Rail transit station three-dimensional space emergency evacuation path planning method

Abstract

The application discloses a kind of three-dimensional space emergency evacuation path planning methods of rail transit station, comprising the following steps: constructing the three-dimensional digital twin model of rail transit station;Establish three-dimensional space coordinate system, the three-dimensional space of station is discretized into a series of three-dimensional body element, establish breathing reference layer;The temperature and its change rate, carbon monoxide concentration and its change rate, visibility of each three-dimensional body element are calculated, the limit residence time of each three-dimensional body element is calculated in real time;Dynamic taboo domain is constructed, and three-dimensional body element in dynamic taboo domain is dynamically updated;Any continuous three-dimensional body element sequence path from evacuation starting point to safety exit, define the cumulative value of respiratory consumption as the path respiratory consumption value of the path;Generate multiple initial paths, based on dynamic taboo domain and cumulative value of respiratory consumption, through multiple rounds of three-dimensional body element combination optimization and reverse respiratory potential field feedback calibration, generate optimal evacuation path.The application realizes the path planning for three-dimensional space of rail transit station.

Description

Technical Field

[0001] This invention belongs to the field of emergency evacuation in rail transit, specifically relating to a three-dimensional spatial emergency evacuation path planning method for rail transit stations. Background Technology

[0002] With the continuous expansion of my country's urban rail transit network, the fire safety risks of stations, as densely populated underground enclosed spaces, are becoming increasingly prominent. When a fire occurs in a rail transit station, the fire environment exhibits dynamic and rapid evolution characteristics, such as smoke diffusion, temperature rise, and accumulation of toxic gases, which can easily lead to casualties. In such extreme environments, scientific and efficient emergency evacuation route planning is a core technical means to ensure personnel safety.

[0003] Currently, the main technical limitations of evacuation route planning methods for rail transit stations are as follows: Firstly, the three-dimensional spatial adaptability is insufficient, making it difficult to support three-dimensional evacuation decisions. Existing methods often simplify station space into a two-dimensional planar network model, using the concourse or platform floor plan as a reference for path searching. However, modern rail transit stations often contain complex three-dimensional structures such as multi-level transfers, grade-separated intersections, stairs, and escalators. The simplified two-dimensional model cannot accurately represent the vertical channel connections and the impact of stair slopes on pedestrian movement, leading to potentially infeasible or suboptimal planned paths in three-dimensional space.

[0004] Secondly, the response to the dynamic evolution of the fire environment is lagging, resulting in poor timeliness of planning results. Existing methods mostly employ static path planning or offline pre-calculation based on historical fire simulation data, failing to dynamically adjust evacuation plans according to real-time sensor data at the fire scene. However, in actual fire scenarios, temperature fields, smoke concentration fields, and visibility all change drastically on a timescale of seconds. Static planning results often become invalid shortly after generation due to environmental changes, making it difficult to provide continuous and effective guidance for evacuees.

[0005] Thirdly, the evaluation indicators for evacuation routes are too simplistic, prioritizing speed over safety. The core objective of conventional route planning is speed, optimizing for the shortest path length or travel time. However, in fire scenarios, safe and rapid evacuation is the fundamental requirement. Existing methods often only consider distance or time factors when planning routes, failing to quantify and incorporate the direct hazards of the fire environment to the human body, such as burns, CO poisoning, and smoke asphyxiation, into the route evaluation system. This easily leads to situations where routes are fast but traverse high-risk areas, resulting in poor synergy between safety and efficiency.

[0006] Fourth, the utilization rate of sensing equipment data is low, and it is disconnected from route planning. While rail transit stations commonly deploy sensing equipment such as temperature sensors and gas monitoring sensors, the real-time data collected by these devices is mostly used for fire alarms or equipment linkage control, and has not been effectively integrated into the dynamic decision-making process of evacuation route planning. Data silos exist between sensor data and route planning algorithms, preventing the planning system from fully utilizing real-time environmental information to accurately correct routes, resulting in deviations between the planned results and the actual fire scene environment.

[0007] In summary, existing emergency evacuation route planning methods for rail transit stations are no longer able to meet the actual evacuation needs under the dynamic evolution of fires. Therefore, it is urgent to design an emergency evacuation route planning method for rail transit stations that can adapt to three-dimensional space, respond to the dynamic evolution of fires, take into account both safety and efficiency objectives, and deeply integrate perception data. Summary of the Invention

[0008] This invention is proposed to solve the problems existing in the prior art, and its purpose is to provide a three-dimensional spatial emergency evacuation route planning method for rail transit stations.

[0009] The technical solution of this invention is: a three-dimensional spatial emergency evacuation route planning method for rail transit stations, comprising the following steps: A. Construct a 1:1 scale 3D digital twin model of the rail transit station. The 3D digital twin model accurately reproduces the 3D structure of the station hall, platform, stairs, passages, and internal facilities and equipment. B. Establish a three-dimensional spatial coordinate system for the three-dimensional digital twin model. x , y , z The station's three-dimensional space is discretized into a series of three-dimensional voxels. The dimensions of each voxel are dynamically adapted according to the spatial geometric features, and the spatial coordinates and volume of each voxel are recorded. V Adjacent volume element set N Local slope angle i And establish a respiratory reference layer; C. Deploy temperature sensors, carbon monoxide sensors, and visibility sensors within the station. Calculate the temperature of each three-dimensional volume element based on the dynamic data collected by the sensors. T ( t ) and its rate of change T ′( t ), carbon monoxide concentration C ( t ) and its rate of change C ′( t ),visibility S ( t The system also calculates the limiting dwell time of each three-dimensional volume element in real time. t ( t),in t For time; D. Constructing dynamic taboo domains F ( t Adaptive periodic dynamic update is adopted. F ( t The three-dimensional element within the fire protection facility will immediately trigger a recalculation of the forbidden domain when the status of the fire protection facility changes. E. Any continuous three-dimensional voxel sequence path from the evacuation start point (Start) to the safety exit (Exit), where all continuous voxel sequences are: P =( v 0 = Start, v 1, v 2,…, v n =Exit), where v 0、 v 1, ... v n Represents each voxel along the path, where n is the number of voxels traversed by the path, and defines the cumulative respiratory consumption value. R ( P , t This represents the path respiratory consumption value and cumulative respiratory consumption value for this path. R ( P , t As the core evaluation indicator for path optimization, it comprehensively represents path safety and evacuation efficiency. F. Based on dynamic taboo domain F ( t ) and cumulative respiratory expenditure R ( P , t Starting from the evacuation point Start, the system generates initial paths in the path set. Through multiple rounds of three-dimensional voxel combination optimization and reverse breathing potential field feedback calibration, it generates the optimal evacuation path from the evacuation point Start to the safety exit Exit. The optimal evacuation path is then dynamically updated based on dynamic data collected by sensors.

[0010] Furthermore, in step B, the dimensions of the 3D volumetric element are dynamically adapted based on spatial geometric features. The specific process is as follows: First, large-scale elements are used in open areas such as platforms and concourses to ensure computational efficiency; Then, small voxels are used in narrow passages, staircases, and escalator entrances to ensure precise calculations of bottleneck areas; Furthermore, columns, turnstiles, and wall obstacles in the station's three-dimensional space are considered as impenetrable boundaries, and their occupied three-dimensional space is not discretized into a three-dimensional volumetric mesh; Finally, it will be above the ground. H min ~H max The three-dimensional volume elements within the range serve as a respiratory reference layer. H min , H max These represent the minimum and maximum values ​​of the vertical height range of the mouth and nose of the crowd, respectively.

[0011] Furthermore, step C calculates the temperature of each three-dimensional volume element. T ( t ) and its rate of change T ′( t ), carbon monoxide concentration C ( t ) and its rate of change C ′( t ),visibility S ( t The specific process is as follows: First, each sensor was calibrated on-site, and its three-dimensional coordinates of physical installation, floor, and fire compartment were recorded. Then, calculate the three-dimensional coordinates of each vertex and geometric center node, the floor to which it belongs, the fire compartment, and the positional relationship with the surrounding impenetrable boundaries for each three-dimensional volume element in the station; Next, the raw data collected by the sensors is first preprocessed in conjunction with the station operation system for the rail transit station scenario, and then outlier removal and time synchronization are completed. Next, the nearest neighbor spatial matching method for rail transit station scenarios is used to match sensors and 3D voxels. Finally, based on the real-time data collected by various sensors within the station, the temperature of each volumetric element within the station was monitored. T ( t ), carbon monoxide concentration C ( t and visibility S ( t The data is dynamically updated, and the rate of temperature change of each volume element is calculated. T ′( t )=d T / d t rate of change in carbon monoxide concentration C ′( t )=d C / d t , where d T d represents the change in temperature. C d represents the change in carbon monoxide concentration. t For time step.

[0012] Furthermore, step C calculates the limiting dwell time of each three-dimensional volume element in real time. t (t This includes calculating the basic limit dwell time. t 0( t The specific process is as follows: First, calculate the rate of degradation of the basic environment. E ( t This method converts multi-dimensional independent fire factors within the enclosed space of a rail transit station into an equivalent air volume consumption rate with a unified dimension. These multi-dimensional independent fire factors include the carbon monoxide concentration change rate. C ′( t ), temperature change rate T ′( t ),visibility S ( t ), rate of basic environmental deterioration E ( t The formula for calculating ) is: ; in, α The conversion coefficient for the rate of change of carbon monoxide concentration was determined through toxicological experiments. β The conversion coefficient for the rate of temperature change was determined through thermophysiological experiments. c The visibility-affected conversion coefficient was determined through behavioral psychology experiments. Then, calculate the carbon monoxide toxicity correction factor. G c ( t This reflects the direct impact of carbon monoxide on the oxygen-carrying capacity of human blood. The calculation formula is: ; in, l The toxicity attenuation index was determined by the carboxyhemoglobin (COHb) assay. Finally, calculate the basic limiting residence time. t 0( t Based on the volumetric effective air volume, the basic environmental degradation rate, and the carbon monoxide toxicity correction coefficient, the basic safe residence time is quantified, and the calculation formula is as follows: ; in, V a The total usable effective air volume within the volume element is the sum of its parts and the volume of the volume element. V The values ​​are the same.

[0013] Furthermore, step C calculates the limiting dwell time of each three-dimensional volume element in real time. t ( t (including the basic limit stay time) t 0( t The correction process is as follows: First, calculate the vertical height correction factor for rail transit stations. W H Considering the characteristic of rail transit stations being deeply buried underground, the reduction effect of vertical distance on personnel tolerance is increased. The farther away from the exit, the shorter the equivalent safe dwell time. The calculation formula is as follows: ; in, k The vertical height influence coefficient, calibrated through human metabolic experiments, reflects the impact of the vertical distance between a person and the ground exit on psychological panic and physical exertion. H z This represents the vertical height difference between the center of the volume element and the nearest ground exit; for ground stations or above-ground sections, this value is 0. Then, the influence correction factor of the rail transit station ventilation system is introduced. W e This reflects the impact of the station's smoke exhaust / makeup air system status on smoke diffusion velocity and visibility recovery, and is linked to the real-time operating condition calibration of the station's smoke exhaust / makeup air system. The volume element is within the effective makeup air zone. W e >1, when in the area of ​​smoke exhaust failure and smoke accumulation 0< W e <1, When the environmental control system is not started W e =1; Next, calculate the correction for crowd density at rail transit stations. To address the high population density characteristic of rail transit stations, a crowd density correction factor is introduced to reflect the additional impact of crowd congestion on air consumption and mobility in the calculation. The higher the density, the shorter the equivalent safe dwell time. The calculation formula is as follows: ; in, m The population density influence coefficient was determined through population behavior experiments. r (t) represents the real-time crowd density within the volume element, obtained through station video analysis; Next, the structural constraint correction for rail transit stations is calculated. W B Considering the structural characteristics of staircases and passageways in rail transit stations, a structural constraint factor is introduced to reflect the amplification effect of personnel congestion on evacuation time in staircases and passageways. The calculation formula is as follows: ; in, x Structural constraint factors, calibrated using social force models or historical data, are used in non-congested areas. x =0, congested areas are set to 0< x <1; Finally, the limiting dwell time of the volume element was determined. t ( t The calculation formula is: .

[0014] Furthermore, step D involves constructing a dynamic taboo domain. F ( t The specific process is as follows: First, the core hazardous elements are identified, and differentiated thresholds are calibrated for rail transit station scenarios. Then, the taboo elements of the rail transit station scene are progressively expanded; Next, determine the natural impassable and unsuitable evacuation conditions for the body elements; Finally, all risk elements across the entire station are uniformly incorporated into the dynamic taboo domain. F ( t Risk elements include all core hazardous elements, all forbidden elements, and elements that are naturally impassable or lack evacuation conditions.

[0015] Furthermore, the cumulative respiratory expenditure in step E R ( P , t The calculation process is as follows: First, the basic travel time of the path volume sequence is calculated, targeting... P =( v 0 = Start, v 1, v 2,…, v n Each three-dimensional volume element on =Exit) v i ( i =0, 1, 2, ..., n), and combined with the pedestrian traffic characteristics of different functional areas of the rail transit station, calculate the basic passage time of pedestrians through this three-dimensional volume element. The calculation formula is: ; in, L i For volume elements along the direction of personnel movement v i Feature length; The average walking speed of people; Then, for the scenario of rail transit stations, the voxel access adaptation coefficient is calculated for each voxel on the path. v i Spatial structure and regional attributes are used to calculate the volume element accessibility coefficient. The formula for quantifying the increase in respiratory consumption under non-ideal travel conditions is as follows: ; in, This is the vertical slope correction factor, based on the local slope angle of the volume element. i Calibration, horizontal area i =0° =1.0, vertical area i >0° and the person is moving upwards =1.0+ q 1·sin i Vertical area i >0° and the person is in the downward direction =1.0+ q 2·sin i , q 1. q 2 represents the passability coefficient calibrated in human metabolic experiments; The correction factor for evacuation bottlenecks is used when the volume element is located in evacuation bottleneck areas such as turnstiles, stairwells, and safety exits. >1.0 indicates a narrower passage width and higher population density in open areas. =1.0; For cross-zone correction coefficients, when the volume element is a transitional volume element connecting fire compartments and floors. >1.0, conforming to the door restrictions for cross-zone and cross-floor passage in rail transit stations, and the increased respiratory consumption caused by the collision of passenger flow, non-transitional body element. =1.0; Next, the instantaneous respiratory consumption value of a single body element is calculated for each body element along the path. v i Based on the voxel travel adaptation coefficient, the instantaneous respiratory consumption of a person passing through that voxel is calculated. R i To achieve the coupling and quantification of security and efficiency, the calculation formula is as follows: ; in, For personnel to reach the body element v i The moment; for Time-based element v i The maximum dwell time; Finally, the cumulative respiratory consumption along the entire pathway is calculated, including the instantaneous respiratory consumption of all body elements along the pathway. R i The values ​​are summed to obtain the cumulative respiratory consumption for the entire path. R ( P , t The calculation formula is: .

[0016] Furthermore, in step F, starting from the evacuation point Start, initial paths are generated in the path set. The specific process is as follows: First, taking the evacuation starting point Start as the path starting point, the set of three-dimensional voxels adjacent to the evacuation starting point Start is... N Not in the dynamic taboo domain F ( t The three-dimensional volume elements within the range are used as initial candidate volume elements; Then, the several paths from the evacuation starting point Start to the initial candidate voxels are taken as the initial paths in the path set.

[0017] Furthermore, step F involves multiple rounds of optimization of the three-dimensional voxel combination, the specific process of which is as follows: First, for each path in the current path set, for the terminal element of a particular path (i.e., the side furthest from the evacuation start point), select its adjacent element set. N Not in the dynamic taboo domain F ( t The three-dimensional volume element set within ) oh j}, j ≥1, construct from the original path end voxel to oh j The new extension path; Then, calculate each one in turn. oh j Cumulative respiratory consumption corresponding to the new pathway R ( P , t (), retaining compliance with station spatial connectivity rules, not crossing non-connected elements and R ( P , t The smallest m ( m ≥1) paths are the optimal expansion paths; Next, the optimal expansion path for each path in the path set is calculated sequentially, and the path set is updated accordingly. Then, repeat the above steps to perform multiple rounds of path end voxel expansion, and after each round of path end voxel expansion, initiate reverse respiratory potential field feedback calibration to guide the path towards the safe exit direction for optimization and avoid getting trapped in local optima. Finally, after multiple rounds of path-end voxel expansion and reverse respiratory potential field feedback calibration, all paths in the path set achieved voxel connectivity from the evacuation start point (Start) to the safety exit (Exit). The cumulative respiratory consumption value was then selected. R ( P , tThe path with the shortest distance is taken as the optimal evacuation path.

[0018] The beneficial effects of this invention are as follows: This invention addresses the core challenge of traditional two-dimensional planar planning methods being unable to adapt to the vertical connectivity and multi-obstacle three-dimensional structures of rail transit station concourses and platforms. By constructing a 1:1 full-element three-dimensional digital twin model of the station, and employing a dynamic adaptable voxel partitioning strategy of large voxels in open areas and small voxels in bottleneck areas, it balances computational efficiency with the accuracy of depicting key evacuation areas. By defining a breathing reference layer that conforms to the breathing characteristics of personnel, the paths planned based on the three-dimensional voxel model can accurately guide the continuous evacuation of personnel in both horizontal and vertical spaces, fully adapting to the unique evacuation scenarios of the multi-level underground three-dimensional structures of rail transit stations.

[0019] This invention solves the problems of existing technologies failing to eliminate interference specific to rail transit station scenarios and the distortion of the matching between sensing data and evacuation space. By linking the station's train operation, environmental control, and AFC passenger flow systems, it performs preprocessing on the raw sensor data specifically for rail transit station scenarios, including piston wind interference removal, environmental control condition filtering, vertical layer calibration, and large passenger flow interference identification, ensuring the authenticity of the basic data from the source. It adopts the nearest neighbor spatial matching method for rail transit station scenarios to achieve precise binding between sensors and voxels. Through Kriging interpolation considering spatial correlation, it achieves full-domain completion of the environmental parameters of accessible voxels throughout the station under sparse sensor deployment. Combined with an adaptive second-level dynamic update mechanism, it ensures that path planning is always synchronized with the real-time status of the fire scene, meeting the high timeliness requirements of emergency evacuation in rail transit stations.

[0020] This invention addresses the industry pain points of existing technologies, such as poor adaptability of general evaluation indicators to rail transit station evacuation scenarios, imbalance between safety and efficiency goals, and susceptibility to local optima. It proposes a volume element-limited dwell time. t ( t This study considers the comprehensive hazards of fire environmental parameters such as temperature, carbon monoxide concentration, and visibility to personnel evacuation, and introduces four correction coefficients specifically for rail transit station scenarios. This enables precise quantification of the safety tolerance limit of personnel in rail transit station fire environments. It also defines a cumulative breathing consumption value that incorporates corrections for slope, bottlenecks, and cross-zone conditions. R ( P , t This transforms the dual objectives of safety and efficiency into a single, quantifiable optimization indicator; and constructs a dynamic taboo domain adapted to the fire protection specifications and smoke diffusion patterns of rail transit stations. F(t) achieves advance avoidance of dangerous areas through directional progressive expansion; a two-way optimization mechanism of positive volume element combination optimization and reverse breathing potential field global calibration is designed, combined with path deviation pruning strategy, which not only retains local optimal candidate paths, but also guides the avoidance of local optima through global situation field, and truly achieves the synergistic optimization of evacuation safety and traffic efficiency in the fire scenario of rail transit station.

[0021] This invention fully integrates real-time sensor data, station environmental control system operating conditions, train signals, passenger flow monitoring data, and fire protection facility status into the entire process of extreme dwell time calculation, dynamic taboo domain update, potential field calibration, and path planning. The sensor data is no longer isolated information but becomes the core driving source of dynamic path planning. At the same time, the solution can be seamlessly embedded into the existing smart emergency command platform of rail transit stations, and is compatible with existing sensing, environmental control, and fire protection systems. It can be implemented without large-scale modifications, forming a complete technology chain of real-time perception, dynamic modeling, accurate planning, and closed-loop updating. The planning results are highly consistent with the actual evacuation scenarios of rail transit stations in fires. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the process of the present invention; Figure 2 This is a flowchart of the multi-round volume element combination optimization and reverse respiratory potential field feedback calibration to generate the optimal evacuation path in this invention. Detailed Implementation

[0023] The present invention will now be described in detail with reference to the accompanying drawings and embodiments: like Figure 1 to Figure 2 As shown, a three-dimensional spatial emergency evacuation route planning method for rail transit stations includes the following steps: A. Construct a 1:1 scale 3D digital twin model of the rail transit station. The 3D digital twin model accurately reproduces the 3D structure of the station hall, platform, stairs, passages, and internal facilities and equipment. B. Establish a three-dimensional spatial coordinate system for the three-dimensional digital twin model. x , y , z The station's three-dimensional space is discretized into a series of three-dimensional voxels. The dimensions of each voxel are dynamically adapted according to the spatial geometric features, and the spatial coordinates and volume of each voxel are recorded. V Adjacent volume element set N Local slope angle i And establish a respiratory reference layer; C. Deploy temperature sensors, carbon monoxide sensors, and visibility sensors within the station. Calculate the temperature of each three-dimensional volume element based on the dynamic data collected by the sensors. T ( t ) and its rate of changeT ′( t ), carbon monoxide concentration C ( t ) and its rate of change C ′( t ),visibility S ( t The system also calculates the limiting dwell time of each three-dimensional volume element in real time. t ( t ),in t For time; D. Constructing dynamic taboo domains F ( t Adaptive periodic dynamic update is adopted. F ( t The three-dimensional element within the fire protection facility will immediately trigger a recalculation of the forbidden domain when the status of the fire protection facility changes. E. Any continuous three-dimensional voxel sequence path from the evacuation start point (Start) to the safety exit (Exit), where all continuous voxel sequences are: P =( v 0 = Start, v 1, v 2,…, v n =Exit), where v 0、 v 1, ... v n Represents each voxel along the path, where n is the number of voxels traversed by the path, and defines the cumulative respiratory consumption value. R ( P , t This represents the path respiratory consumption value and cumulative respiratory consumption value for this path. R ( P , t As the core evaluation indicator for path optimization, it comprehensively represents path safety and evacuation efficiency. F. Based on dynamic taboo domain F ( t ) and cumulative respiratory expenditure R ( P , t Starting from the evacuation point Start, the system generates initial paths in the path set. Through multiple rounds of three-dimensional voxel combination optimization and reverse breathing potential field feedback calibration, it generates the optimal evacuation path from the evacuation point Start to the safety exit Exit. The optimal evacuation path is then dynamically updated based on dynamic data collected by sensors.

[0024] In step B, the dimensions of the 3D volumetric element are dynamically adapted based on spatial geometric features. The specific process is as follows: First, large-scale elements are used in open areas such as platforms and concourses to ensure computational efficiency; Then, small voxels are used in narrow passages, staircases, and escalator entrances to ensure precise calculations of bottleneck areas; Furthermore, columns, turnstiles, and wall obstacles in the station's three-dimensional space are considered as impenetrable boundaries, and their occupied three-dimensional space is not discretized into a three-dimensional volumetric mesh; Finally, it will be above the ground. H min ~ H max The three-dimensional volume elements within the range serve as a respiratory reference layer. H min , H max These represent the minimum and maximum values ​​of the vertical height range of the mouth and nose of the crowd, respectively.

[0025] Step C involves calculating the temperature of each three-dimensional volume element. T ( t ) and its rate of change T ′( t ), carbon monoxide concentration C ( t ) and its rate of change C ′( t ),visibility S ( t The specific process is as follows: First, each sensor was calibrated on-site, and its three-dimensional coordinates of physical installation, floor, and fire compartment were recorded. Then, calculate the three-dimensional coordinates of each vertex and geometric center node, the floor to which it belongs, the fire compartment, and the positional relationship with the surrounding impenetrable boundaries for each three-dimensional volume element in the station; Next, the raw data collected by the sensors is first preprocessed in conjunction with the station operation system for the rail transit station scenario, and then outlier removal and time synchronization are completed. Next, the nearest neighbor spatial matching method for rail transit station scenarios is used to match sensors and 3D voxels. Finally, based on the real-time data collected by various sensors within the station, the temperature of each volumetric element within the station was monitored. T ( t ), carbon monoxide concentration C ( t and visibility S ( t The data is dynamically updated, and the rate of temperature change of each volume element is calculated. T ′( t )=d T / d t rate of change in carbon monoxide concentration C ′(t )=d C / d t , where d T d represents the change in temperature. C This represents the change in carbon monoxide concentration, expressed in ppm and d. t For time step.

[0026] In step C, the ultimate dwell time of each three-dimensional voxel is calculated in real time. t ( t This includes calculating the basic limit dwell time. t 0( t The specific process is as follows: First, calculate the rate of degradation of the basic environment. E ( t This method converts multi-dimensional independent fire factors within the enclosed space of a rail transit station into an equivalent air volume consumption rate with a unified dimension. These multi-dimensional independent fire factors include the carbon monoxide concentration change rate. C ′( t ), temperature change rate T ′( t ),visibility S ( t ), rate of basic environmental deterioration E ( t The formula for calculating ) is: ; in, α The conversion coefficient for the rate of change of carbon monoxide concentration was determined through toxicological experiments. β The conversion coefficient for the rate of temperature change was determined through thermophysiological experiments. c The visibility-affected conversion coefficient was determined through behavioral psychology experiments. Then, calculate the carbon monoxide toxicity correction factor. G c ( t This reflects the direct impact of carbon monoxide on the oxygen-carrying capacity of human blood. The calculation formula is: ; in, l The toxicity attenuation index was determined by the carboxyhemoglobin (COHb) assay. Finally, calculate the basic limiting residence time. t 0( t Based on the volumetric effective air volume, the basic environmental degradation rate, and the carbon monoxide toxicity correction coefficient, the basic safe residence time is quantified, and the calculation formula is as follows: ; in, Va The total usable effective air volume within the volume element is the sum of its parts and the volume of the volume element. V The values ​​are the same.

[0027] In step C, the ultimate dwell time of each three-dimensional voxel is calculated in real time. t ( t (including the basic limit stay time) t 0( t The correction process is as follows: First, calculate the vertical height correction factor for rail transit stations. W H Considering the characteristic of rail transit stations being deeply buried underground, the reduction effect of vertical distance on personnel tolerance is increased. The farther away from the exit, the shorter the equivalent safe dwell time. The calculation formula is as follows: ; in, k The vertical height influence coefficient, calibrated through human metabolic experiments, reflects the impact of the vertical distance between a person and the ground exit on psychological panic and physical exertion. H z This represents the vertical height difference between the center of the volume element and the nearest ground exit; for ground stations or above-ground sections, this value is 0. Then, the influence correction factor of the rail transit station ventilation system is introduced. W e This reflects the impact of the station's smoke exhaust / makeup air system status on smoke diffusion velocity and visibility recovery, and is linked to the real-time operating condition calibration of the station's smoke exhaust / makeup air system. The volume element is within the effective makeup air zone. W e >1, when in the area of ​​smoke exhaust failure and smoke accumulation 0< W e <1, When the environmental control system is not started W e =1; Next, calculate the correction for crowd density at rail transit stations. To address the high population density characteristic of rail transit stations, a crowd density correction factor is introduced to reflect the additional impact of crowd congestion on air consumption and mobility in the calculation. The higher the density, the shorter the equivalent safe dwell time. The calculation formula is as follows: ; in, m The population density influence coefficient was determined through population behavior experiments. r (t) represents the real-time crowd density within the volume element, obtained through station video analysis; Next, the structural constraint correction for rail transit stations is calculated. W BConsidering the structural characteristics of staircases and passageways in rail transit stations, a structural constraint factor is introduced to reflect the amplification effect of personnel congestion on evacuation time in staircases and passageways. The calculation formula is as follows: ; in, x Structural constraint factors, calibrated using social force models or historical data, are used in non-congested areas. x =0, congested areas are set to 0< x <1; Finally, the limiting dwell time of the volume element was determined. t ( t The calculation formula is: .

[0028] In step D, construct the dynamic taboo domain. F ( t The specific process is as follows: First, the core hazardous elements are identified, and differentiated thresholds are calibrated for rail transit station scenarios. Then, the taboo elements of the rail transit station scene are progressively expanded; Next, determine the natural impassable and unsuitable evacuation conditions for the body elements; Finally, all risk elements across the entire station are uniformly incorporated into the dynamic taboo domain. F ( t Risk elements include all core hazardous elements, all forbidden elements, and elements that are naturally impassable or lack evacuation conditions.

[0029] Cumulative respiratory consumption in step E R ( P , t The calculation process is as follows: First, the basic travel time of the path volume sequence is calculated, targeting... P =( v 0 = Start, v 1, v 2,…, v n Each three-dimensional volume element on =Exit) v i ( i =0, 1, 2, ..., n), and combined with the pedestrian traffic characteristics of different functional areas of the rail transit station, calculate the basic passage time of pedestrians through this three-dimensional volume element. The calculation formula is: ; in, L i For volume elements along the direction of personnel movementv i Feature length; The average walking speed of people; Then, for the scenario of rail transit stations, the voxel access adaptation coefficient is calculated for each voxel on the path. v i Spatial structure and regional attributes are used to calculate the volume element accessibility coefficient. The formula for quantifying the increase in respiratory consumption under non-ideal travel conditions is as follows: ; in, This is the vertical slope correction factor, based on the local slope angle of the volume element. i Calibration, horizontal area i =0° =1.0, vertical area i >0° and the person is moving upwards =1.0+ q 1·sin i Vertical area i >0° and the person is in the downward direction =1.0+ q 2·sin i , q 1. q 2 represents the passability coefficient calibrated in human metabolic experiments; The correction factor for evacuation bottlenecks is used when the volume element is located in evacuation bottleneck areas such as turnstiles, stairwells, and safety exits. >1.0 indicates a narrower passage width and higher population density in open areas. =1.0; For cross-zone correction coefficients, when the volume element is a transitional volume element connecting fire compartments and floors. >1.0, conforming to the door restrictions for cross-zone and cross-floor passage in rail transit stations, and the increased respiratory consumption caused by the collision of passenger flow, non-transitional body element. =1.0; Next, the instantaneous respiratory consumption value of a single body element is calculated for each body element along the path. v i Based on the voxel travel adaptation coefficient, the instantaneous respiratory consumption of a person passing through that voxel is calculated. R i To achieve the coupling and quantification of security and efficiency, the calculation formula is as follows: ; in, For personnel to reach the body element v i The moment; for Time-based element v i The maximum dwell time; Finally, the cumulative respiratory consumption along the entire pathway is calculated, including the instantaneous respiratory consumption of all body elements along the pathway. R i The values ​​are summed to obtain the cumulative respiratory consumption for the entire path. R ( P , t The calculation formula is: .

[0030] In step F, the evacuation starting point Start is used as the path starting point to generate initial paths in the path set. The specific process is as follows: First, taking the evacuation starting point Start as the path starting point, the set of three-dimensional voxels adjacent to the evacuation starting point Start is... N Not in the dynamic taboo domain F ( t The three-dimensional volume elements within the range are used as initial candidate volume elements; Then, the several paths from the evacuation starting point Start to the initial candidate voxels are taken as the initial paths in the path set.

[0031] Step F involves multiple rounds of 3D voxel combination optimization, and the specific process is as follows: First, for each path in the current path set, for the terminal element of a particular path (i.e., the side furthest from the evacuation start point), select its adjacent element set. N Not in the dynamic taboo domain F ( t The three-dimensional volume element set within ) oh j}, j ≥1, construct from the original path end voxel to oh j The new extension path; Then, calculate each one in turn. oh j Cumulative respiratory consumption corresponding to the new pathway R ( P , t (), retaining compliance with station spatial connectivity rules, not crossing non-connected elements and R ( P , t The smallest m ( m ≥1) paths are the optimal expansion paths; Next, the optimal expansion path for each path in the path set is calculated sequentially, and the path set is updated accordingly. Then, repeat the above steps to perform multiple rounds of path end voxel expansion, and after each round of path end voxel expansion, initiate reverse respiratory potential field feedback calibration to guide the path towards the safe exit direction for optimization and avoid getting trapped in local optima. Finally, after multiple rounds of path-end voxel expansion and reverse respiratory potential field feedback calibration, all paths in the path set achieved voxel connectivity from the evacuation start point (Start) to the safety exit (Exit). The cumulative respiratory consumption value was then selected. R ( P , t The path with the shortest distance is taken as the optimal evacuation path.

[0032] Specifically, in step D, the core hazard elements are identified, and differentiated thresholds for rail transit station scenarios are calibrated, as follows: Establish the limit of the dwell time of the volume element t ( t A tiered safety threshold system is established, which differentiates threshold values ​​based on the varying evacuation importance of different functional areas within a rail transit station. First, a first safety threshold is set for evacuation bottleneck areas such as stairwells, safety exits, turnstile passages, and platform screen door end doors. t min1 For open public areas such as platforms and station halls, a second safety threshold is set. t min2 ,in t min1 > t min2 ; Real-time traversal of all passable elements in the entire station, when the element's maximum dwell time... t ( t When the value is less than or equal to the safety threshold of the corresponding area, the element is marked as a core hazardous element. Then, within the body element C ( t )≥ C max (The human tolerance threshold for carbon monoxide concentration, experimentally calibrated recommended value 800 ppm) or T ( t )≥ T max When the temperature reaches the human tolerance threshold (experimentally calibrated recommended value of 45℃), it is directly marked as an immediately lethal element and included in the core hazardous element range.

[0033] Specifically, step D involves a progressive expansion of the taboo elements in the rail transit station scenario, as follows: Based on the smoke diffusion patterns in enclosed spaces of rail transit stations and the fire prevention and smoke control zoning management rules of national rail transit station fire protection standards, the core hazardous elements are expanded in a directional and progressive manner using prohibited elements. The specific rules are as follows: Firstly, based on the expansion of the smoke control zone, adjacent elements within the same smoke control zone that are three-dimensionally connected to the core hazardous element are marked as prohibited elements. Secondly, the boundaries of fire compartments are forcibly locked, and all cross-compartment passage elements that connect to the fire compartment where the core hazardous element is located are marked as prohibited elements, in accordance with the fire protection requirements of closing fire doors and prohibiting cross-compartment passages in the event of a fire in a rail transit station. Third, vertical spread prediction and expansion: For the staircase and escalator area that connects the station hall and platform vertically, if the core hazardous element is located at the downward end of the staircase platform, two additional continuous elements are added along the upward direction of the staircase and marked as forbidden elements to adapt to the characteristics of vertical upward spread of smoke. Fourth, the piston wind influences directional expansion. If the core hazardous element is located at the end of the platform, all passable elements along the longitudinal direction of the platform corresponding to where the train stops will be marked as prohibited elements, adapting to the unique pattern of the train piston wind driving the smoke to spread longitudinally along the platform.

[0034] Specifically, step D determines the naturally impassable and non-evacuation-conditional body elements, as follows: all body elements within the non-breathing reference layer, body elements in non-evacuation public areas such as the track area and equipment management rooms, undivided body elements corresponding to impenetrable boundaries such as columns, turnstiles, and walls, and evacuation bottleneck body elements whose personnel congestion rate exceeds the threshold as determined by the AFC passenger flow system.

[0035] Specifically, the reverse respiratory potential field feedback calibration in step F is performed as follows: f1. Perform respiratory potential field baseline initialization. Taking the volume element where the safety exit (Exit) is located as the reference point, the respiratory potential field strength of a certain volume element F This is the baseline value for respiratory consumption from the body element to the safety exit (Exit). F The smaller the volume element, the closer it is to the exit, and the lower the passage cost. The volume element containing the safe exit (Exit) is... F =0, meaning there is no additional consumption upon reaching the exit.

[0036] f2. Define the potential field constraint range during iteration. Based on the constructed dynamic taboo domain F ( t Fire prevention and smoke control zoning in rail transit station fire protection specifications, station floor structure, and determination of the effective range and diffusion constraint rules for potential field iteration: Firstly, it applies only to non-dynamic taboo domains. F ( t Potential field calculation is performed on the passable volume elements within the (). F ( t Internal elements do not participate in the iteration; Secondly, the diffusion of the potential field must strictly follow the hierarchical order of the same smoke control zone - the same fire control zone - the same floor, and unrestrained diffusion across fire control zones and floors is prohibited; Third, for impenetrable boundaries such as platform screen doors, track area boundaries, and equipment room walls, potential field diffusion blocking rules should be set to prohibit the transmission of potential fields across boundaries.

[0037] f3. Calculate the basic respiratory passage cost of adjacent voxels. For volume elements whose potential field assignment has not yet been completed v Its adjacent set of volume elements N v The set of already assigned values ​​in the middle is { u k}, k ≥1, Calculate the basic breathing passage cost between two voxels, taking into account the passage characteristics of rail transit station areas. Cost uk,v The calculation formula is: ; in, For personnel through body element u k Passage time; This is the start time of the potential field calibration. Lower body element u k The maximum dwell time; For body element u k The passability adaptation coefficient.

[0038] f4. Perform three-dimensional volumetric respiratory potential field strength calculation. For volume elements whose potential field assignment has not yet been completed v Its respiratory potential field strength F v for{ u k Intensity of respiratory potential field of each element in} and basic respiratory access costs Cost uk,v The minimum value is calculated using the following formula: ; in, for{ u k The values ​​have already been assigned to the elements in the} u k The intensity of the respiratory potential field.

[0039] f5. Layer-by-layer iterative calculation and global result locking Using the volume element containing the safety exit (Exit) as the reference point, a layer-by-layer diffusion method is adopted, that is, expanding outwards in layers according to their distance from the exit. During the calculation of each layer, adjacent volume elements that have already been assigned potential fields are used. F The value is iteratively assigned to the voxels within the effective range until the respiratory potential field strength of all voxels within the effective range is obtained. F The calculation locks the breathing potential field intensity of all volume elements within the non-dynamic taboo domain of the entire station; When the dynamic taboo domain of step f4 F ( t Upon completion of an update or a change in the status of fire protection facilities, immediately restart the entire potential field calculation process and update the values ​​of all elements within the effective range. F value.

[0040] f6. Path-end volumetric potential field calibration and pruning First, the potential field strength of the terminal volume element of each path in the path set. Perform calibration and evaluate the global optimality of each candidate path; Then, after each round of path expansion, the end volume element of each path in the current path set is calculated. value; Then, the current path With the end voxel of the previous round on this path Compare; Finally, if a certain path in three consecutive calibration rounds... All are greater than the previous round This indicates that the path gradually deviates from the exit direction and moves further away from the safe exit Exit during the expansion process. It only satisfies the local optimum in a small area around the evacuation starting point Start, but ignores global safety and efficiency. The path is removed from the path set to achieve path pruning and avoid the continuous expansion of invalid paths.

[0041] Specifically, the local slope angle in step B i Let be the angle between the direction of passage within this element and the horizontal ground. For vertical areas such as stairs, i >0°, for horizontal areas such as platforms and concourses. i = 0°.

[0042] Specifically, the preprocessing for the rail transit station scenario in step C is as follows: First, piston wind interference data is removed: Link the station platform screen doors / train signaling system to block the parameter jump data of sensors during the time when the train enters and leaves the station, and eliminate non-fire-related parameter fluctuations caused by piston wind; Then, perform environmental control condition linkage filtering: The system is linked to the station's ventilation and smoke exhaust environmental control system. Based on the start / stop status of the air vents and data on wind speed and direction, the system filters the airflow disturbance data of the sensors around the air vents and smoke exhaust vents to eliminate changes in non-fire parameters caused by the operation of the environmental control system. Next, vertical stratified data calibration is performed: For the multi-layered three-dimensional structure of the station concourse and platform, sensor data at different vertical heights are calibrated layer by layer according to floor to eliminate parameter deviations caused by vertical temperature differences and flue gas stratification. Finally, a screening process for large passenger flow interference is conducted: The system is linked to the station's AFC passenger flow system to eliminate data on normal increases in carbon monoxide concentration caused by high population density during morning and evening rush hours, thus avoiding parameter calculation distortions caused by non-fire factors.

[0043] Specifically, in step B, the nearest neighbor spatial matching method for rail transit station scenarios is used for sensor-3D volume element matching, as follows: First, perform pre-screening of the candidate matching range: In view of the multi-layered, three-dimensional, and multi-fire compartment structure of rail transit stations, only visible connected elements that are on the same floor and in the same fire compartment as the sensor, within the vertical height range of the breathing reference layer, and without any impenetrable boundary such as walls, columns, turnstiles, or equipment rooms between them and the sensor are included in the matching candidate range. Then, the priority of key evacuation areas is determined: For sensors located in evacuation bottleneck areas such as evacuation stairwells, safety exits, turnstile passages, and platform screen door end doors, a dedicated matching priority is assigned. Priority is given to matching these sensors with voxels in the same area, and the spatial radius of their matching candidate range is appropriately expanded to ensure the parameter matching accuracy of key evacuation areas. Next, perform nearest neighbor spatial mapping matching: Within the pre-screened matching candidate range, the spatial straight-line distance between the center node of each voxel and the sensor calibration coordinates is calculated, and the sensor is mapped to the nearest voxel center node to complete the one-to-one precise matching between a single sensor and a single voxel. Finally, after completing the above sensor-3D voxel matching, for the scenario where sensors are sparsely deployed in rail transit stations and it is impossible to achieve one-to-one matching between a single voxel and a sensor for the entire station, global parameter completion is performed on the voxels without sensor matching. All passable cubic elements within the breathing reference layer of the station that have been 3D subdivided are defined as interpolation candidate ranges; Based on the measured data of the voxels within the same fire compartment where sensor matching has been completed, the temperature data of the adjacent voxels without sensor matching is supplemented using the Kriging interpolation method that considers spatial correlation. T ( t (Unit: °C), carbon monoxide concentrationC ( t (Unit: ppm) and visibility S ( t (Unit: m) data; Then, using the interpolated volumetric data as the new benchmark, the process is iteratively repeated in the order of fire compartment - floor - entire station until the temperature of all volumetric data within the station's interpolation candidate range is reached. T ( t ), carbon monoxide concentration C ( t and visibility S ( t All data have been interpolated and completed.

[0044] Specifically, the dynamic taboo domain in step D F ( t The periodic updates are as follows: First, the adaptive update cycle, in the early stages of a fire. t ( t When the decay is gradual, the update cycle is 5 seconds; when the fire enters the rapid development stage... t ( t When decay accelerates, the update cycle is shortened to 1 second; Then, it is verified whether each element within the station meets the judgment criteria for core hazardous elements, forbidden elements, naturally impassable elements, and elements lacking evacuation conditions, based on the dynamic forbidden domain. F ( t ) Delete the voxels that do not meet the conditions and add the voxels that do meet the conditions; Finally, when the status of fire protection facilities changes, such as when the fireproof roller shutter falls or the smoke exhaust system is activated, a complete recalculation of the taboo domain is immediately triggered.

[0045] Example 1 The subway station selected in this embodiment is a two-level underground island platform subway station. The first underground level is the concourse level: it has 4 sets of evacuation staircases from the concourse to the ground exit, 4 ground safety exits, 2 sets of evacuation staircases between the concourse and the platform, and 8 entrance and exit turnstiles, divided into 2 fire compartments and 4 smoke compartments; the second underground level is the platform level: an island platform with full-height platform screen doors and end doors at both ends, connected to the concourse level by 2 sets of evacuation staircases between the concourse and the platform, divided into 2 fire compartments and 4 smoke compartments; the station's core evacuation bottlenecks are: 4 staircase entrances of 2 sets of inter-platform staircases (2 on the platform side and 2 on the concourse side, all located in the underground non-safe zone), and 4 entrances on the concourse side of 4 sets of exit staircases (core control nodes in the underground non-safe zone), totaling 8 core control staircase entrances, 8 turnstile channels, and end doors at both ends of the platform.

[0046] The core parameters of the subway station are calibrated as follows: Voxel subdivision parameters: Voxel dimensions for open areas in the central area of ​​the station hall and platform are 1.5m × 1.5m × 1.2m; general evacuation routes such as stair platforms and connecting passages use 1m × 1m × 1.2m voxels; core bottleneck areas such as the interior of turnstiles, stair treads, and safety exit doorways use ultra-fine voxels of 0.5m × 0.5m × 1.2m; breathing reference layer. H min =1.0m H max =2.2m, which fits the vertical height range of the mouth and nose of most people; the vertical dimension of 1.2m of the voxel completely covers the respiratory reference layer without cross-layer deviation; columns, turnstiles, walls and other impenetrable boundaries are not divided into voxels.

[0047] Sensing module deployment: A total of 48 temperature sensors, 24 carbon monoxide (CO) sensors, and 24 visibility sensors are deployed in the station. The sensors are located at the four corners of the station hall, on both sides of the eight core stairwells, next to the four safety exits, at both ends of the platform, and in the middle of the platform. All sensors are connected to the station's environmental control and emergency command system, which can achieve millisecond-level data synchronization.

[0048] Experimental calibration coefficients: α =0.02m³ / ppm, β =0.001m³ / ℃, c =0.05m 4 / s, l =0.0015ppm -1 , k =0.02m -1 , m =0.1m³ / person q 1 = 0.4 q =2=0.2, and all coefficients were calibrated through standardized toxicology, human metabolism, and human behavior experiments.

[0049] Tiered safety thresholds: First safety thresholds are set for evacuation bottleneck areas such as stairwells, safety exits, and turnstiles. t min1 =60s; In open public areas such as platforms and concourses, the second safety threshold is set. t min2 =30s; Human tolerance threshold for carbon monoxide concentration C max =800ppm, temperature human tolerance threshold T max =45℃.

[0050] Path optimization parameters: The number of optimal paths retained in each round of path expansion is m=3.

[0051] Based on the aforementioned subway stations, the specific implementation steps are as follows: A. Constructing a 1:1 three-dimensional digital twin model of a rail transit station: Using terrestrial 3D laser point cloud scanning and BIM forward modeling technology, a 1:1 3D reconstruction was performed on all elements of the station, including the station concourse level, platform level, two sets of evacuation staircases between the concourse and platform, four sets of evacuation staircases between the concourse and ground exit, turnstiles, platform screen doors, columns, safety exits, and equipment management rooms. After eliminating point cloud noise, the model was globally registered with a spatial accuracy error of ≤±0.05m. This accurately restored the station's 3D structure, the spatial location of facilities and equipment, and the complete vertical connectivity of the two-level evacuation staircases, providing a high-precision basic model for subsequent voxel subdivision and path planning.

[0052] B. Station 3D spatial volume element subdivision and attribute recording: 1. Establish a three-dimensional rectangular coordinate system (x, y, z) with the southwest corner of the station hall as the origin. The x-axis represents the east-west direction of the station, the y-axis represents the north-south direction, and the z-axis represents the vertical direction. 2. An optimized dynamic adaptive voxel partitioning strategy is adopted, balancing computational efficiency and accuracy: 1.5m×1.5m×1.2m voxels are used in the open areas of the station hall and platform center; 1m×1m×1.2m voxels are used in general evacuation passages such as stair platforms and connecting passages; 0.5m×0.5m×1.2m ultra-fine voxels are used in core bottleneck areas such as the interior of turnstiles, stair treads, and safety exit doorways; impenetrable boundaries such as columns, turnstiles, walls, and equipment rooms are not partitioned into voxels and are regarded as blocking boundaries for path planning. 3. The area from 1.0m to 2.2m above the ground is designated as the breathing reference layer. Subsequent environmental parameter calculations, safety threshold determinations, and path planning will only be carried out on passable elements within this layer. 4. Assign a unique ID to each subdivided volume element and record the spatial coordinates and volume of each volume element. V Adjacent volume element set N Local slope angle i (Horizontal area) i =0°, staircase area i =30°), the floor it belongs to, fire compartments and smoke control compartments, and complete the construction of the entire station's element attribute library.

[0053] C. Environmental parameter acquisition and voxel dwell time limit t ( t )calculate: 1. Sensor data preprocessing, matching, and interpolation completion: (1) Data preprocessing: For the raw data collected by the sensors, the station operation system is first linked to complete four preprocessing steps, and then outlier removal and millisecond-level time synchronization are performed: the platform screen doors and train signal system are linked to remove parameter jump data caused by piston wind during train entry / exit; the environmental control system is linked to filter the airflow disturbance around the ventilation / smoke exhaust outlets according to the opening and closing status of the air outlets and wind speed and direction; vertical data calibration is completed by the concourse / platform layer to eliminate parameter deviations caused by smoke layering; the AFC passenger flow system is linked to remove data on normal rise in carbon monoxide concentration caused by large passenger flow during morning and evening peak hours, ensuring the authenticity of basic data from the source; (2) Sensor-3D Voxel Matching: First, pre-screen the matching candidate range and include only visible connected voxels that are on the same floor as the sensor, in the same fire compartment, in the breathing reference layer, and without obstruction. For sensors in evacuation bottleneck areas such as the 8 core stairwells and safety exits, the highest matching priority is set, and the matching is completed first and the coverage radius is appropriately expanded. Finally, within the candidate range, the sensor is mapped to the center node of the voxel closest to its coordinates to complete the one-to-one accurate matching between the sensor and the voxel. (3) Global parameter interpolation and completion: All passable voxels within the breathing reference layer are defined as interpolation candidate ranges. Based on the measured data of voxels with matched sensors within the same fire compartment, the Kriging interpolation method considering spatial correlation is used to complete the temperature of adjacent voxels without matched sensors. T ( t ), carbon monoxide concentration C ( t ),visibility S ( t The data is then used as a new benchmark, and the process is repeated in the order of fire compartment - floor - entire station until all voxels within the interpolation candidate range have completed the environmental parameter completion. (4) Calculate the temperature change rate of each volume element based on real-time updated environmental parameters at a fixed time step. T ′( t )=d T / d t rate of change in carbon monoxide concentration C ′( t )=d C / d t This provides input for subsequent calculations of the maximum dwell time.

[0054] 2. Limiting dwell time of the body element t ( t Step-by-step calculation: by t Taking the ultra-fine element T-012 of the stair treads on the west side of the platform level at 30s as an example: (1) Calculate the rate of basic environmental degradation: This volume elementC ′(30)=2.5ppm / s、 T ′(30)=1.0℃ / s, S(30)=18m, substituting into the formula, we get: E (30)=0.002×2.5+0.001×1.0+0.05 / 18≈0.005+0.001+0.0028≈0.0088m³ / s; (2) Calculate the carbon monoxide toxicity correction factor: This volume element C (30) = 100ppm, substituting into the formula, we get: ; (3) Calculate the basic limit dwell time: This volume element is a hyperfine volume element of the stair tread, with a volume of V a =0.5×0.5×1.2=0.3m³, substituting into the formula, we get: t 0(30)=0.3 / 0.0088×0.8607≈29.3s; (4) Calculate the vertical height correction factor: the vertical height of the volume element from the ground outlet H z =12m, substituting into the formula, we get: W H =1 / (1+0.02×12)≈0.8065; (5) Ventilation system influence correction factor: This volume element is located in the effective air supply area of ​​the environmental control system, and the calibration... W e =1.15; (6) Crowd density correction: Real-time crowd density of the voxel. r (30) = 0.3 people / m³, substituting into the formula, we get: ; (7) Structural constraint correction: This element represents the bottleneck area of ​​the staircase, with no severe congestion, and is calibrated. W B =0.95; (8) Determine the final maximum dwell time: t (30)=29.3×0.8065×1.15×0.9704×0.95≈24.9s; This element represents the evacuation bottleneck area, corresponding to the safety threshold. t min1 =60s, t (30)≤60s, therefore marked as a core hazardous element, fully complying with the patent's graded threshold control rules; if the element C(30) ≥800ppm or T (30)≥45℃, directly set t (30)=0, marked as immediately lethal body element.

[0055] D. Dynamic taboo domain F ( t )Build and update: 1. Identification of core hazardous elements and differential threshold calibration: A hierarchical safety threshold system is adopted to traverse all passable elements in the station in real time and to evacuate bottleneck areas. t ( t ≤60s, open area t ( t Voxel elements with a duration of ≤30s are marked as core hazardous voxel elements; t =30s, a total of 28 core hazardous elements were marked at the station, all of which were located in the fire initiation area at the end of the platform and the stair treads; 2. Progressive Expansion of Prohibited Elements for Rail Transit Scenarios: Based on the smoke diffusion patterns and fire safety regulations in enclosed subway spaces, directional expansion is performed on core hazardous elements: Adjacent elements within the same smoke control zone that are three-dimensionally connected to the core hazardous element are marked as prohibited elements; all cross-zone passage elements connecting to the fire compartment where the core hazardous element is located are marked as prohibited elements; if a core hazardous element appears at the platform-side downhill end of the staircase between the concourse and the platform, two additional continuous elements are added along the uphill direction of the staircase and marked as prohibited elements; if a core hazardous element appears at the concourse-side entrance of the staircase between the concourse and the ground exit, two additional continuous elements are added along the uphill direction of the staircase and marked as prohibited elements; core hazardous elements at the platform end are marked as prohibited elements along the entire longitudinal length of the platform screen doors; a total of 67 prohibited elements are marked in this expansion. 3. Delineate naturally impassable and evacuation-incompatible body elements: The criteria are all body elements within the non-breathing reference layer, body elements in non-evacuation public areas such as the track area and equipment management rooms, undivided body elements corresponding to impenetrable boundaries such as columns, turnstiles, and walls, and evacuation bottleneck body elements whose personnel congestion rate exceeds the preset threshold as determined by the AFC passenger flow system. 4. Dynamic taboo domain F ( t ) Defining the overall scope: All core hazardous elements, all forbidden elements, and all naturally impassable elements and elements lacking evacuation conditions are uniformly included in the dynamic forbidden domain. F ( t Crossing is strictly prohibited in the route planning; 5. Periodic Adaptive Updates: Set the adaptive update cycle, especially in the early stages of a fire. t ( t When the decay is gradual, the update cycle is 5 seconds; when the fire enters the rapid development stage... t (t During accelerated decay, the update cycle is shortened to 1 second; the judgment conditions for each element within the station are verified in each update cycle, from... F ( t ) Delete the voxels that do not meet the conditions and add the voxels that meet the conditions; when the status of fire protection facilities such as fireproof roller shutters falling or smoke exhaust systems starting or stopping changes, immediately trigger a complete recalculation of the taboo domain.

[0056] E. Cumulative respiratory expenditure R ( P , t )calculate: For any continuum element sequence path from the evacuation start point (Start) to the safety exit (Exit), all continuum element sequences on it... P =( v 0 = Start, v 1, v 2,…, v n =Exit), where v 0、 v 1, ... v n Represents each voxel along the path, where n is the number of voxels traversed by the path, and defines the cumulative respiratory consumption value. R ( P , t The path breathing consumption value is the core evaluation index for path optimization, which comprehensively represents the path safety and evacuation efficiency.

[0057] Taking a certain route as an example: the evacuation starting point in the middle of the platform level (Start) v 0) - Platform open area - Station hall - Platform section stairs - Station hall level passage - Station hall - Ground exit stairs - Ground safety exit No. 1 (Exit) v 8, final volume element), the continuous volume element sequence is P =( v 0 = Start, v 1, v 2, v 3, v 4, v 5, v 6, v 7, v 8 = Exit), a total of 9 continuum elements. R ( P , t The calculation is as follows: 1. Calculation of basic travel time for path volume sequence Based on the optimized dynamically adapted voxel size, the base passage time for each voxel is calculated. The average walking speed of people p =0.9m / s, and the parameters of each volume element and the passage time are as follows:

[0058] 2. Calculation of voxel-based accessibility coefficient for rail transit station scenarios According to the formula The calculations are performed using the following core parameters: Horizontal open area volume element ( v 0, v 1, v 2, v 4, v 5): , , Therefore ; Staircase platform element v 3: (Bottleneck area) Therefore ; Stair treads v 6: Upward direction =1.0 + 0.4 × sin30° = 1.2 =1.2、 =1.0, therefore =1.44; Safety exit doorway element v 7: Therefore =1.1; Ground safety exit body v 8: We have reached a safe outdoor area with no obstacles to passage, therefore... =1.0.

[0059] 3. Calculation of instantaneous respiratory consumption of a single body element According to the formula Calculation, where For personnel to reach the body element v i The time is expressed in seconds (s). for Time-based element v i The maximum dwell time, measured in seconds. Safety exit unit. v 8. Located in a safe outdoor area. =+∞, therefore R 8=0, the instantaneous respiratory consumption of each body element is as follows: R 0 = 1.67 / 620 × 1.0 ≈ 0.00269,R 1 = 1.67 / 610 × 1.0 ≈ 0.00274, R 2 = 1.67 / 600 × 1.0 ≈ 0.00278, R 3 = 1.11 / 83.2 × 1.2 ≈ 0.0160, R 4 = 1.67 / 580 × 1.0 ≈ 0.00288, R 5 = 1.67 / 570 × 1.0 ≈ 0.00293, R 6 = 0.56 / 24.9 × 1.44 ≈ 0.0324 R 7 = 0.56 / 300 × 1.1 ≈ 0.00205, R 8 = 0.

[0060] 4. Final calculation of cumulative respiratory consumption throughout the entire pathway The instantaneous respiratory consumption values ​​of all cells along the path are summed to obtain the cumulative respiratory consumption value for the entire path: .

[0061] F. Multi-round volume element combination optimization and reverse respiratory potential field feedback calibration. Flowchart is attached. Figure 2 1. Using the evacuation starting point Start (middle of the platform level) as the path start point and Exit No. 1 ground safety exit as the path end point, filter the adjacent voxels that are not in the dynamic taboo domain. F ( t The 6 voxels within the range are the initial candidate voxels. The 6 paths from the evacuation starting point Start to the initial candidate voxels are used as the initial paths in the path set.

[0062] 2. For each path in the current path set, for the terminal element of a particular path (farthest from the evacuation start point), select its adjacent element set. N Not in the dynamic taboo domain F ( t The set of volume elements within ) oh j}, j ≥1, construct from the original path end voxel to oh j The new extension path.

[0063] 3. Calculate each one in turn. oh j Cumulative respiratory consumption corresponding to the new pathway R ( P ,t Preserve the station's spatial connectivity rules (do not cross non-connected elements) and R ( P , t The three smallest paths are the optimal expansion paths.

[0064] 4. Calculate the optimal expansion path for each path in the path set in turn, and update the path set accordingly.

[0065] 5. Repeat steps 2-4 above to perform multiple rounds of path end-voxel expansion. After each round of path end-voxel expansion, initiate reverse respiratory potential field feedback calibration to guide the path towards the safe exit direction and avoid getting trapped in local optima. The steps for reverse respiratory potential field feedback calibration are as follows: Respiratory potential field baseline initialization: Using the volume element where the safety exit (Exit) is located as the core baseline point, calibration is performed. F (Exit)=0.

[0066] Defining the range of constraints for potential field iteration: only for non-potential fields. F ( t The potential field is calculated for the passable volume elements within the area. The diffusion of the potential field strictly follows the hierarchical order of the same smoke control zone - same fire control zone - same floor. Potential field diffusion blocking rules are set for impenetrable boundaries.

[0067] Calculation of basic respiratory passage cost for adjacent volume elements: For volume elements whose potential field assignment has not yet been completed. v Its adjacent set of volume elements N v The set of already assigned values ​​in the middle is { u k}, k ≥1, calculate the basic respiratory passage cost between two voxels. .in, For personnel through body element u k The passage time is expressed in seconds. This is the start time of the potential field calibration. Lower body element u k The maximum dwell time, expressed in seconds; For body element u k The passability coefficient is dimensionless.

[0068] Iterative assignment of respiratory potential field intensity of volume element: Volume element to be assigned v respiratory potential field strength The layer-by-layer diffusion method is used for iterative calculation until the respiratory potential field intensity of all volume elements within the effective range is calculated. F The calculation locks the global potential field result; when F (t When updates are made or the status of fire protection facilities changes, the entire potential field calculation process should be restarted immediately, updating the values ​​of all volume elements within the effective range. F value.

[0069] Path-end voxel potential field calibration and pruning: After each round of path expansion, calculate the potential field of each path-end voxel in the current path set. Value; will be the current path With the end voxel of the previous round on this path Compare; if a certain path in three consecutive calibration rounds... All are greater than the previous round This indicates that the path gradually deviates from the exit direction and moves further away from the safe exit Exit during the expansion process. It only satisfies the local optimum in a small area around the evacuation starting point Start, but ignores global safety and efficiency. The path is removed from the path set to achieve path pruning and avoid the continuous expansion of invalid paths.

[0070] 6. After multiple rounds of path end-vowel expansion and reverse respiratory potential field feedback calibration, all paths in the path set achieved voxel connectivity from the evacuation start point (Start) to the safety exit (Exit). The cumulative respiratory consumption value was selected from these paths. R ( P , t The path with the shortest distance is taken as the optimal evacuation path.

[0071] The complete evacuation route is as follows: Evacuation start point in the middle of the platform level (Start) - North-South passageway on the platform - West concourse - Evacuation staircase between platforms - Southwest passageway on the concourse level - Southwest concourse - Ground exit evacuation staircase - Ground safety exit No. 1 (Exit). The total evacuation time for this route is 46 seconds. R ( P , t =0.0645, with no crossing of dynamic taboo domains throughout the entire process. F ( t (area). Based on the dynamic data collected by the sensors, the optimal evacuation route is dynamically updated.

[0072] This invention addresses the core challenge of traditional two-dimensional planar planning methods being unable to adapt to the vertical connectivity and multi-obstacle three-dimensional structures of rail transit station concourses and platforms. By constructing a 1:1 full-element three-dimensional digital twin model of the station, and employing a dynamic adaptable voxel partitioning strategy of large voxels in open areas and small voxels in bottleneck areas, it balances computational efficiency with the accuracy of depicting key evacuation areas. By defining a breathing reference layer that conforms to the breathing characteristics of personnel, the paths planned based on the three-dimensional voxel model can accurately guide the continuous evacuation of personnel in both horizontal and vertical spaces, fully adapting to the unique evacuation scenarios of the multi-level underground three-dimensional structures of rail transit stations.

[0073] This invention solves the problems of existing technologies failing to eliminate interference specific to rail transit station scenarios and the distortion of the matching between sensing data and evacuation space. By linking the station's train operation, environmental control, and AFC passenger flow systems, it performs preprocessing on the raw sensor data specifically for rail transit station scenarios, including piston wind interference removal, environmental control condition filtering, vertical layer calibration, and large passenger flow interference identification, ensuring the authenticity of the basic data from the source. It adopts the nearest neighbor spatial matching method for rail transit station scenarios to achieve precise binding between sensors and voxels. Through Kriging interpolation considering spatial correlation, it achieves full-domain completion of the environmental parameters of accessible voxels throughout the station under sparse sensor deployment. Combined with an adaptive second-level dynamic update mechanism, it ensures that path planning is always synchronized with the real-time status of the fire scene, meeting the high timeliness requirements of emergency evacuation in rail transit stations.

[0074] This invention addresses the industry pain points of existing technologies, such as poor adaptability of general evaluation indicators to rail transit station evacuation scenarios, imbalance between safety and efficiency goals, and susceptibility to local optima. It proposes a volume element-limited dwell time. t ( t This study considers the comprehensive hazards of fire environmental parameters such as temperature, carbon monoxide concentration, and visibility to personnel evacuation, and introduces four correction coefficients specifically for rail transit station scenarios. This enables precise quantification of the safety tolerance limit of personnel in rail transit station fire environments. It also defines a cumulative breathing consumption value that incorporates corrections for slope, bottlenecks, and cross-zone conditions. R(P , t This transforms the dual objectives of safety and efficiency into a single, quantifiable optimization indicator; and constructs a dynamic taboo domain adapted to the fire protection specifications and smoke diffusion patterns of rail transit stations. F ( t The system achieves proactive avoidance of hazardous areas through directional progressive expansion; it designs a two-way optimization mechanism of positive volume element combination optimization and reverse breathing potential field global calibration, combined with path deviation pruning strategy, which not only retains local optimal candidate paths, but also guides the avoidance of local optima through global situation field, truly achieving the synergistic optimization of evacuation safety and traffic efficiency in rail transit station fire scenarios.

[0075] This invention fully integrates real-time sensor data, station environmental control system operating conditions, train signals, passenger flow monitoring data, and fire protection facility status into the entire process of extreme dwell time calculation, dynamic taboo domain update, potential field calibration, and path planning. The sensor data is no longer isolated information but becomes the core driving source of dynamic path planning. At the same time, the solution can be seamlessly embedded into the existing smart emergency command platform of rail transit stations, and is compatible with existing sensing, environmental control, and fire protection systems. It can be implemented without large-scale modifications, forming a complete technology chain of real-time perception, dynamic modeling, accurate planning, and closed-loop updating. The planning results are highly consistent with the actual evacuation scenarios of rail transit stations in fires.

Claims

1. A method for planning emergency evacuation routes in three-dimensional space at rail transit stations, characterized in that: Includes the following steps: A. Construct a 1:1 scale 3D digital twin model of the rail transit station. The 3D digital twin model accurately reproduces the 3D structure of the station hall, platform, stairs, passages, and internal facilities and equipment. B. Establish a three-dimensional spatial coordinate system for the three-dimensional digital twin model. x , y , z The station's three-dimensional space is discretized into a series of three-dimensional voxels. The dimensions of each voxel are dynamically adapted according to the spatial geometric features, and the spatial coordinates and volume of each voxel are recorded. V Adjacent volume element set N Local slope angle θ And establish a respiratory reference layer; C. Deploy temperature sensors, carbon monoxide sensors, and visibility sensors within the station. Calculate the temperature of each three-dimensional volume element based on the dynamic data collected by the sensors. T ( t ) and its rate of change T ′( t ), carbon monoxide concentration C ( t ) and its rate of change C ′( t ),visibility S ( t The system also calculates the limiting dwell time of each three-dimensional volume element in real time. τ ( t ),in t For time; D. Constructing dynamic taboo domains F ( t Adaptive periodic dynamic update is adopted. F ( t The three-dimensional element within the fire protection facility will immediately trigger a recalculation of the forbidden domain when the status of the fire protection facility changes. E. Any continuous three-dimensional voxel sequence path from the evacuation start point (Start) to the safety exit (Exit), where all continuous voxel sequences are: P =( v 0 = Start, v 1, v 2,…, v n =Exit), where v 0、 v 1, ... v n Represents each voxel along the path, where n is the number of voxels traversed by the path, and defines the cumulative respiratory consumption value. R ( P , t This represents the path respiratory consumption value and cumulative respiratory consumption value for this path. R ( P , t As the core evaluation indicator for path optimization, it comprehensively represents path safety and evacuation efficiency. F. Based on dynamic taboo domain F ( t ) and cumulative respiratory expenditure R ( P , t Starting from the evacuation point Start, the system generates initial paths in the path set. Through multiple rounds of three-dimensional voxel combination optimization and reverse breathing potential field feedback calibration, it generates the optimal evacuation path from the evacuation point Start to the safety exit Exit. The optimal evacuation path is then dynamically updated based on dynamic data collected by sensors.

2. The method for planning three-dimensional emergency evacuation routes in rail transit stations according to claim 1, characterized in that: In step B, the dimensions of the 3D volumetric element are dynamically adapted based on spatial geometric features. The specific process is as follows: First, large-scale elements are used in open areas such as platforms and concourses to ensure computational efficiency; Then, small voxels are used in narrow passages, staircases, and escalator entrances to ensure precise calculations of bottleneck areas; Furthermore, columns, turnstiles, and wall obstacles in the station's three-dimensional space are considered as impenetrable boundaries, and their occupied three-dimensional space is not discretized into a three-dimensional volumetric mesh; Finally, it will be above the ground. H min ~ H max The three-dimensional volume elements within the range serve as a respiratory reference layer. H min , H max These represent the minimum and maximum values ​​of the vertical height range of the mouth and nose of the crowd, respectively.

3. The method for planning three-dimensional emergency evacuation routes in rail transit stations according to claim 1, characterized in that: Step C involves calculating the temperature of each three-dimensional volume element. T ( t ) and its rate of change T ′( t ), carbon monoxide concentration C ( t ) and its rate of change C ′( t ),visibility S ( t The specific process is as follows: First, each sensor was calibrated on-site, and its three-dimensional coordinates of physical installation, floor, and fire compartment were recorded. Then, calculate the three-dimensional coordinates of each vertex and geometric center node, the floor to which it belongs, the fire compartment, and the positional relationship with the surrounding impenetrable boundaries for each three-dimensional volume element in the station; Next, the raw data collected by the sensors is first preprocessed in conjunction with the station operation system for the rail transit station scenario, and then outlier removal and time synchronization are completed. Next, the nearest neighbor spatial matching method for rail transit station scenarios is used to match sensors and 3D voxels. Finally, based on the real-time data collected by various sensors within the station, the temperature of each volumetric element within the station was monitored. T ( t ), carbon monoxide concentration C ( t and visibility S ( t The data is dynamically updated, and the rate of temperature change of each volume element is calculated. T ′( t )=d T / d t rate of change in carbon monoxide concentration C ′( t )=d C / d t , where d T d represents the change in temperature. C d represents the change in carbon monoxide concentration. t For time step.

4. The method for planning three-dimensional emergency evacuation routes in rail transit stations according to claim 1, characterized in that: In step C, the ultimate dwell time of each three-dimensional voxel is calculated in real time. τ ( t This includes calculating the basic limit dwell time. τ 0( t The specific process is as follows: First, calculate the rate of degradation of the basic environment. E ( t This method converts multi-dimensional independent fire factors within the enclosed space of a rail transit station into an equivalent air volume consumption rate with a unified dimension. These multi-dimensional independent fire factors include the carbon monoxide concentration change rate. C ′( t ), temperature change rate T ′( t ),visibility S ( t ), rate of basic environmental deterioration E ( t The formula for calculating ) is: ； in, α The conversion coefficient for the rate of change of carbon monoxide concentration was determined through toxicological experiments. β The conversion coefficient for the rate of temperature change was determined through thermophysiological experiments. γ The visibility-affected conversion coefficient was determined through behavioral psychology experiments. Then, calculate the carbon monoxide toxicity correction factor. G c ( t This reflects the direct impact of carbon monoxide on the oxygen-carrying capacity of human blood. The calculation formula is: ； in, λ The toxicity attenuation index was determined by the carboxyhemoglobin (COHb) assay. Finally, calculate the basic limiting residence time. τ 0( t Based on the volumetric effective air volume, the basic environmental degradation rate, and the carbon monoxide toxicity correction coefficient, the basic safe residence time is quantified, and the calculation formula is as follows: ； in, V a The total usable effective air volume within the volume element is the sum of its parts and the volume of the volume element. V The values ​​are the same.

5. The method for planning three-dimensional emergency evacuation routes in rail transit stations according to claim 4, characterized in that: In step C, the ultimate dwell time of each three-dimensional voxel is calculated in real time. τ ( t (including the basic limit stay time) τ 0( t The correction process is as follows: First, calculate the vertical height correction factor for rail transit stations. W H Considering the characteristic of rail transit stations being deeply buried underground, the reduction effect of vertical distance on personnel tolerance is increased. The farther away from the exit, the shorter the equivalent safe dwell time. The calculation formula is as follows: ； in, κ The vertical height influence coefficient, calibrated through human metabolic experiments, reflects the impact of the vertical distance between a person and the ground exit on psychological panic and physical exertion. H z This represents the vertical height difference between the center of the volume element and the nearest ground exit; for ground stations or above-ground sections, this value is 0. Then, the influence correction factor of the rail transit station ventilation system is introduced. W e This reflects the impact of the station's smoke exhaust / makeup air system status on smoke diffusion velocity and visibility recovery, and is linked to the real-time operating condition calibration of the station's smoke exhaust / makeup air system. The volume element is within the effective makeup air zone. W e >1, when in the area of ​​smoke exhaust failure and smoke accumulation 0< W e <1, When the environmental control system is not started W e =1; Next, calculate the correction for crowd density at rail transit stations. To address the high population density characteristic of rail transit stations, a crowd density correction factor is introduced to reflect the additional impact of crowd congestion on air consumption and mobility in the calculation. The higher the density, the shorter the equivalent safe dwell time. The calculation formula is as follows: ； in, μ The population density influence coefficient was determined through population behavior experiments. ρ (t) represents the real-time crowd density within the volume element, obtained through station video analysis; Next, the structural constraint correction for rail transit stations is calculated. W B Considering the structural characteristics of staircases and passageways in rail transit stations, a structural constraint factor is introduced to reflect the amplification effect of personnel congestion on evacuation time in staircases and passageways. The calculation formula is as follows: ； in, ξ Structural constraint factors, calibrated using social force models or historical data, are used in non-congested areas. ξ =0, congested areas are set to 0< ξ <1; Finally, the limiting dwell time of the volume element was determined. τ ( t The calculation formula is: 。 6. The method for planning three-dimensional emergency evacuation routes in rail transit stations according to claim 1, characterized in that: In step D, construct the dynamic taboo domain. F ( t The specific process is as follows: First, the core hazardous elements are identified, and differentiated thresholds are calibrated for rail transit station scenarios. Then, the taboo elements of the rail transit station scene are progressively expanded; Next, determine the natural impassable and unsuitable evacuation conditions for the body elements; Finally, all risk elements across the entire station are uniformly incorporated into the dynamic taboo domain. F ( t Risk elements include all core hazardous elements, all forbidden elements, and elements that are naturally impassable or lack evacuation conditions.

7. The method for planning three-dimensional emergency evacuation routes in rail transit stations according to claim 1, characterized in that: Cumulative respiratory consumption in step E R ( P , t The calculation process is as follows: First, the basic travel time of the path volume sequence is calculated, targeting... P =( v 0 = Start, v 1, v 2,…, v n Each three-dimensional volume element on =Exit) v i ( i =0, 1, 2, ..., n), and combined with the pedestrian traffic characteristics of different functional areas of the rail transit station, calculate the basic passage time of pedestrians through this three-dimensional volume element. The calculation formula is: ； in, L i For volume elements along the direction of personnel movement v i Feature length; The average walking speed of people; Then, for the scenario of rail transit stations, the voxel access adaptation coefficient is calculated for each voxel on the path. v i Spatial structure and regional attributes are used to calculate the volume element accessibility coefficient. The formula for quantifying the increase in respiratory consumption under non-ideal travel conditions is as follows: ； in, This is the vertical slope correction factor, based on the local slope angle of the volume element. θ Calibration, horizontal area θ =0° =1.0, vertical area θ >0° and the person is moving upwards =1.0+ q 1·sin θ Vertical area θ >0° and the person is in the downward direction =1.0+ q 2·sin θ , q 1. q 2 represents the passability coefficient calibrated in human metabolic experiments; The correction factor for evacuation bottlenecks is used when the volume element is located in evacuation bottleneck areas such as turnstiles, stairwells, and safety exits. >1.0 indicates a narrower passage width and higher population density in open areas. =1.0; For cross-zone correction coefficients, when the volume element is a transitional volume element connecting fire compartments and floors. >1.0, conforming to the door restrictions for cross-zone and cross-floor passage in rail transit stations, and the increased respiratory consumption caused by the collision of passenger flow, non-transitional body element. =1.0; Next, the instantaneous respiratory consumption value of a single body element is calculated for each body element along the path. v i Based on the voxel travel adaptation coefficient, the instantaneous respiratory consumption of a person passing through that voxel is calculated. R i To achieve the coupling and quantification of security and efficiency, the calculation formula is as follows: ； in, For personnel to reach the body element v i The moment; for Time-based element v i The maximum dwell time; Finally, the cumulative respiratory consumption along the entire pathway is calculated, including the instantaneous respiratory consumption of all body elements along the pathway. R i The values ​​are summed to obtain the cumulative respiratory consumption for the entire path. R ( P , t The calculation formula is: 。 8. The method for planning three-dimensional emergency evacuation routes in rail transit stations according to claim 1, characterized in that: In step F, the evacuation starting point Start is used as the path starting point to generate initial paths in the path set. The specific process is as follows: First, taking the evacuation starting point Start as the path starting point, the set of three-dimensional voxels adjacent to the evacuation starting point Start is... N Not in the dynamic taboo domain F ( t The three-dimensional volume elements within the range are used as initial candidate volume elements; Then, the several paths from the evacuation starting point Start to the initial candidate voxels are taken as the initial paths in the path set.

9. The method for planning three-dimensional emergency evacuation routes in rail transit stations according to claim 1, characterized in that: Step F involves multiple rounds of 3D voxel combination optimization, and the specific process is as follows: First, for each path in the current path set, for the terminal element of a particular path (i.e., the side furthest from the evacuation start point), select its adjacent element set. N Not in the dynamic taboo domain F ( t The three-dimensional volume element set within ) ω j }, j ≥1, construct from the original path end voxel to ω j The new extension path; Then, calculate each one in turn. ω j Cumulative respiratory consumption corresponding to the new pathway R ( P , t (), retaining compliance with station spatial connectivity rules, not crossing non-connected elements and R ( P , t The smallest m ( m ≥1) paths are the optimal expansion paths; Next, the optimal expansion path for each path in the path set is calculated sequentially, and the path set is updated accordingly. Then, repeat the above steps to perform multiple rounds of path end voxel expansion, and after each round of path end voxel expansion, initiate reverse respiratory potential field feedback calibration to guide the path towards the safe exit direction for optimization and avoid getting trapped in local optima. Finally, after multiple rounds of path-end voxel expansion and reverse respiratory potential field feedback calibration, all paths in the path set achieved voxel connectivity from the evacuation start point (Start) to the safety exit (Exit). The cumulative respiratory consumption value was then selected. R ( P , t The path with the shortest distance is taken as the optimal evacuation path.

Images