a time window threshold is set for link verification. The time window threshold is a predefined duration parameter, for example, set to 30 seconds. The subsequent action must begin within the time window threshold after the preceding action has ended; otherwise, even if a logical dependency exists, it is not considered a direct causal link.

[0025] If a causal relationship exists, a knowledge graph is constructed using individual firefighter actions as nodes. During knowledge graph construction, each firefighter action deemed to have a causal relationship is transformed into a node in the knowledge graph. Nodes have attributes, with attribute values ​​derived from the corresponding firefighter action data records. Attributes include at least a node identifier, a corresponding firefighter identity identifier, an action type code, an action start timestamp, and an action end timestamp. The node identifier is a globally unique string or number; for example, it may use a unique identifier recorded in the database. The knowledge graph architecture is defined using an attribute graph model and stored and managed using a graph database.

[0026] In the knowledge graph, edges connect the corresponding individual firefighter action nodes, with each edge representing a causal link. Edges are created based on the causal links identified in the previous step. For each pair of firefighter actions with a causal link, a directed edge is created in the knowledge graph. The starting point of the directed edge is the node representing the preceding action, and the ending point is the node representing the following action. This directed edge represents the causal link between the individual firefighter actions. Edges can also carry attributes that describe the specific characteristics of the causal link. Attributes include at least link type and link strength. Link type is a classification label used to distinguish between different types, such as physical condition dependence or information condition dependence. The link type label is determined during the verification step of determining the causal link. Link strength is a numerical attribute used to quantify the confidence or importance of the causal link. The link strength is assigned based on the sufficiency of evidence during the verification process. Sufficiency of evidence is measured by the quantity and quality of evidence types. For example, if both location sensor data and video data confirm physical condition dependence, the association strength is assigned a value of 0.9; if only a single type of data confirms it, the association strength is assigned a value of 0.6. The specific numerical range of the association strength is between 0 and 1, and this range is preset during system initialization. By traversing all determined causal transmission association pairs, corresponding edges are created in the knowledge graph, ultimately forming a directed graph structure connecting multiple individual firefighter action behavior nodes. This directed graph structure is the knowledge graph representing the causal relationships of actions in this training.

[0027] Based on the mutual exclusion constraint of resource occupancy and the consistency constraint of tactical objectives in the knowledge graph, the key causal chain from individual firefighter actions to team task results is identified, specifically through the following methods: The process iterates through all causal paths in the knowledge graph from individual firefighter actions to team task outcomes. Each individual firefighter action node in the knowledge graph is used as a potential starting node, and the team task outcome associated with that node is used as the target endpoint. A depth-first search (DFS) or breadth-first search (BFS) algorithm is employed to find all paths in the directed graph that originate from the starting node, pass through directed edges representing causal relationships, and ultimately reach the endpoint representing the same team task outcome. During the search, the sequence of individual firefighter actions along the path and the sequence of directed edges connecting them are recorded. Each complete path constitutes a candidate causal path from an individual firefighter action to a team task outcome. To control path complexity and ensure logical consistency, a maximum path length threshold is set. This threshold is a positive integer parameter, determined based on the typical length distribution of effective causal chains in historical training data; for example, based on historical data analysis, a maximum path length threshold of 10 is set. The search will stop when the number of individual firefighter action nodes on the search path exceeds the maximum path length threshold.

[0028] The process involves filtering causal transmission paths that do not conflict with individual firefighter actions in terms of resource usage in both time and space. Resource usage conflict refers to two or more individual firefighter actions competing for the same non-shareable physical or spatial resource within the same or overlapping time period. The judgment process is performed on each candidate causal transmission path. The start and end timestamps of each individual firefighter action node are extracted from the candidate causal transmission paths, forming a series of time intervals. Any two time intervals are checked for overlap. The criterion for time overlap is that the end timetamp of the earlier-ending action is no later than the start timetamp of the later-starting action. If two time intervals overlap, it is further checked whether the two overlapping individual firefighter actions occupy the same critical spatial resource. Critical spatial resources are determined through a predefined spatial resource mapping table, which associates different action type codes with one or more spatial location labels. For example, breaching actions are associated with door / window location labels, water spraying actions with fire source location labels, and interior attack actions with corridor location labels or room entrance location labels. Simultaneously, coordinate data from position sensors in individual firefighter action data records are used to obtain the core position coordinates of each action. If two firefighter actions overlap in time, have the same associated spatial location label, or the Euclidean distance between their core position coordinates is less than a preset spatial conflict distance threshold, then these two firefighter actions are determined to have a resource occupation conflict. The spatial conflict distance threshold is a length parameter, set based on the physical dimensions of the firefighter's personal protective equipment and the minimum safe interval required for coordinated operations; for example, the spatial conflict distance threshold is set to 2 meters. If no resource occupation conflict satisfying the above conditions exists in a candidate causal transmission path, then the path is determined to satisfy the mutual exclusion constraint of resource occupation.

[0029] From causal transmission paths that satisfy the mutual exclusion constraint of resource usage, the process filters out causal transmission paths where the tactical intentions of individual firefighter actions all point to the same team task outcome. This filtering process includes the following sub-steps: First, the top-level tactical objective is parsed from the team task outcome set in the training. The team task outcome is defined and stored in the training management terminal before training. The top-level tactical objective is the highest-level task purpose description extracted from the team task outcome. For example, for a team task outcome of successfully conducting indoor search and rescue and firefighting, the parsed top-level tactical objective could be controlling the fire and rescuing people. The parsing process is completed by querying a predefined task objective mapping table, which associates different team task outcome types with a standardized top-level tactical objective description string. Second, each individual firefighter action in the causal transmission path is mapped to a preset standard operating procedure library to obtain its corresponding standard tactical intent. The preset standard operating procedure library is a structured database that stores various standard firefighting action type codes and their corresponding standard tactical intent descriptions. The standard tactical intent description is a text string used to explain the tactical purpose to be achieved by performing the action. For example, the standard tactical intent description could be to open a passage, suppress the fire, search for and rescue people, or provide cover. The mapping process is achieved by matching the action type codes of individual firefighter actions in the causal transmission path with the action type codes in the standard operating procedure library. Upon successful matching, the corresponding standard tactical intent description is read. It then determines whether the standard tactical intents corresponding to all individual firefighter actions collectively support the top-level tactical objective. This determination relies on a pre-constructed tactical intent support relationship matrix. This matrix defines the support relationship and strength of various standard tactical intent descriptions for different top-level tactical objectives. Support strength is a value between 0 and 1, with higher values ​​indicating stronger support. The support strength values ​​are assigned based on the consensus assessment of domain experts regarding the correlation between firefighting tactical actions and global objectives, and are solidified into the matrix after multiple rounds of expert scoring and consistency verification. During the determination, the set of standard tactical intent descriptions for all individual firefighter actions on the current path is extracted, and the overall support of this set for the current top-level tactical objective is calculated. The overall support is calculated as follows: First, find the support strength value of each standard tactical intention description for the current top-level tactical objective from the tactical intention support relationship matrix. Then, sum all support strength values ​​to obtain a total. Finally, divide the total by the number of standard tactical intention descriptions in the set to obtain the average value. This average value is the overall support. An overall support threshold is set, which is a numerical parameter between 0 and 1. The threshold is determined by analyzing the tactical intentions in historical successful training cases. Figure 1The distribution characteristics of consistency scores are used to determine this, for example, setting the overall support threshold to 0.7. If the calculated overall support is greater than or equal to the overall support threshold, it is determined that the standard tactical intentions corresponding to all individual firefighter actions collectively support the top-level tactical objective. The causal transmission path of the standard tactical intentions collectively supporting the top-level tactical objective is preserved.

[0030] Causal transmission paths that satisfy the tactical objective consistency constraint are identified as critical causal chains. The causal transmission paths that remain after filtering by resource usage mutual exclusion constraints and tactical objective consistency constraints are the critical causal chains from individual firefighter actions to team mission outcomes. Each critical causal chain is stored as an independent data object, containing a unique identifier for the chain, a sequence of individual firefighter action nodes along the path, a sequence of directed edges connecting these nodes, and a calculated overall support value. These critical causal chain data objects will serve as explicit inputs for subsequent Monte Carlo perturbation simulations.

[0031] Monte Carlo perturbation simulation is performed on the execution process of the key causal chain to quantify the sensitivity of the key causal chain to execution perturbations and generate process sensitivity indicators. This is achieved in the following ways: Define time perturbation parameters and success rate perturbation parameters for individual firefighter actions included in the key causal chain. The time perturbation parameter is a mathematical description that specifies the range and distribution of the simulated execution time of each individual firefighter action, randomly fluctuating around its baseline execution time. The baseline execution time is the difference between the action end timestamp and the action start timestamp in the individual firefighter's action data record, measured in seconds. The specific form of the time perturbation parameter can be an interval range and a probability distribution type. For example, the time perturbation parameter can be defined as the simulated execution time following a normal distribution with a mean of 0.2 times the baseline execution time and a standard deviation of 0.5 times the baseline execution time, and restricted to between 0.5 times and 2 times the baseline execution time. The success rate perturbation parameter specifies the probability that each individual firefighter action may fail in a single simulation. The baseline success rate of each individual firefighter action is calculated by statistically analyzing the total number of times the firefighter performed the same action type encoding action in historical training data and the number of successful attempts, by dividing the number of successful attempts by the total number of attempts. If a firefighter lacks historical data on a specific action, the average success rate of that type of action, calculated based on historical data from all firefighters, is used as their baseline success rate. For example, if the overall average success rate of breaching actions is 0.88, obtained by analyzing the historical database, this average success rate of 0.88 can be used as the baseline success rate for breaching actions for new firefighters.

[0032] Based on defined time perturbation and success rate perturbation parameters, multiple random sampling simulations are performed on the execution process of the critical causal chain. Each random sampling simulation is a complete execution extrapolation of the critical causal chain under a random perturbation environment. At the start of the simulation, a random time delay following a specific distribution is superimposed on the baseline execution time of each individual firefighter's action in the critical causal chain to simulate time perturbation. The value of the random time delay is generated according to the defined time perturbation parameter for that individual firefighter's action. The generation process involves calling a random number generator to generate a random time change according to the probability distribution type and parameters specified by the time perturbation parameter; this time change is the random time delay. For example, if the time perturbation parameter is defined as the simulated execution time being uniformly randomized between 0.8 and 1.5 times the baseline execution time, a random number uniformly distributed within the interval 0.8 to 1.5 is generated as a scaling factor. The baseline execution time is multiplied by this scaling factor to obtain the simulated execution time, and then the simulated execution time is subtracted from the baseline execution time to obtain the random time delay. For each individual firefighter's action, a random number is generated based on its baseline success rate to determine the success of the action in the simulation, thus simulating success rate perturbations. Specifically, a random number uniformly distributed between 0 and 1 is generated and compared to the baseline success rate of the individual firefighter's action. If the random number is less than or equal to the baseline success rate, the individual firefighter's action is considered successful in the simulation; otherwise, it is considered a failure. Based on the randomly generated time delay and the success status of the action, the execution process of the entire critical causal chain and the final team task result are sequentially deduced. The deduction process follows the sequence of individual firefighter action nodes stored in the critical causal chain data object. The simulation start time for each individual firefighter's action is equal to the simulation end time of the previous individual firefighter's action plus any possible coordination interval, which can be set to a fixed value, such as 2 seconds, according to standard operating procedures. The simulation end time for each individual firefighter's action is equal to its simulation start time plus the baseline execution time plus the random time delay generated for that action. If a firefighter's individual action is deemed a failure, a predefined failure impact mapping table is used to look up the corresponding failure consequence for that action type code. The failure impact mapping table defines the effect of different action failures on subsequent actions; for example, the mapping table specifies that a failed breaching action will prevent all subsequent actions from starting. Through sequential calculations, the team's task achievement status for that simulation is finally obtained, and the criteria for judging the achievement status are completely consistent with the training settings.

[0033] The proportion of team task outcomes that failed to be achieved in multiple random sampling simulations is calculated. The total number of random sampling simulations is controlled by a pre-set simulation count threshold. This threshold is a positive integer, set based on the statistical precision requirements of the Monte Carlo method. Specifically, it can be set by calculating the minimum number of simulations required, given a confidence level (e.g., 95%) and an acceptable error range (e.g., 1%), using a sample size estimation formula in statistics. This minimum number of simulations is then rounded up to the nearest integer as the threshold. For example, if the minimum number of simulations is calculated to be 9604, the threshold is set to 10000. After completing the random sampling simulations up to the threshold count, the number of times the team task outcome failed to be achieved in all simulations is counted. The proportion of team task outcomes that failed to be achieved is equal to the number of times the team task outcome failed divided by the simulation count threshold. For example, after 10000 simulations, if the number of times the team task outcome failed to be achieved is 3200, the proportion of team task outcomes that failed to be achieved is calculated as 0.32.

[0034] A process sensitivity index is calculated based on the proportion of team task outcomes that were not achieved. This index, derived from the proportion of team task outcomes that were not achieved, is used to visually represent the stability of key causal chains. One direct method is to define the process sensitivity index as the proportion of team task outcomes that were not achieved, in which case the index ranges from 0 to 1. Another method, to enhance the differentiation between different key causal chains, is to linearly scale the proportion of team task outcomes that were not achieved, for example, by multiplying it by a scaling factor of 10, resulting in a process sensitivity index ranging from 0 to 10. The specific value of the scaling factor is pre-set according to the numerical range requirements of the actual scoring system. The final generated process sensitivity index is stored in association with a unique identifier for the key causal chain data object, serving as a quantitative output of the key causal chain's sensitivity to execution disturbances.

[0035] The contribution weight of individual firefighter actions on key causal chains to the team's task outcome is evaluated based on process sensitivity indicators, specifically through the following methods: The process sensitivity index is parsed to determine the perturbation sensitivity value of the critical causal chain. The process sensitivity index is a numerical value stored in association with the critical causal chain data object. The parsing process directly reads the value of the process sensitivity index as the perturbation sensitivity value of the critical causal chain. The perturbation sensitivity value is a dimensionless numerical value. For example, if the process sensitivity index of a critical causal chain is 0.25, then the perturbation sensitivity value of the critical causal chain is 0.25.

[0036] The weighting of individual firefighter actions within a critical causal chain is determined based on the perturbation sensitivity values ​​of that chain. This determination process employs a weighting mapping function. The input to this function is the perturbation sensitivity value of the critical causal chain, and the output is a baseline weight coefficient between 0 and 1. The weighting mapping function is a piecewise linear function. First, a high-sensitivity threshold and a low-sensitivity threshold need to be set. These thresholds are determined based on the distribution analysis of the perturbation sensitivity values ​​of all critical causal chains in historical training. Specifically, the perturbation sensitivity values ​​of all critical causal chains in historical training are sorted by numerical value, and the values ​​in the top 20% of the sorted values ​​are used as the high-sensitivity threshold, while the values ​​in the bottom 20% are used as the low-sensitivity threshold. For example, based on historical data, the high-sensitivity threshold is calculated to be 0.8, and the low-sensitivity threshold is 0.2. The specific calculation rules for the weight mapping function are as follows: If the perturbation sensitivity value of the input key causal chain is greater than or equal to the high sensitivity threshold, the output baseline weight coefficient is set to 0.9; if the perturbation sensitivity value of the input key causal chain is less than or equal to the low sensitivity threshold, the output baseline weight coefficient is set to 0.1; if the perturbation sensitivity value of the input key causal chain is between the low sensitivity threshold and the high sensitivity threshold, the baseline weight coefficient is obtained through linear interpolation. The specific steps of linear interpolation are: First, calculate the difference between the perturbation sensitivity value of the key causal chain and the low sensitivity threshold, i.e., the perturbation sensitivity value minus the low sensitivity threshold; Second, calculate the difference between the high sensitivity threshold and the low sensitivity threshold, i.e., the high sensitivity threshold minus the low sensitivity threshold; Third, divide the difference obtained in the first step by the difference obtained in the second step to obtain a ratio; Fourth, multiply the ratio by 0.8; Fifth, add 0.1 to the result obtained in the fourth step, and the final value is the baseline weight coefficient. For example, if the perturbation sensitivity value of a key causal chain is 0.5, the low sensitivity threshold is 0.2, and the high sensitivity threshold is 0.8, then the calculated ratio is (0.5-0.2) / (0.8-0.2)=0.5, 0.5×0.8=0.4, 0.4+0.1=0.5, so the baseline weight coefficient is 0.5.

[0037] Assigning baseline weight coefficients to individual firefighter actions on the critical causal chain requires two auxiliary factors: a positional influence factor and an edge strength factor. The positional influence factor is calculated based on the sequential position of the individual firefighter action within the critical causal chain node sequence. The first node in the sequence is assigned a positional influence factor of 1.5, and the last node is assigned a positional influence factor of 1.0. For nodes in the middle of the sequence, the positional influence factor decreases linearly from the first node to the last. The linear decrease is calculated as follows: if the node's position number in the sequence is i, and the total number of nodes in the sequence is n, then the positional influence factor of that node is equal to 1.5 - 0.5 × [(i-1) / (n-1)]. The edge strength factor is calculated based on the association strength attribute values ​​of the directed edges originating from the node of the individual firefighter action. The association strength attribute values ​​of all outgoing edges from that node are read, and the arithmetic mean of these attribute values ​​is calculated. This arithmetic mean is used as the edge strength factor. The initial weight allocation ratio for each individual firefighter action is equal to the baseline weight coefficient × the node's positional influence factor × the node's edge strength factor. Next, the initial weight allocation ratios calculated for all nodes are normalized. The specific steps of the normalization process are as follows: First, calculate the sum of the initial weight allocation ratios for all nodes in the critical causal chain; second, divide the initial weight allocation ratio of each node by the sum calculated in the first step to obtain a normalized ratio; third, multiply the normalized ratio of each node by the baseline weight coefficient to obtain the final weight allocation ratio of the firefighter's individual action behavior within the critical causal chain.

[0038] The contribution weight of each firefighter's individual actions on the key causal chain to the team's task outcome is calculated according to the weight allocation ratio. Calculating the contribution weight requires a global contribution budget value. The contribution budget value is a pre-set positive constant based on the scoring system requirements; for example, the contribution budget value is set to 100. First, the sum of the baseline weight coefficients of all key causal chains needs to be calculated. Then, the contribution share that the current key causal chain should receive is calculated. The contribution share equals the contribution budget value multiplied by the baseline weight coefficient of the current key causal chain divided by the sum of the baseline weight coefficients of all key causal chains. Finally, the contribution weight of each firefighter's individual actions on the key causal chain is calculated. The contribution weight equals the contribution share multiplied by the final weight allocation ratio of that firefighter's individual action within the current key causal chain. For example, if there are two key causal chains, chain A has a baseline weight coefficient of 0.8, and chain B has a baseline weight coefficient of 0.6, then the sum of the baseline weight coefficients is 1.4. Assuming the contribution budget value is 100, then the contribution share of chain A is 100 × 0.8 / 1.4 ≈ 57.14. If the final weight allocation ratio of a firefighter's individual action on chain A is 0.2, then the contribution weight of that individual action is 57.14 × 0.2 = 11.428. After the contribution weight of each firefighter's individual action is calculated, it is stored in association with the unique identifier of that individual action data record.

[0039] The individual firefighter's multidimensional comprehensive score and the team's multidimensional comprehensive score are calculated based on contribution weights, specifically through the following methods: For each firefighter, the contribution weights of their individual actions along the key causal chain are aggregated to obtain a total individual contribution weight. The aggregation process uses each firefighter's identifier as the search key. All calculated individual firefighter action data records are retrieved from the storage system. Each record contains an identifier field and a contribution weight field. All retrieved records are grouped according to the identifier field value, with records having the same identifier field value grouped together. For each group, the contribution weight field values ​​of all firefighter action data records within that group are summed. The summation operation involves creating an accumulator variable with an initial value of 0, then iterating through each record in the group, adding the record's contribution weight field value to the accumulator variable. After the iteration is complete, the final value of the accumulator variable is the total individual contribution weight of the firefighter corresponding to that identifier. For example, a firefighter with the firefighter identification number 001 has 5 individual action behavior data records in his group, with contribution weight field values ​​of 3.2, 5.1, 2.8, 4.0, and 6.5 respectively. The accumulation process is 0 + 3.2 = 3.2, 3.2 + 5.1 = 8.3, 8.3 + 2.8 = 11.1, 11.1 + 4.0 = 15.1, 15.1 + 6.5 = 21.6, and the total individual contribution weight is 21.6.

[0040] The individual multidimensional comprehensive score of firefighters is calculated based on the sum of their individual contribution weights. The calculation process uses a linear scaling method to map the sum of individual contribution weights to a score range of 0 to 100. First, a global reference value needs to be calculated. The global reference value is the maximum value among the sums of individual contribution weights of all firefighters in this training. The maximum value is obtained by creating a list to store all the sums of individual contribution weights after calculating the sums of individual contribution weights for all firefighters, and then iterating through the list to find the maximum value as the global reference value. Then, for each firefighter, their individual multidimensional comprehensive score is calculated using the following formula: Firefighter Individual Multidimensional Comprehensive Score = 100 × Sum of Individual Contribution Weights / Global Reference Value. The division operation uses floating-point arithmetic, and the result is rounded to one decimal place. Since the sum of individual contribution weights is always less than or equal to the global reference value, the calculated score ranges from 0 to 100. For example, if the maximum sum of individual contribution weights for all firefighters in this training is 25.0, and the sum of individual contribution weights for firefighter identifier 001 is 21.6, then their individual multidimensional comprehensive score is 100 × 21.6 / 25.0 = 86.4. If the sum of individual contribution weights is 0, then their individual multidimensional comprehensive score is directly determined to be 0.

[0041] The team's multidimensional comprehensive score is obtained by combining the individual multidimensional comprehensive scores of all firefighters. The team's multidimensional comprehensive score is calculated using the arithmetic mean method. First, the individual multidimensional comprehensive scores of all firefighters in this training are collected, forming a score list. The score list contains N elements, where N equals the total number of firefighters participating in the training. The arithmetic mean of all elements in the score list is calculated. The calculation steps are: first, initialize a summation variable to 0; second, iterate through the score list, adding each score value in the list to the summation variable; third, after completing the iteration, divide the summation variable by N, and the result is the team's multidimensional comprehensive score. Floating-point arithmetic is used for division, and the result is rounded to one decimal place. For example, if five firefighters participate in a training exercise, and their individual multidimensional comprehensive scores are listed as [86.4, 78.2, 92.0, 81.5, 88.9], then the summation variable is calculated as 86.4 + 78.2 + 92.0 + 81.5 + 88.9 = 427.0. Since N equals 5, the team multidimensional comprehensive score is 427.0 / 5 = 85.4. After the team multidimensional comprehensive score is calculated, it is stored in conjunction with the team task result data record for this training exercise.

[0042] Example 2: Figure 2 A schematic diagram of a multi-dimensional index comprehensive scoring management system for fire training is provided according to the present invention. The multi-dimensional index comprehensive scoring management system for fire training includes: The data acquisition module is used to acquire individual action and behavior data of firefighters and team task results during real fire and smoke heat simulation training; The association judgment module is used to determine whether there is a causal transmission relationship between individual firefighter actions based on individual firefighter action data. If so, a knowledge graph is constructed with individual firefighter actions as nodes and causal transmission relationships between individual firefighter actions as edges. The causal chain identification module is used to identify key causal chains from individual firefighter actions to team mission outcomes in a knowledge graph based on mutual exclusion constraints of resource usage and consistency constraints of tactical objectives. The disturbance simulation module is used to perform Monte Carlo disturbance simulation on the execution process of key causal chains to quantify the sensitivity of key causal chains to execution disturbances and generate process sensitivity indicators. The weighting evaluation module is used to evaluate the contribution weight of individual firefighter actions on key causal chains to the team's task results based on process sensitivity indicators. The scoring calculation module is used to comprehensively calculate the individual multidimensional comprehensive score of firefighters and the multidimensional comprehensive score of teams based on their contribution weights.

[0043] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0044] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0045] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. Computer-readable storage media can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0046] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0047] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0048] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0049] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0050] If a function is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0051] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0052] In conclusion, the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-dimensional index-based comprehensive scoring management method for fire training, characterized in that, include: S1. Obtain individual action and behavior data of firefighters and team task results during real fire smoke and heat simulation training; S2. Based on the individual firefighter action data, determine whether there is a causal transmission relationship between individual firefighter actions. If so, construct a knowledge graph with individual firefighter actions as nodes and causal transmission relationships between individual firefighter actions as edges. S3. Based on the mutual exclusion constraint of resource occupation and the consistency constraint of tactical objectives in the knowledge graph, identify the key causal chain from individual firefighter actions to team task results; S4. Perform Monte Carlo perturbation simulation on the execution process of the key causal chain to quantify the sensitivity of the key causal chain to execution perturbations and generate process sensitivity indicators. S5. Evaluate the contribution weight of individual firefighter actions on key causal chains to team task results based on process sensitivity indicators. S6. Based on the contribution weight, calculate the individual multidimensional comprehensive score of the firefighter and the multidimensional comprehensive score of the team.

2. The multi-dimensional index comprehensive scoring management method for fire training according to claim 1, characterized in that, S1 includes: Data on individual firefighter actions and behaviors are collected using sensors and audio / video recording devices in training facilities. Record team task results through the training management terminal; The individual actions and behaviors of firefighters are time-stamped and linked with the team's task results for storage.

3. The fire training multi-dimensional index comprehensive scoring management method according to claim 1, characterized in that, S2 include: Analyze the dependencies between actions in individual firefighter behavior data; By considering the chronological order, determine whether there is a causal relationship where the subsequent action depends on the conditions created by the preceding action. If a causal relationship exists, a knowledge graph is constructed using individual firefighter actions as nodes. In the knowledge graph, the corresponding individual firefighter action nodes are connected by edges based on the causal transmission relationship that is determined to exist.

4. The multi-dimensional index comprehensive scoring management method for fire training according to claim 1, characterized in that, S3 includes: Traverse the entire causal transmission path in the knowledge graph from individual firefighter actions to team task results; We selected causal transmission paths that satisfy the condition that there is no conflict in resource occupation between individual firefighter actions in the temporal and spatial dimensions; From the causal transmission paths that satisfy the mutual exclusion constraint of resource occupation, we select the causal transmission paths in which the tactical intentions of individual firefighter actions all point to the same team task outcome; The causal transmission paths that satisfy the tactical objective consistency constraint are identified as key causal chains.

5. The multi-dimensional index comprehensive scoring management method for fire training according to claim 4, characterized in that, The selection process involves filtering out causal transmission paths where the tactical intentions of individual firefighter actions all point to the same team mission outcome. This includes: analyzing the top-level tactical objective based on the team mission outcome set in the training; mapping each individual firefighter action in the causal transmission path to a pre-set standard operating procedure library to obtain its corresponding standard tactical intention; determining whether the standard tactical intentions corresponding to all individual firefighter actions collectively support the top-level tactical objective; and retaining causal transmission paths where the standard tactical intentions collectively support the top-level tactical objective.

6. The fire training multi-dimensional index comprehensive scoring management method according to claim 1, characterized in that, S4 include: Define time perturbation parameters and success rate perturbation parameters for individual firefighter actions included in the key causal chain; Based on the defined time perturbation parameters and success rate perturbation parameters, the execution process of the key causal chain is simulated by multiple random sampling. The percentage of team task results that were not achieved in multiple random sampling simulations was statistically analyzed. A process sensitivity index is generated based on the proportion of team task results that were not achieved.

7. The multi-dimensional index comprehensive scoring management method for fire training according to claim 6, characterized in that, Based on the defined time perturbation parameters and success rate perturbation parameters, the execution process of the critical causal chain is simulated by multiple random sampling, including: for each individual firefighter's action in the critical causal chain, a random time delay following a specific distribution is superimposed on its baseline execution time to simulate time perturbation; for each individual firefighter's action, a random number is generated based on its baseline success rate to determine whether the action was successful in this simulation to simulate success rate perturbation; based on the randomly generated time delay and the success status of the action, the execution process of the entire critical causal chain and the final team task result are sequentially deduced.

8. The fire training multi-dimensional index comprehensive scoring management method according to claim 1, characterized in that, S5 include: Analyze process sensitivity metrics to determine the degree of sensitivity to disturbances in key causal chains; The weighting ratio of individual firefighter actions on the critical causal chain is set according to the perturbation sensitivity value of the critical causal chain; The contribution weight of each firefighter's individual actions to the team's task outcome is calculated according to the weight allocation ratio.

9. The multi-dimensional index comprehensive scoring management method for fire training according to claim 1, characterized in that, S6 include: For each firefighter, the contribution weight of each firefighter's individual actions and behaviors on the key causal chain is summarized to obtain the sum of individual contribution weights; Calculate the firefighter's individual multidimensional comprehensive score based on the sum of individual contribution weights; The team's multidimensional comprehensive score is obtained by combining the individual multidimensional comprehensive scores of all firefighters.

10. A fire training multi-dimensional index comprehensive scoring management system, used to implement the fire training multi-dimensional index comprehensive scoring management method according to any one of claims 1-9, characterized in that, include: The data acquisition module is used to acquire individual action and behavior data of firefighters and team task results during real fire and smoke heat simulation training; The association judgment module is used to determine whether there is a causal transmission relationship between individual firefighter actions based on individual firefighter action data. If so, a knowledge graph is constructed with individual firefighter actions as nodes and causal transmission relationships between individual firefighter actions as edges. The causal chain identification module is used to identify key causal chains from individual firefighter actions to team mission outcomes in a knowledge graph based on mutual exclusion constraints of resource usage and consistency constraints of tactical objectives. The disturbance simulation module is used to perform Monte Carlo disturbance simulation on the execution process of key causal chains to quantify the sensitivity of key causal chains to execution disturbances and generate process sensitivity indicators. The weighting evaluation module is used to evaluate the contribution weight of individual firefighter actions on key causal chains to the team's task results based on process sensitivity indicators. The scoring calculation module is used to comprehensively calculate the individual multidimensional comprehensive score of firefighters and the multidimensional comprehensive score of teams based on their contribution weights.

Images