A Multi-Agent-Based Method and Device for Trajectory Prediction and Interception Control of Minor Injury Groups

By modeling minor injuries as intelligent agents and combining explicit and implicit risk fields, the irrational escape trajectories of minor injury groups can be predicted and intercepted. This solves the problem of lagging risk assessment of secondary injuries in fire and explosion accidents, and achieves high-precision risk prediction and rescue intervention.

CN123118334APending Publication Date: 2026-11-17BEIJING GLOBAL SAFETY TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611303965.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-26
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In large-scale fire and explosion accidents, the irrational escape behavior of the lightly injured makes it difficult to predict the direction of their escape during rescue operations, and the subsequent rescue and dispatch may cause secondary damage. Existing technologies cannot perform dynamic analysis within a unified spatiotemporal framework, resulting in delayed risk warnings.

Method used

Each person with minor injuries in the group is modeled as an independent intelligent agent. The escape trajectory is predicted by the combined force of explicit risk repulsion, safety exit attraction, and panic herd mentality. Explicit and implicit risk fields are constructed, and the high-risk escape trajectory is traced back to select the best blocking point for interception and guidance.

Benefits of technology

It improves the accuracy of predicting the escape trajectory of lightly injured groups, enables high-precision dynamic assessment of the risk of secondary injury, and allows for intervention before the optimal blocking point, thereby increasing the success rate of rescue.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123118334A_ABST
    Figure CN123118334A_ABST
Patent Text Reader

Abstract

This application discloses a method and device for predicting and intercepting the trajectory of a group of lightly injured individuals based on multi-agent intelligence, belonging to the fields of emergency management and emergency medicine. The method involves acquiring disaster environment data and group data of the lightly injured group in the target area; dividing the target area into grids based on the disaster environment data; calculating explicit and implicit risk values ​​based on the disaster environment data; calculating the resultant force on the agents within the group of lightly injured individuals based on the disaster environment data, group data, and explicit risk values; generating escape trajectories based on the grid sequence traversed by the agents during movement, using the resultant force prediction; identifying high-risk escape trajectories that could worsen the injuries based on the escape trajectories and implicit risk values; and traversing back along the high-risk escape trajectories to select grids that meet preset safety constraints as optimal blocking points, intercepting and guiding the evacuation of the lightly injured group at these points. This application improves the accuracy of assessing group escape trajectories and the risk of secondary injury, thereby increasing the success rate of rescue efforts.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the fields of emergency management and emergency medicine, and in particular to a method and device for predicting and intercepting the trajectory of a group of people with minor injuries based on multi-agent systems. Background Technology

[0002] In emergency rescue operations following large-scale fires and explosions, rescue efforts primarily focus on life support and rapid evacuation of the seriously injured. However, in the complex and ever-changing disaster scene, the mobile, lightly injured population constitutes a significant blind spot in rescue management. Under the dual pressures of extreme panic and limited environmental perception, such as the invisibility of colorless toxic gases, the lightly injured are highly susceptible to irrational escape behaviors, including blind running and unwittingly entering dangerous areas. This uncontrolled movement not only leads to secondary injuries such as the spread of toxic gases and secondary explosions, causing further damage and even death, but also obstructs rescue channels due to disorderly crowd movement, severely disrupting the overall rescue operation. Therefore, guiding the safe and orderly evacuation of the lightly injured has become an urgent problem to be solved.

[0003] In related technologies, it is generally assumed that lightly injured individuals will remain in place awaiting rescue, or that they will have access to complete disaster environmental data and evacuate safely along the shortest path. However, in real disaster scenarios, driven by the shockwave of an explosion and panic, lightly injured individuals often exhibit strong blindness and herd mentality. Currently, it is difficult to characterize this irrational escape trajectory caused by localized information asymmetry (such as avoiding only visible fire sources while ignoring hidden toxic gases), making it impossible for rescue command centers to predict the direction of lightly injured individuals' dispersal. Furthermore, current dispatch and command rely on the location of lightly injured individuals reported back from the front end for post-disaster dispatch and rescue. By the time rescue forces arrive at the current location of lightly injured individuals, they may have already moved to a danger zone. In addition, existing technologies cannot overlay and analyze dynamically changing personnel trajectories and dynamically evolving disaster fields within a unified spatiotemporal framework. This makes it difficult for rescue command centers to quantitatively assess the risk of "light injury turning into serious injury" faced by lightly injured groups during movement, resulting in a significant lag in early warning of the risk of secondary injury to lightly injured individuals. Therefore, there is an urgent need for a method that can predict and intervene in the irrational escape behavior of slightly injured individuals. Summary of the Invention

[0004] This application provides a method and device for predicting, intercepting, and controlling the trajectory of lightly injured groups based on multi-agent systems. It addresses the problem that the irrational escape trajectories of lightly injured groups under conditions of limited perception and panic cannot be predicted, and that subsequent rescue and dispatch may lead to secondary injuries for these groups. The technical solution is as follows: According to a first aspect of this application, a method for predicting and intercepting the trajectory of a group of lightly injured individuals based on a multi-agent system is provided, the method comprising: Acquire disaster environmental data and minor injury group data of the target area, and divide the target area into multiple grids based on the disaster environmental data; The explicit risk value and implicit risk value of each grid are calculated based on the disaster environment data. The explicit risk value represents the degree of risk of the grid being affected by fire or smoke, and the implicit risk value represents the degree of risk of the grid being affected by toxic gas diffusion and / or secondary explosion. Based on the disaster environment data, the group data, and the explicit risk value, the resultant force on each agent in the slightly injured group is calculated. The resultant force includes explicit risk repulsion force, safe exit attraction force, and panic herd force. Based on the resultant force, the grid sequence that the agent will move through is predicted, and the grid sequence is used as the agent's escape trajectory. Based on the escape trajectory and the hidden risk value, the escape trajectory that would lead to the deterioration of the agent's injury is identified as a high-risk escape trajectory. Tracing back along the direction of movement of the high-risk escape trajectory, a grid that meets the preset safety constraints is selected as the optimal blocking point. The group of lightly injured people is intercepted and guided to evacuate at the optimal blocking point.

[0005] In one possible implementation, the calculation of the explicit risk value and implicit risk value for each grid based on the disaster environment data includes: The fire spread model is used to calculate the fire source location, heat release rate and smoke generation rate in the disaster environment data to obtain the explicit risk value of each grid. An obstacle-corrected Gaussian plume model is used to calculate the hazardous chemical data, ambient wind speed, and ambient wind direction in the disaster environment data to obtain the toxic gas concentration in each grid. A cumulative heat dose triggering model is used to calculate the hazardous chemical data and heat release rate in the disaster environment data to obtain the cumulative heat dose of each explosive object. When the cumulative heat dose of any explosive object reaches a critical value, a secondary explosion probability is set for the grid within a preset radius around the explosive object. The toxic gas concentration and / or the secondary explosion probability are used as the implicit risk value of each grid.

[0006] In one possible implementation, calculating the resultant force on each agent in the lightly injured group based on the disaster environment data, the group data, and the explicit risk value includes: Based on the agent distribution data in the group data, generate a corresponding number of agents in each passable grid, and obtain the current position and panic value of each agent from the group data; Based on the current location, the disaster environment data, and the explicit risk value, determine the explicit risk sources within the perception range of each agent, and calculate the explicit risk repulsion force from the explicit risk sources to the agent; Based on the current location, the location of safety exits in the disaster environment data, and obstacle data, calculate the safety exit attractiveness from which the agent points to the nearest unobstructed safety exit. The panic value is updated based on the explicit risk value, and the panic herd force experienced by the agent is calculated based on the updated panic value and the overall movement direction of other agents in the agent's neighborhood. The combined force is obtained by vector-superimposing the explicit risk repulsion force, the safe exit attraction force, and the panic herd mentality force.

[0007] In one possible implementation, predicting the grid sequence traversed by the agent during its movement based on the resultant force includes: The acceleration of the agent is calculated based on the resultant force and the agent's mass in the group data. Calculate the agent's current velocity based on the acceleration; Calculate the current position of the agent based on the current speed; The current position at different time steps is mapped to the corresponding grid, and multiple grids are combined into a grid sequence according to the time step order.

[0008] In one possible implementation, determining an escape trajectory that would worsen the agent's condition as a high-risk escape trajectory based on the escape trajectory and the latent risk value includes: Calculate the agent's breathing enhancement coefficient based on the current speed in the escape trajectory; Based on the latent risk value of the grid where the agent is located and the breathing enhancement coefficient, the cumulative inhaled toxic dose of the agent during the movement is calculated. If the cumulative inhaled toxic dose exceeds the dose threshold, or if the escape trajectory contains a grid that will fall into the secondary explosion's influence area in the future, then the escape trajectory will be determined as a candidate escape trajectory. Spatiotemporal clustering is performed on all candidate escape trajectories, and the resulting group movement streamlines are identified as high-risk escape trajectories.

[0009] In one possible implementation, the step of tracing back along the movement direction corresponding to the high-risk escape trajectory and selecting a grid that meets preset safety constraints as the optimal blocking point includes: Determine the entry point from the high-risk escape route into the hidden risk area; Starting from the entry point, perform grid backtracking in the opposite direction to the movement direction corresponding to the high-risk escape trajectory; The first grid that satisfies all preset safety constraints is determined as the optimal blocking point. The preset safety constraints include at least one of the following: the distance between the optimal blocking point and the group of minor injuries is greater than the safe buffer distance; the optimal blocking point is located within the effective operating range of the rescue force; and the openness around the optimal blocking point meets the group's stay conditions.

[0010] In one possible implementation, intercepting and guiding the lightly injured group to evacuate at the optimal blocking point includes: Calculate the movement direction corresponding to the high-risk escape trajectory at the optimal blocking point, and the safe direction from the optimal blocking point to the nearest safe area; Calculate the deflection angle from the moving direction to the safe direction; Generate an interception and control command that includes the location of the optimal blocking point and the deflection angle, so as to intercept and guide the evacuation of the lightly injured group through the interception and control command.

[0011] According to a second aspect of this application, a device for predicting, intercepting, and controlling the trajectory of a group of minor injuries based on a multi-agent system is provided, the device comprising: The data acquisition module is used to acquire disaster environmental data and minor injury group data of the target area, and divide the target area into multiple grids based on the disaster environmental data; The risk calculation module is used to calculate the explicit risk value and implicit risk value of each grid based on the disaster environment data. The explicit risk value represents the degree of risk of the grid being affected by fire or smoke, and the implicit risk value represents the degree of risk of the grid being affected by toxic gas diffusion and / or secondary explosion. The trajectory prediction module is used to calculate the resultant force on each agent in the slightly injured group based on the disaster environment data, the group data, and the explicit risk value. The resultant force includes explicit risk repulsion force, safety exit attraction force, and panic herd mentality force. Based on the resultant force, the module predicts the grid sequence that the agent will move through, and uses the grid sequence as the agent's escape trajectory. The trajectory filtering module is used to identify high-risk escape trajectories as those that would worsen the injury of the intelligent agent, based on the escape trajectory and the hidden risk value. The interception and control module is used to trace back along the movement direction of the high-risk escape trajectory, select a grid that meets the preset safety constraints as the best blocking point, and intercept and guide the lightly injured group to evacuate at the best blocking point.

[0012] According to a third aspect of this application, a computer-readable storage medium is provided, wherein at least one instruction is stored therein, the at least one instruction being loaded and executed by a processor to implement the multi-agent-based method for predicting and intercepting the trajectory of a group of minor injuries as described above.

[0013] According to a fourth aspect of this application, an electronic device is provided, the electronic device comprising the above-described multi-agent-based trajectory prediction and interception control device for minor injuries.

[0014] The beneficial effects of the technical solution provided in this application include at least the following: By modeling each slightly injured person in a group of individuals as an independent intelligent agent, and then predicting the escape trajectory of the agent based on the combined force of explicit risk repulsion, safety exit attraction, and panic herd mentality, it is possible to simulate the irrational escape behavior of the slightly injured group under limited perception and panic-driven conditions, thereby improving the accuracy of group escape trajectory prediction in complex stress environments.

[0015] By constructing two logically decoupled evolutionary perspectives—an explicit risk field and a implicit risk field—where the explicit risk field acts as the driving force determining the direction of crowd movement, and the implicit risk field acts as the injury source determining changes in the crowd's injuries, a logical separation of behavioral drivers and injury sources is achieved. Based on this, the respiratory enhancement coefficient is calculated according to the current speed in the escape trajectory, and the cumulative inhaled toxic dose is calculated by combining it with the implicit risk value. Furthermore, the system monitors whether the escape trajectory falls into the area affected by a secondary explosion, enabling a high-precision dynamic assessment of the risk of secondary injury to the lightly injured group.

[0016] Pre-set safety constraints ensure that rescue forces can reach the optimal containment point before the lightly injured group arrives, and that a safe buffer distance is maintained between the optimal containment point and the entry point, while also satisfying the requirement for open space to prevent trampling. This allows for the determination of intervention locations before the lightly injured group enters the hidden risk area. Based on this, the deflection angle between the high-risk escape trajectory's movement direction at the optimal containment point and the safe direction pointing to the nearest safe area is calculated. This generates an interception and control command that includes the optimal containment point location and the deflection angle, achieving source-level containment of risk while minimizing resource consumption and improving the rescue success rate. Attached Figure Description

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

[0018] Figure 1This is a flowchart of a method for predicting and intercepting the trajectory of a group of minor injuries based on multi-agent systems, provided in one embodiment of this application; Figure 2 This is a flowchart of a method for predicting and intercepting the trajectory of a group of minor injuries based on multi-agent systems, provided in one embodiment of this application; Figure 3 This is a structural block diagram of a multi-agent-based trajectory prediction and interception control device for minor injuries provided in one embodiment of this application; Figure 4 This is a structural block diagram of an electronic device provided in one embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0020] This application proposes a method for predicting and intercepting the trajectory of minorly injured groups based on the evolution of physical disaster fields and the dynamics of mass stress behavior. The overall process includes: multi-source data acquisition and two-dimensional gridded scene reconstruction, construction of a dynamic risk field based on the explicit-implicit separation mechanism, evolution of irrational crowd trajectories driven by panic, spatiotemporal collision detection and calculation of injury status changes, and generation of proactive control strategies. The core of this application lies in constructing two logically decoupled but spatiotemporally coupled evolutionary perspectives: one is the explicit risk field (fire / smoke) and socio-psychological field as the driving source, which determines where the crowd runs; the other is the implicit risk field (poison / explosion) as the injury source, which determines the severity of the crowd's injuries. By calculating the spatiotemporal intersection of the two through a "risk-behavior coupling engine," the quantitative prediction and proactive prevention of secondary injuries can be achieved. The method for predicting and intercepting the trajectory of minorly injured groups is described in detail below.

[0021] like Figure 1 The diagram illustrates a flowchart of a method for predicting and intercepting the trajectory of a minor injury group based on a multi-agent system, according to an embodiment of this application. This method can be applied to electronic devices. The method may include: Step 101: Obtain disaster environmental data and group data of the slightly injured population in the target area, and divide the target area into multiple grids based on the disaster environmental data.

[0022] The target area is the area where a disaster such as a fire, explosion, or toxic gas leak has occurred.

[0023] The group with minor injuries consists of injured persons within the target area who are mobile but have impaired judgment, typically corresponding to those in the green zone in the on-site triage.

[0024] Disaster environmental data is used to describe the environment of a disaster, including but not limited to target area data, ignition source data, environmental data, physical property data, and hazardous chemical data. These data will be explained below.

[0025] (1) Target area data The target area data is oblique photogrammetry or Geographic Information System (GIS) data of the target area.

[0026] When collecting data from the target area, you can select the main activity plane of people (such as the ground level or a specific floor plan), extract the building outline, wall and obstacle information of the target area, and then generate a two-dimensional obstruction layer through rasterization. obs ), and clearly define a unified coordinate system.

[0027] (2) Fire source data (D fire ) Fire source data (D) fire ) at least includes the location of the fire source (P) fire ), heat release rate (HRR(t)) and flue gas formation rate (S smoke (t)).

[0028] Location of fire source (P) fire ) is the location coordinates (x0, y0) of the fire source within the target area.

[0029] The heat release rate (HRR(t)) is a function that describes the change of fire intensity with time t, and its unit is MW.

[0030] Flue gas generation rate (S) smoke (t) is a function describing the change of flue gas output with time t, and its unit is kg / s.

[0031] (3) Environmental data (D) env ) Environmental data (D) env At least ambient wind speed (U) ∞ ), environmental wind direction (D) wind ), ambient temperature (T) ∞ ) and environmental pressure (P ∞ ).

[0032] Ambient wind speed (U ∞ () is the wind speed at the reference height, in m / s.

[0033] Environmental wind direction (D) wind ) is the unit vector of wind direction (d) x ,d y ).

[0034] Ambient temperature (T) ∞ () is the atmospheric temperature of the target area, in K.

[0035] Environmental stress (P) ∞ () is the atmospheric pressure of the target area, in Pa.

[0036] (4) Physical attribute data (D prop ) Physical attribute data (D) prop At least air density (ρ) air ), specific heat capacity (C) p,air ), dynamic viscosity (μ) air Fluid properties such as fluid physical parameters.

[0037] (5) Hazardous Chemicals Data (D) chem ) Hazardous Chemicals Data (D) chem This should at least include the storage location, type, explosive equivalent (e.g., TNT equivalent), leakage threshold, and critical heating parameter (Q) of hazardous chemicals within the target area. crit )wait.

[0038] After acquiring disaster environmental data, the target area is defined as a two-dimensional computational domain Ω. 2D The grid is discretized into M×N grid cells, denoted as Cell(i,j), using a square grid. Then, each cell Cell(i,j) is traversed. If there are walls, obstacles, or impassable terrain within the cell, it is marked as impassable; otherwise, it is marked as passable.

[0039] The group data includes at least the agent distribution data of the slightly injured group within the target area and the status data of each slightly injured person.

[0040] The agent's distributed data is an external injury prediction model (D). triage The output of the two-dimensional injury distribution heatmap or distribution data can indicate the number or distribution density of minor injuries at different locations, which can be used to generate the corresponding number of agents in each accessible grid Cell(i,j). Here, an agent is an independent individual formed by modeling each minor injury patient in the minor injury group.

[0041] State data includes the position vector of each agent. Velocity vector Health status H m Panic level ε m respiratory enhancement coefficient α RMV,m and cumulative exposure dose Dose m .

[0042] Position vector The coordinates (x, y) of the agent within a two-dimensional grid. m ,y m ).

[0043] velocity vector It is the speed (v) of the intelligent agent. x ,v y ), with an initial value of 0.

[0044] Health status H m The values ​​{G, Y, R, B} correspond to minor injury (Green), moderate injury (Yellow), serious injury (Red), and death (Black), respectively. In this embodiment, we focus on H. m =G group.

[0045] Panic level ε m It is a value within the range [0,1], set according to the planar distance between the agent and the fire source. Initial value ε m (0) =exp(-d fire_to_m / L decay ), where d fire_to_m L is the Euclidean distance between the agent and the fire source. decay It is the panic decay constant.

[0046] respiratory enhancement coefficient α RMV,m The initial value is 1.0.

[0047] Cumulative exposure dose Dose m The initial value is 0.

[0048] Step 102: Calculate the explicit risk value and implicit risk value of each grid based on the disaster environment data. The explicit risk value represents the degree of risk of the grid being affected by fire or smoke, and the implicit risk value represents the degree of risk of the grid being affected by the spread of toxic gas and / or secondary explosion.

[0049] The explicit risk value represents the degree of risk that the grid will be affected by fire or smoke, and is used to drive the movement behavior of the agent.

[0050] The latent risk value represents the degree of risk that the grid will be affected by the spread of toxic gas and / or secondary explosions, and is used to assess the risk of secondary damage to the agent.

[0051] The calculation of explicit and implicit risk values ​​is explained in detail below.

[0052] Step 103: Calculate the resultant force on each agent in the lightly injured group based on disaster environment data, group data, and explicit risk values. The resultant force includes explicit risk repulsion force, safety exit attraction force, and panic herd mentality force. Predict the grid sequence that the agent will move through based on the resultant force, and use the grid sequence as the agent's escape trajectory.

[0053] The explicit risk repulsion force is the repulsive force directed from the explicit risk source within the agent's perception range towards the agent. The safety exit attraction force is the attractive force that pulls the agent towards the nearest unobstructed safety exit. The panic herd mentality force is the following force acting on the agent based on the overall movement direction of other agents in the agent's neighborhood. The resultant force is obtained by vector-wise superimposing the explicit risk repulsion force, the safety exit attraction force, and the panic herd mentality force, which drives the agent's movement. Since the agent can only perceive explicit risk sources and not implicit risk sources, and the higher the panic level, the stronger the herd mentality, it is possible to simulate the irrational escape behavior exhibited by a slightly injured group under conditions of limited perception and panic.

[0054] After obtaining the resultant force, the position of the agent is updated according to the resultant force. The positions at different time steps are mapped to the corresponding grids. Multiple grids are combined into a grid sequence according to the time step order to obtain the escape trajectory of the agent.

[0055] Step 104: Based on the escape trajectory and the hidden risk value, the escape trajectory that will cause the agent's injury to worsen is identified as a high-risk escape trajectory.

[0056] The worsening of the injury refers to the change in the health status of an intelligent agent from minor injury to serious injury or death.

[0057] Specifically, each escape trajectory can be judged based on the escape path and the hidden risk value. If the escape path will lead to the worsening of the agent's injury, these escape paths will be aggregated to obtain the high-risk escape path. For details on the specific judgment method, please refer to the explanation below.

[0058] Step 105: Backtrack along the direction of movement of the high-risk escape trajectory, select the grid that meets the preset safety constraints as the best blocking point, and intercept and guide the lightly injured group to evacuate at the best blocking point.

[0059] Reverse tracing refers to finding the best blocking point on a high-risk escape route, starting from the end closest to the danger zone and moving away from it, so as to intercept and guide the lightly injured group before they enter the danger zone.

[0060] Pre-set safety constraints are designed to ensure that rescue forces can reach the optimal containment point before the lightly injured group arrives, and that there is a safe buffer distance between the optimal containment point and the danger zone, while also meeting the requirement of open space to avoid trampling.

[0061] This embodiment adopts a discrete spatiotemporal inference framework based on multi-agent (Agent-based) to treat lightly injured personnel as intelligent agents with perception, decision-making and mobility capabilities. By step-by-step coupling of physical field simulation and behavioral dynamics model, a closed loop of the entire process from trajectory prediction to control generation is realized.

[0062] In summary, the multi-agent-based trajectory prediction and interception control method for minor injuries provided in this application modeles each minor injured person in the group as an independent agent, and then predicts the escape trajectory of the agents based on the combined force of explicit risk repulsion, safety exit attraction, and panic herd mentality. This simulates the irrational escape behavior of the minor injured group under limited perception and panic-driven conditions, thereby improving the accuracy of group escape trajectory prediction in complex stress environments.

[0063] like Figure 2 The diagram illustrates a flowchart of a multi-agent-based method for predicting and intercepting the trajectory of a minor injury group, as provided in an embodiment of this application. This method can be applied to electronic devices. The multi-agent-based method for predicting and intercepting the trajectory of a minor injury group may include: Step 201: Obtain disaster environmental data and group data of the slightly injured population in the target area, and divide the target area into multiple grids based on the disaster environmental data.

[0064] Step 201 is implemented in the same way as step 101, as detailed in the description above, and will not be repeated here.

[0065] Step 202: The fire spread model is used to calculate the fire source location, heat release rate and smoke generation rate in the disaster environment data to obtain the explicit risk value of each grid.

[0066] In calculating the explicit risk value (E vis When a fire spread model is used, the location of the fire source (P) is considered. fire ), heat release rate (HRR(t)) and flue gas formation rate (S smoke Using (t) as input, calculate the spatial spread of the fire. At each time step Δt, update the cell(i,j) covered by the fire, and set the explicit risk value E for the cell(i,j) directly covered by the fire. vis (i,j,t) = 1.0 (highest risk). For cell(i,j) in the smoke-covered area far from the fire source, set the corresponding explicit risk value according to the smoke concentration.

[0067] The fire spread model can be an improved model, such as a simplified version of the Fire Dynamics Simulator (FDS) or an empirical diffusion model.

[0068] Step 203: Using an obstacle-corrected Gaussian plume model, the hazardous chemical data, ambient wind speed, and ambient wind direction in the disaster environment data are calculated to obtain the toxic gas concentration in each grid. Using a cumulative heat dose triggering model, the hazardous chemical data and heat release rate in the disaster environment data are calculated to obtain the cumulative heat dose of each explosive object. When the cumulative heat dose of any explosive object reaches a critical value, a secondary explosion probability is set for the grid within a preset radius around the explosive object. The toxic gas concentration and / or secondary explosion probability are used as the implicit risk value of each grid.

[0069] The latent risk value represents the degree of risk that the grid will be affected by the spread of toxic gas and / or secondary explosions, and is used to assess the risk of secondary damage to the agent.

[0070] In calculating the concentration of toxic gas (C) gas When using hazardous chemical data (D) chem Starting from the leak source in ), combined with the ambient wind speed (U) ∞ ), environmental wind direction (D) wind The gas concentration C in each cell (i,j) at time t is calculated using a Gaussian plume model based on obstacle correction, along with obstacle information in each cell (i,j). gas (i,j,t).

[0071] In calculating the probability of a secondary explosion (P) blast When a potentially explosive target s is identified from hazardous chemical data, its current cumulative heat dose D is calculated at each time step t based on the heat release rate (HRR(t)). thermal,s The calculation formula is: When the cumulative heat dose reaches a critical value (D) thermal,s ≥Q crit,s When the explosive target s is in a critical failure state, it is determined that the explosive target s has entered a critical failure state, and an area with a preset radius (R) centered on the explosive target s is determined. blast All cells (i,j) within a given cell in the future T danger The time window is marked as a high-risk deflagration zone, i.e., a secondary explosion probability P is set. blast (i,j,t)=1.0.

[0072] Step 204: Calculate the resultant force on each agent in the lightly injured group based on disaster environment data, group data, and explicit risk values. The resultant force includes explicit risk repulsion force, safety exit attraction force, and panic herd mentality force. Predict the grid sequence that the agent will move through based on the resultant force, and use the grid sequence as the agent's escape trajectory.

[0073] Specifically, calculating the combined force on each agent in the lightly injured group based on disaster environmental data, population data, and explicit risk values ​​can include: (1) Generate a corresponding number of agents in each passable grid based on the agent distribution data in the group data, and obtain the current position and panic value of each agent from the group data.

[0074] Model each agent based on the agent distribution data, and define the agent set A = {a1, a2, ..., a...} n}. Wherein, the state vector S of each agent... m (t) includes: position vector Velocity vector Health status H m Panic level ε m respiratory enhancement coefficient α RMV,m and cumulative exposure dose Dose m .

[0075] (2) Based on the current location, disaster environment data and explicit risk value, determine the explicit risk sources within the perception range of each agent, and calculate the explicit risk repulsion force from the explicit risk sources to the agents.

[0076] agent a m By simulating its field of view using a ray projection algorithm, it can perceive the location of fire sources or dense smoke in a visible risk field.

[0077] Overt risk repulsion ( The direction of the hazard is the unit vector pointing from the nearest visible hazard source within the agent's field of vision to the agent. Size K rep ×exp(-d risk_to_m / r rep ), where d risk_to_m It is the Euclidean distance r between the agent and a visible, obvious hazard source (fire or dense smoke). rep These are model constants.

[0078] (3) Based on the current location, the location of the safety exit in the disaster environment data, and the obstacle data, calculate the safety exit attraction of the agent pointing to the nearest safety exit that is not blocked by obstacles.

[0079] Safe Exit Attractiveness The direction is from the agent towards the nearest safe exit that is not blocked by obstacles. The size is the model constant k att .

[0080] (4) Update the panic value based on the explicit risk value, and calculate the panic herd force on the agent based on the updated panic value and the overall movement direction of other agents in the agent's neighborhood.

[0081] Panic level ε m The update formula is: εm (t) = ε m (t-Δt)+ Δε stimulus -Δε decay , where Δε stimulus With explicit risk value (E vis The panic increment, Δε, is related to the density of the surrounding population and is triggered by the intensity of the overt risk field. decay This is the natural decay over time. Then, a factor based on ε is set. m The weight w of (t) social (ε m Furthermore, the higher the panic value, the greater the weight.

[0082] Panic and herd mentality ( The direction of ) is determined by agent a m Within the neighborhood (radius R) neighbor All other intelligent agents a j Overall movement direction Where j is the index of other agents in the neighborhood; the size is w social (ε m )×k soc , where k soc These are model constants.

[0083] (5) The combined force is obtained by vector superimposing the repulsive force of explicit risks, the attractiveness of safe exits, and the herd mentality of panic.

[0084] agent a m The net force acting on it at time t .

[0085] Specifically, predicting the grid sequence that the agent will traverse during its movement based on the combined force can include: (1) Calculate the acceleration of the agent based on the resultant force and the mass of the agent in the group data.

[0086] acceleration = / M m , of which M m It refers to the quality of the intelligent agent.

[0087] (2) Calculate the current velocity of the agent based on the acceleration.

[0088] Current speed = + The speed is limited to [0, V]. max Within the range.

[0089] (3) Calculate the current position of the agent based on the current speed.

[0090] Current location = + .

[0091] (4) Map the current position of different time steps to the corresponding grids respectively, and combine multiple grids into a grid sequence according to the time step order.

[0092] Based on current location Determine the cell (i,j) where the agent is located, and assign each agent's time (t,j) to cells at different time steps. , The sequence is stored as trajectory data, resulting in the dynamic escape trajectory set D of all agents within the time window. trajectory .

[0093] Step 205: Based on the escape trajectory and the hidden risk value, the escape trajectory that will cause the agent's injury to worsen is identified as a high-risk escape trajectory.

[0094] Specifically, based on the escape trajectory and the implicit risk value, escape trajectories that would worsen the agent's condition are identified as high-risk escape trajectories, which may include: (1) Calculate the breathing enhancement coefficient of the agent based on the current speed in the escape trajectory.

[0095] According to agent a m Current speed | | Calculate its respiratory enhancement coefficient α in real time RMV,m (t)=1+λ RMV ×(| | / V max ), where λ RMV It is a physiological correction constant that reflects the effect of running on respiratory volume.

[0096] (2) Calculate the cumulative dose of inhaled toxins during the movement of the agent based on the latent risk value and respiratory enhancement coefficient of the grid where the agent is located.

[0097] At each time step Δt, according to agent a m The concentration of poison gas C in the cell (i,j) is... gas (i,j,t) and respiratory enhancement coefficient α RMV,m (t), calculate the cumulative inhaled toxicant dose Dose m (t+Δt)= Dose m (t)+[C gas (i,j,t)×α RMV,m (t)]×Δt.

[0098] (3) If the cumulative dose of inhaled toxic substance exceeds the dose threshold, or if there is a grid in the escape trajectory that will fall into the area affected by the secondary explosion in the future, then the escape trajectory will be determined as a candidate escape trajectory.

[0099] If the cumulative inhaled dose of toxic substances is Dose m (t) > Dose threshold D threshold Then agent a m Health status H m G changes to R or B.

[0100] If the escape trajectory is in the future T danger P falls within the time window blast Within the grid of the high-risk combustion and explosion zone (i,j,t)=1.0, the agent a is directly determined. m Health status H m G changes to B.

[0101] Optionally, the number of people in each health state (G, Y, R, B) within the predicted time window can be statistically analyzed over time to obtain an injury evolution matrix, and a secondary injury warning report can be generated based on the injury evolution matrix. For example, it is predicted that between T+15min and T+30min, 30 lightly injured individuals will develop serious injuries due to entering the toxic gas zone.

[0102] (4) Perform spatiotemporal clustering on all candidate escape trajectories, and determine the group movement streamlines obtained after clustering as high-risk escape trajectories.

[0103] First, all agent trajectories whose health status changed during the injury evolution process (especially from G to R or B) were selected to form a high-risk trajectory set D. fatal_trajectory Then, density-based spatial clustering algorithms (such as DBSCAN) are used to cluster D. fatal_trajectory Spatiotemporal clustering was performed to identify the densest and most dangerous group movement flows, and their centerlines were defined as high-risk escape trajectories (Paths). trap .

[0104] Step 206: Backtrack along the direction of movement of the high-risk escape trajectory, select the grid that meets the preset safety constraints as the best blocking point, and intercept and guide the lightly injured group to evacuate at the best blocking point.

[0105] Specifically, by tracing back along the movement direction corresponding to the high-risk escape trajectory, the optimal blocking point can be selected from the grid that meets the preset safety constraints. This can include: (1) Determine the entry point from the high-risk escape route to the hidden risk area.

[0106] For a high-risk escape route L, determine the entry point P on this route into the hidden risk zone (toxic gas / explosion zone).danger .

[0107] (2) Starting from the entry point, perform grid backtracking in the opposite direction of the movement direction corresponding to the high-risk escape trajectory.

[0108] From the entry point P danger We began by tracing back along the high-risk escape route L, against the flow of people.

[0109] (3) The first grid that meets all the preset safety constraints is determined as the optimal blocking point. The preset safety constraints include at least one of the following: the distance between the optimal blocking point and the group with minor injuries is greater than the safe buffer distance; the optimal blocking point is located within the effective working range of the rescue force; and the openness around the optimal blocking point meets the group's stay conditions.

[0110] Preset safety constraints may include at least one of safety distance verification, accessibility verification, and spatial verification.

[0111] Safety distance verification: Finding a point P candidate This makes the distance between this point and the entry point P danger The path distance is greater than the safety buffer distance D safe Among them, the safety buffer distance D safe This allows for leeway in crowd response and deceleration, for example, it can be set to 50 meters.

[0112] Reachability check: Check P candidate Whether it is within the effective operational range of rescue forces (such as drones or ground teams), i.e., the rescue arrival time T. rescue <crowd arrival time T crowd .

[0113] Spatial verification: Check P candidate Are the grid in question and its surrounding grids sufficiently open to allow people to stay or turn around, thus preventing stampedes?

[0114] The first point that satisfies all the above conditions is the optimal blocking point P. block .

[0115] Specifically, intercepting and guiding lightly injured individuals to evacuate at the optimal point of contact can include: (1) Calculate the movement direction corresponding to the high-risk escape trajectory at the best blocking point, and the safe direction from the best blocking point to the nearest safe area.

[0116] Find the optimal blocking point P block Movement direction corresponding to high-risk escape trajectory Calculate from the optimal blocking point P block Safe direction leading to the nearest safe zone .

[0117] (2) Calculate the deflection angle when the moving direction turns to the safe direction.

[0118] Calculate from the direction of movement Turn to safe direction The deflection angle θ.

[0119] (3) Generate interception and control instructions containing the location and deflection angle of the optimal blocking point, so as to intercept and guide the evacuation of the lightly injured group through the interception and control instructions.

[0120] The interception and control command should include at least the location of the optimal blocking point (x,y) and the deflection angle θ.

[0121] In one example, the interception control instruction could be: set an interception at coordinates (x,y) and instruct the crowd to deflect θ degrees to the right and evacuate.

[0122] In summary, the multi-agent-based trajectory prediction and interception control method for minor injuries provided in this application modeles each minor injured person in the group as an independent agent, and then predicts the escape trajectory of the agents based on the combined force of explicit risk repulsion, safety exit attraction, and panic herd mentality. This simulates the irrational escape behavior of the minor injured group under limited perception and panic-driven conditions, thereby improving the accuracy of group escape trajectory prediction in complex stress environments.

[0123] By constructing two logically decoupled evolutionary perspectives—an explicit risk field and a implicit risk field—where the explicit risk field acts as the driving force determining the direction of crowd movement, and the implicit risk field acts as the injury source determining changes in the crowd's injuries, a logical separation of behavioral drivers and injury sources is achieved. Based on this, the respiratory enhancement coefficient is calculated according to the current speed in the escape trajectory, and the cumulative inhaled toxic dose is calculated by combining it with the implicit risk value. Furthermore, the system monitors whether the escape trajectory falls into the area affected by a secondary explosion, enabling a high-precision dynamic assessment of the risk of secondary injury to the lightly injured group.

[0124] Pre-set safety constraints ensure that rescue forces can reach the optimal containment point before the lightly injured group arrives, and that a safe buffer distance is maintained between the optimal containment point and the entry point, while also satisfying the requirement for open space to prevent trampling. This allows for the determination of intervention locations before the lightly injured group enters the hidden risk area. Based on this, the deflection angle between the high-risk escape trajectory's movement direction at the optimal containment point and the safe direction pointing to the nearest safe area is calculated. This generates an interception and control command that includes the optimal containment point location and the deflection angle, achieving source-level containment of risk while minimizing resource consumption and improving the rescue success rate.

[0125] like Figure 3The diagram illustrates a structural block diagram of a multi-agent-based minor injury group trajectory prediction and interception control device according to an embodiment of this application. This multi-agent-based minor injury group trajectory prediction and interception control device can be applied to electronic devices. The device includes: The data acquisition module 310 is used to acquire disaster environmental data and minor injury group data of the target area, and divide the target area into multiple grids based on the disaster environmental data; The risk calculation module 320 is used to calculate the explicit risk value and implicit risk value of each grid based on the disaster environment data. The explicit risk value represents the degree of risk of the grid being affected by fire or smoke, and the implicit risk value represents the degree of risk of the grid being affected by the spread of toxic gas and / or secondary explosion. The trajectory prediction module 330 is used to calculate the resultant force on each agent in the slightly injured group based on disaster environment data, group data and explicit risk values. The resultant force includes explicit risk repulsion force, safety exit attraction force and panic herd force. Based on the resultant force, the module predicts the grid sequence that the agent will move through when moving, and uses the grid sequence as the agent's escape trajectory. The trajectory filtering module 340 is used to identify high-risk escape trajectories that could worsen the agent's condition based on the escape trajectory and the hidden risk value. The interception and control module 350 is used to trace back along the movement direction of the high-risk escape trajectory, select the grid that meets the preset safety constraints as the best blocking point, and intercept and guide the evacuation of the lightly injured group at the best blocking point.

[0126] In an optional embodiment, the risk calculation module 320 is further configured to: A fire spread model was used to calculate the location of the fire source, the heat release rate, and the smoke generation rate in the disaster environment data, and the explicit risk value of each grid was obtained. An obstacle-corrected Gaussian plume model was used to calculate the toxic gas concentration in each grid cell based on hazardous chemical data, ambient wind speed, and ambient wind direction in the disaster environment data. A cumulative heat dose triggering model was used to calculate the cumulative heat dose of each explosive object based on the hazardous chemical data and heat release rate in the disaster environment data. When the cumulative heat dose of any explosive object reaches a critical value, a secondary explosion probability was set for the grid cells within a preset radius around the explosive object. The toxic gas concentration and / or secondary explosion probability were used as the implicit risk value for each grid cell.

[0127] In an optional embodiment, the trajectory prediction module 330 is further configured to: Based on the agent distribution data in the swarm data, generate a corresponding number of agents in each passable grid, and obtain the current position and panic value of each agent from the swarm data; Based on the current location, disaster environment data, and explicit risk values, identify the explicit risk sources within the perception range of each agent, and calculate the explicit risk repulsion force from the explicit risk sources to the agents. Based on the current location, the location of safety exits in the disaster environment data, and obstacle data, calculate the safety exit attractiveness from which the agent points to the nearest safety exit that is not blocked by obstacles. The panic value is updated based on the explicit risk value. Based on the updated panic value and the overall movement direction of other agents in the agent's neighborhood, the panic herd force experienced by the agent is calculated. The combined force of the repulsive force of explicit risks, the attractiveness of safe exits, and the herd mentality of panic creates a synergistic effect.

[0128] In an optional embodiment, the trajectory prediction module 330 is further configured to: Calculate the agent's acceleration based on the resultant force and the agent's mass in the group data; Calculate the agent's current velocity based on acceleration; Calculate the agent's current position based on the current speed; The current position at different time steps is mapped to the corresponding grid, and multiple grids are combined into a grid sequence according to the time step order.

[0129] In an optional embodiment, the trajectory filtering module 340 is further configured to: Calculate the agent's breathing enhancement coefficient based on the current speed in the escape trajectory; Based on the latent risk value and respiratory enhancement coefficient of the grid where the agent is located, calculate the cumulative dose of inhaled toxic substance during the agent's movement. If the cumulative dose of inhaled toxic substance exceeds the dose threshold, or if there is a grid in the escape trajectory that will fall into the area affected by a secondary explosion in the future, then the escape trajectory will be identified as a candidate escape trajectory. Spatiotemporal clustering is performed on all candidate escape trajectories, and the resulting group movement streamlines are identified as high-risk escape trajectories.

[0130] In an optional embodiment, the interception and control module 350 is further configured to: Determine the entry point from high-risk escape routes into areas with hidden risks; Starting from the entry point, perform grid backtracking in the reverse direction of the movement direction corresponding to the high-risk escape trajectory; The first grid that meets all the preset safety constraints is determined as the optimal blocking point. The preset safety constraints include at least one of the following: the distance between the optimal blocking point and the group of minor injuries is greater than the safe buffer distance; the optimal blocking point is located within the effective operating range of the rescue force; and the openness around the optimal blocking point meets the group's stay conditions.

[0131] In an optional embodiment, the interception and control module 350 is further configured to: Calculate the movement direction corresponding to the high-risk escape trajectory at the optimal blocking point, and the safe direction from the optimal blocking point to the nearest safe area; Calculate the deflection angle when the direction of movement turns to the safe direction; Generate interception and control instructions containing the location and deflection angle of the optimal blocking point, so as to intercept and guide the evacuation of lightly injured groups through the interception and control instructions.

[0132] In summary, the multi-agent-based trajectory prediction and interception control device for minor injuries provided in this application modeles each minor injured person in the group as an independent agent, and then predicts the escape trajectory of the agents based on the combined force of explicit risk repulsion, safety exit attraction, and panic herd mentality. This allows the device to simulate the irrational escape behavior of the minor injured group under limited perception and panic-driven conditions, thereby improving the accuracy of group escape trajectory prediction in complex stress environments.

[0133] By constructing two logically decoupled evolutionary perspectives—an explicit risk field and a implicit risk field—where the explicit risk field acts as the driving force determining the direction of crowd movement, and the implicit risk field acts as the injury source determining changes in the crowd's injuries, a logical separation of behavioral drivers and injury sources is achieved. Based on this, the respiratory enhancement coefficient is calculated according to the current speed in the escape trajectory, and the cumulative inhaled toxic dose is calculated by combining it with the implicit risk value. Furthermore, the system monitors whether the escape trajectory falls into the area affected by a secondary explosion, enabling a high-precision dynamic assessment of the risk of secondary injury to the lightly injured group.

[0134] Pre-set safety constraints ensure that rescue forces can reach the optimal containment point before the lightly injured group arrives, and that a safe buffer distance is maintained between the optimal containment point and the entry point, while also satisfying the requirement for open space to prevent trampling. This allows for the determination of intervention locations before the lightly injured group enters the hidden risk area. Based on this, the deflection angle between the high-risk escape trajectory's movement direction at the optimal containment point and the safe direction pointing to the nearest safe area is calculated. This generates an interception and control command that includes the optimal containment point location and the deflection angle, achieving source-level containment of risk while minimizing resource consumption and improving the rescue success rate.

[0135] One embodiment of this application provides a computer-readable storage medium storing at least one instruction, which is loaded and executed by a processor to implement the multi-agent-based method for predicting and intercepting the trajectory of a group of minor injuries as described above.

[0136] Figure 4A schematic block diagram of an example electronic device 400 that can be used to implement embodiments of this application is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the application described and / or claimed herein.

[0137] like Figure 4 As shown, the electronic device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 into a random access memory (RAM) 403. The RAM 403 may also store various programs and data required for the operation of the electronic device 400. The computing unit 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0138] Multiple components in electronic device 400 are connected to I / O interface 405, including: input unit 406, such as keyboard, mouse, etc.; output unit 407, such as various types of displays, speakers, etc.; storage unit 408, such as disk, optical disk, etc.; and communication unit 409, such as network card, modem, wireless transceiver, etc. Communication unit 409 allows electronic device 400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0139] The computing unit 401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as data processing methods. For example, in some embodiments, the data processing methods may be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by the computing unit 401, one or more steps of the data processing methods described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to perform data processing methods by any other suitable means (e.g., by means of firmware).

[0140] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0141] The program code used to implement the methods of this application may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0142] In the context of this application, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0143] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0144] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0145] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0146] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this application can be achieved, and this is not limited herein.

[0147] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for trajectory prediction and interception control of a group of lightly injured individuals based on multi-agent systems, characterized in that, The method includes: Acquire disaster environmental data and minor injury group data of the target area, and divide the target area into multiple grids based on the disaster environmental data; The explicit risk value and implicit risk value of each grid are calculated based on the disaster environment data. The explicit risk value represents the degree of risk of the grid being affected by fire or smoke, and the implicit risk value represents the degree of risk of the grid being affected by toxic gas diffusion and / or secondary explosion. Based on the disaster environment data, the group data, and the explicit risk value, the resultant force on each agent in the slightly injured group is calculated. The resultant force includes explicit risk repulsion force, safe exit attraction force, and panic herd force. Based on the resultant force, the grid sequence that the agent will move through is predicted, and the grid sequence is used as the agent's escape trajectory. Based on the escape trajectory and the hidden risk value, the escape trajectory that would lead to the deterioration of the agent's injury is identified as a high-risk escape trajectory. Tracing back along the direction of movement of the high-risk escape trajectory, a grid that meets the preset safety constraints is selected as the optimal blocking point. The group of lightly injured people is intercepted and guided to evacuate at the optimal blocking point.

2. The method for trajectory prediction and interception control of a group of lightly injured individuals based on multi-agent systems according to claim 1, characterized in that, The calculation of the explicit and implicit risk values ​​for each grid based on the disaster environment data includes: The fire spread model is used to calculate the fire source location, heat release rate and smoke generation rate in the disaster environment data to obtain the explicit risk value of each grid. An obstacle-corrected Gaussian plume model is used to calculate the hazardous chemical data, ambient wind speed, and ambient wind direction in the disaster environment data to obtain the toxic gas concentration in each grid. A cumulative heat dose triggering model is used to calculate the hazardous chemical data and heat release rate in the disaster environment data to obtain the cumulative heat dose of each explosive object. When the cumulative heat dose of any explosive object reaches a critical value, a secondary explosion probability is set for the grid within a preset radius around the explosive object. The toxic gas concentration and / or the secondary explosion probability are used as the implicit risk value of each grid.

3. The method for predicting and intercepting the trajectory of a group of minor injuries based on multi-agent systems according to claim 1, characterized in that, The calculation of the resultant force on each agent in the lightly injured group based on the disaster environment data, the group data, and the explicit risk value includes: Based on the agent distribution data in the group data, generate a corresponding number of agents in each passable grid, and obtain the current position and panic value of each agent from the group data; Based on the current location, the disaster environment data, and the explicit risk value, determine the explicit risk sources within the perception range of each agent, and calculate the explicit risk repulsion force from the explicit risk sources to the agent; Based on the current location, the location of safety exits in the disaster environment data, and obstacle data, calculate the safety exit attractiveness from which the agent points to the nearest unobstructed safety exit. The panic value is updated based on the explicit risk value, and the panic herd force experienced by the agent is calculated based on the updated panic value and the overall movement direction of other agents in the agent's neighborhood. The combined force is obtained by vector-superimposing the explicit risk repulsion force, the safe exit attraction force, and the panic herd mentality force.

4. The method for predicting and intercepting the trajectory of a group of minor injuries based on multi-agent systems according to claim 1, characterized in that, The step of predicting the grid sequence traversed by the agent during its movement based on the combined force includes: The acceleration of the agent is calculated based on the resultant force and the agent's mass in the group data. Calculate the agent's current velocity based on the acceleration; Calculate the current position of the agent based on the current speed; The current position at different time steps is mapped to the corresponding grid, and multiple grids are combined into a grid sequence according to the time step order.

5. The method for trajectory prediction and interception control of a group of lightly injured individuals based on multi-agent systems according to claim 1, characterized in that, The method of identifying escape trajectories that would worsen the agent's condition based on the escape trajectory and the latent risk value as high-risk escape trajectories includes: Calculate the agent's breathing enhancement coefficient based on the current speed in the escape trajectory; Based on the latent risk value of the grid where the agent is located and the breathing enhancement coefficient, the cumulative inhaled toxic dose of the agent during the movement is calculated. If the cumulative inhaled toxic dose exceeds the dose threshold, or if the escape trajectory contains a grid that will fall into the secondary explosion's influence area in the future, then the escape trajectory will be determined as a candidate escape trajectory. Spatiotemporal clustering is performed on all candidate escape trajectories, and the resulting group movement streamlines are identified as high-risk escape trajectories.

6. The method for predicting and intercepting the trajectory of a group of minor injuries based on multi-agent systems according to claim 1, characterized in that, The step of tracing back along the movement direction corresponding to the high-risk escape trajectory and selecting a grid that meets preset safety constraints as the optimal blocking point includes: Determine the entry point from the high-risk escape route into the hidden risk area; Starting from the entry point, perform grid backtracking in the opposite direction to the movement direction corresponding to the high-risk escape trajectory; The first grid that satisfies all preset safety constraints is determined as the optimal blocking point. The preset safety constraints include at least one of the following: the distance between the optimal blocking point and the group of minor injuries is greater than the safe buffer distance; the optimal blocking point is located within the effective operating range of the rescue force; and the openness around the optimal blocking point meets the group's stay conditions.

7. The method for predicting and intercepting the trajectory of a group of minor injuries based on multi-agent systems according to any one of claims 1 to 6, characterized in that, The interception and guidance of the lightly injured group at the optimal blocking point includes: Calculate the movement direction corresponding to the high-risk escape trajectory at the optimal blocking point, and the safe direction from the optimal blocking point to the nearest safe area; Calculate the deflection angle from the moving direction to the safe direction; Generate an interception and control command that includes the location of the optimal blocking point and the deflection angle, so as to intercept and guide the evacuation of the lightly injured group through the interception and control command.

8. A device for predicting and intercepting the trajectory of a group of lightly injured individuals based on multi-agent intelligence, characterized in that, The device includes: The data acquisition module is used to acquire disaster environmental data and minor injury group data of the target area, and divide the target area into multiple grids based on the disaster environmental data; The risk calculation module is used to calculate the explicit risk value and implicit risk value of each grid based on the disaster environment data. The explicit risk value represents the degree of risk of the grid being affected by fire or smoke, and the implicit risk value represents the degree of risk of the grid being affected by toxic gas diffusion and / or secondary explosion. The trajectory prediction module is used to calculate the resultant force on each agent in the slightly injured group based on the disaster environment data, the group data, and the explicit risk value. The resultant force includes explicit risk repulsion force, safety exit attraction force, and panic herd mentality force. Based on the resultant force, the module predicts the grid sequence that the agent will move through, and uses the grid sequence as the agent's escape trajectory. The trajectory filtering module is used to identify high-risk escape trajectories as those that would worsen the injury of the intelligent agent, based on the escape trajectory and the hidden risk value. The interception and control module is used to trace back along the movement direction of the high-risk escape trajectory, select a grid that meets the preset safety constraints as the best blocking point, and intercept and guide the lightly injured group to evacuate at the best blocking point.

9. A computer-readable storage medium, characterized in that, The storage medium stores at least one instruction, which is loaded and executed by a processor to implement the multi-agent-based trajectory prediction and interception control method for minor injuries as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, The electronic device includes: the multi-agent-based trajectory prediction and interception control device for minor injuries as described in claim 8.