Methods and systems for predicting and optimizing emergency evacuation of personnel during subway renovation and construction phases

By constructing a directed graph network model and a social force model, the spatiotemporal distribution of congestion waves was predicted, evacuation routes were optimized, and the evacuation simulation problem during the renovation and construction of subway underground station groups was solved. This enabled proactive evacuation prediction and intervention, and improved the intelligence and safety of evacuation.

CN123117488APending Publication Date: 2026-11-17SHENZHEN UNIV +1
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611629079.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-21
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing evacuation simulation technologies cannot adapt to changes in spatial topology during the renovation and construction of subway underground station complexes. They lack the ability to predict and proactively intervene in surges in passenger flow, resulting in the inability to scientifically formulate evacuation plans and accurately assess evacuation safety.

Method used

By constructing a directed graph network model through an intelligent control chain of 'semantic topology modeling—S-FSM state modeling—congestion prediction—path optimization—guided execution—simulation evaluation', the model predicts the spatiotemporal distribution of congestion waves, optimizes evacuation paths, and combines a social force model for microscopic simulation to achieve dynamic evacuation decision-making.

Benefits of technology

It enables proactive evacuation prediction and intervention during construction, improving the intelligence and safety of evacuation and adapting to the renovation and construction scenarios of subway station clusters with multiple intersecting lines.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117488A_ABST
    Figure CN123117488A_ABST
Patent Text Reader

Abstract

This invention relates to a method and system for predicting and optimizing emergency evacuation of personnel during the construction phase of subway renovation projects. The method includes a spatial network modeling stage, a semantic rule constraint system establishment stage, a pedestrian congestion wave prediction and quantitative analysis stage, a multi-exit dynamic path allocation optimization stage, a micro-level pedestrian simulation stage based on a social force model, and a total evacuation completion time assessment and scheme output stage. Through an intelligent control chain of "semantic topology modeling—S-FSM state modeling—congestion prediction—path optimization—guided execution—simulation evaluation," the method achieves a leap from "passive response" to "active prediction and intervention" in evacuation decision-making during subway underground station group renovation construction scenarios, significantly improving the intelligence level, safety, and engineering applicability of emergency evacuation during renovation construction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of emergency evacuation technology for subway personnel, and in particular to a method and system for predicting and optimizing emergency evacuation of personnel during the subway renovation and construction phase. Background Technology

[0002] With the continuous expansion of urban rail transit networks, many early-built subway stations have entered the renovation and upgrading phase. Subway underground station renovation projects often involve multiple lines converging and being renovated in phases; that is, some lines need to maintain normal operation and service during construction, while others are closed for construction. This special condition of "operating while renovating" is becoming increasingly common in urban rail transit upgrades, posing serious challenges to operational safety and emergency management during construction.

[0003] During the renovation and construction of subway underground station complexes, the core challenge of passenger evacuation exhibits significant spatiotemporal coupling characteristics. On the one hand, construction barriers and temporary partitions significantly alter the passageway within the station, creating a series of "static bottlenecks." Some passageways are closed or rerouted, potentially rendering normal evacuation routes ineffective. On the other hand, subway operations rely on strict timetables. Train arrivals and departures, transfers, and other factors cause passenger flow to concentrate at specific times, forming passenger flow time pulses. These pulses, coupled with the staggered opening times of exits / passageways, collide and overlap within confined spaces, easily triggering sudden congestion waves. In the event of emergencies such as fires, structural instability, or equipment failures, how to scientifically formulate passenger evacuation plans and accurately assess evacuation safety under the dual uncertainties of dynamic changes in the construction environment and passenger flow pulses remains a critical technical problem to be solved.

[0004] Regarding the dynamic evacuation of pedestrians, although existing evacuation simulation technology provides some support for emergency evacuation in subway stations, it still has the following significant shortcomings under the special circumstances of subway underground station group renovation and construction: Existing systems (such as the simulation path planning method and system for passenger evacuation in subway stations under multiple hazard sources, CN114298438A) optimize paths based on fixed spatial layouts. They do not consider the impact of phased changes in spatial topology such as enclosures and temporary passages during the renovation and construction of subway underground station groups on the evacuation network. They cannot adapt to dynamic evacuation scenarios under special conditions of "partial line operation and partial line closure". Their path planning relies on statically preset evacuation networks and has not established a congestion prediction mechanism based on the temporal distribution law of pulse surge passenger flow. They lack the ability to quantitatively predict the congestion wave driven by the S-FSM semantic state transition caused by the collision between the influx pulse and the phased release of the exit after the sudden event is triggered. It is difficult to achieve proactive path intervention and exit load balancing optimization based on congestion prediction.

[0005] Existing systems (such as a simulation analysis method for emergency evacuation in rail transit, CN118278217A) focus on simulation analysis of fire scenarios at a single station, without addressing special conditions such as the cross-influence of multiple lines and partial area closures during the renovation and construction of underground subway station groups. Their path verification is a post-event evaluation-type decision-making process, lacking the intelligent decision-making ability to predict and proactively intervene in congestion waves driven by the four-stage semantic state transition of S-FSM based on constraint-aware semantic topological bottlenecks and the temporal distribution patterns of surge passenger flow before evacuation is initiated. Furthermore, no congestion evolution prediction model based on semantic rule constraints has been established.

[0006] Existing systems (such as the Metro Evacuation Dynamic Decision Simulation System CN120874364A based on virtual reality and artificial intelligence) focus on enhancing the realism of drills through VR interaction. Their path planning and decision-making are still based on preset evacuation scenarios and fixed spatial layouts, without addressing the dynamic changes in spatial topology during the construction of metro underground station groups as construction progresses. They have not established a congestion wave prediction mechanism based on pulse surge passenger flow (influence pulses triggered by sudden events / timetables and phased releases due to changes in exit / passage status), and have not formed a knowledge-driven spatiotemporal control framework for "prediction-optimization-evaluation" intelligent closed-loop evacuation decision-making capabilities. As a result, they are unable to cope with complex evacuation problems involving spatiotemporal coupling under special construction and renovation conditions.

[0007] In summary, on the one hand, existing evacuation simulation methods in the context of subway underground station renovation and construction mostly rely on the final spatial layout to establish a fixed network model. This fails to reflect the dynamic changes in spatial topology caused by construction barriers, temporary passages, and closed areas as the renovation progresses. Furthermore, these methods fail to incorporate passenger flow time pulses driven by subway train timetables and changes in exit / passage status into the congestion prediction model. They largely depend on the assumption of uniform distribution or instantaneous density data for static extrapolation and lack the ability to predict the time and location of congestion waves using known timetable information. On the other hand, existing methods are mostly "post-event response" decision-making models, which passively adjust the plan after congestion is discovered through simulation. They have not yet formed the intelligent decision-making ability to proactively optimize path allocation and dynamically balance exit load based on bottleneck distribution and pulse patterns before evacuation begins. Moreover, they do not involve a closed-loop intervention mechanism that transforms the optimization results into dynamic information dissemination and guidance strategies and considers differences in passenger compliance rates.

[0008] Therefore, existing technologies still have significant gaps in intelligent evacuation decision-making under the special conditions of "operation and renovation" of subway underground station clusters, especially lacking an integrated system that can combine spatial dynamic modeling, pulse surge passenger flow prediction and proactive intervention decision-making. Summary of the Invention

[0009] This invention provides a method and system for predicting and optimizing emergency evacuation of personnel during the construction phase of subway renovation. It aims to achieve a leap from "passive response" to "active prediction and intervention" in the evacuation decision-making process during the renovation of subway underground station groups through an intelligent control chain of "semantic topology modeling - S-FSM state modeling - congestion prediction - path optimization - guided execution - simulation evaluation". This significantly improves the intelligence level, safety and engineering applicability of emergency evacuation during the renovation construction period.

[0010] According to a first aspect of the present disclosure, a method for predicting and optimizing emergency evacuation of personnel during the construction phase of subway renovation is provided, comprising the following steps: Spatial network modeling stage: Obtain spatial layout information of each construction stage of the subway underground station group renovation project, and construct a directed graph network model based on the boundary information of the spatial layout; The stage of establishing a semantic rule constraint system includes setting two types of triggering conditions for influx pulses: one is the regular passenger flow pulse, and the other is the emergency evacuation pulse. Pedestrian congestion wave prediction quantitative analysis stage: The evolution pattern of queuing state caused by limited exit capacity is identified through the four-stage semantic state transition of S-FSM. The two shock waves of the influx pulse are quantitatively analyzed. For the bottleneck edge with limited capacity, the queue formation time and queue number are calculated based on the number of people entering and the actual capacity of the bottleneck. The number of people staying is calculated based on the integral of the flow rate of people entering the waiting area from the queue formation to the stage release trigger time. The duration of the release process and the exit evacuation rate are calculated based on the ratio of the number of people staying to the actual capacity of the exit. The spatiotemporal distribution prediction results of the congestion wave driven by S-FSM semantic state transition are output. The four-stage semantic state of S-FSM includes the initial response and waiting stage, the approach and queuing stage, the passage release and passage stage, and the arrival at the safe area stage. Multi-exit dynamic path allocation optimization stage: Based on the spatiotemporal distribution prediction results of congestion waves, the maximum evacuation completion time of each exit is minimized as the optimization variable. All available evacuation exits and their actual capacity under the current working conditions are determined, as well as the total number of people to be evacuated. The number of people allocated to each exit is used as the optimization variable. Equality constraints are established, and a load balancing optimization model is constructed. The optimal solution satisfies that the ratio of the number of people allocated to each exit to its capacity is equal. The passenger compliance rate parameter is introduced. The corrected actual number of people allocated to each exit is the weighted sum of the number of people allocated by the optimized scheme and the number of people allocated by inertial behavior. The microscopic pedestrian simulation stage based on the social force model: Based on spatial layout information and a directed graph network model, pedestrians are made to conform to the social force model by being given self-driving force, interaction force between pedestrians, and interaction force between pedestrians and obstacles. The speed and position of pedestrians are updated according to the motion equation to realize the simulation and deduction of the evacuation process. Total evacuation completion time assessment and plan output stage: The total evacuation completion time is used as the evaluation index, calculated as the difference between the time when the last passenger arrives at the safe exit and the evacuation trigger time; if it exceeds the safety standard, it returns to the multi-exit dynamic path allocation optimization stage for re-optimization; the deviation between the predicted congestion parameters and the actual simulation values ​​is fed back to the parameter library for subsequent stage accuracy optimization; when switching construction stages, the closed area weights and temporary edges of the directed graph network model are updated, and the semantic rule constraint system stage is repeatedly established until the total evacuation completion time assessment and plan output stage, forming an evacuation plan system covering the entire transformation cycle.

[0011] A further technical solution of the present invention is: in the pedestrian congestion wave prediction and quantitative analysis stage, the phased release represents the batch activation events that indicate the switch of an exit or passage from an impassable state to a passable state.

[0012] A further technical solution of the present invention is as follows: In the pedestrian congestion wave prediction quantization stage, the conditions for the semantic state transition of the four stages of S-FSM are: When pedestrians i If the pedestrian's current location is not within the safe zone or waiting area node, and the safety exit has not yet been opened, i Currently in the initial response and waiting phase; When pedestrians i When the pedestrian's current location is in a safe zone or waiting area node, and the safety exit has been partially opened, i In the approaching and queuing stage; When pedestrians i The current location is within the queuing buffer and the target exit is the currently active exit. Pedestrian i It is currently in the stage of allowing passage and clearing. When pedestrian i's current position is at the target exit position, pedestrian i is in the stage of reaching the safe area.

[0013] A further technical solution of the present invention is: pedestrian congestion wave prediction and quantitative analysis stage: Queue formation time With the number of people in the queue The expression is: , ,in, t This refers to the moment when the inrush pulse reaches the upstream of the bottleneck. q For the number of people injected; The actual traffic capacity at the bottleneck; τ This represents the average travel time for pedestrians from the influx source to the bottleneck. q> At that time, a queue is formed; Calculate the number of residents The expression is: ,in, For the time when the queue is formed, The triggering time for the inrush pulse; The flow rate of people entering the waiting area is expressed in people / second, and the flow rate of people leaving before the phased release trigger is 0. Duration of release process With exit evacuation flow rate The specific expression is: , ,in, To stimulate exports The actual traffic capacity, when Exceeding exports Actual traffic capacity When the product of the waiting time and the waiting time is multiplied, secondary congestion will occur at the exit.

[0014] A further technical solution of the present invention is as follows: the load balancing optimization model specifically includes: Let the set of available exits be O = {o1, o2, ..., o...} m The actual throughput capacity of each exit is: j=1, ..., m, where m is the number of exits; pedestrians are divided into R starting groups according to their injection source, and the number of people to be evacuated in the r-th starting group is... r=1, ...,R, satisfying , Let r be the total number of people to be evacuated from the starting group, and let r be the set of exits that can be reached by the pedestrians in the r-th group. ;set up Let the number of people in group r assigned to exit j be the optimization variable, satisfying the following condition: (All pedestrians in each group were evacuated.) ≥0, and the actual number of people allocated to each export Limited by its service capacity, it can meet the needs of customers. , Given the target evacuation time window, the optimization objective is to minimize the maximum estimated travel time at each exit. The optimal solution satisfies the condition that the ratio of the total number of people allocated to each exit to its travel capacity is equal. ; Optimal allocation of personnel to each export Calculate as follows: ; The obtained optimal route allocation scheme is transformed into an information dissemination and guidance strategy, introducing a passenger compliance rate parameter λ, 0≤λ≤1, which characterizes the actual response rate of passengers to dynamic guidance information. The value of λ is determined based on actual operational statistics of similar hubs or on-site test calibration. λ=1 indicates complete compliance, and λ=0 indicates complete non-compliance. The proportion of passengers who comply with the guidance is... Evacuation will be carried out according to the optimized plan; the proportion of passengers who do not follow the guidance is [missing information]. Evacuation will be carried out according to the inertial path selection behavior; Revised actual number of personnel allocated to each export for: ,in To optimize the allocation of personnel at exit j in the plan, The number of people allocated to exit j under inertial behavior.

[0015] A further technical solution of the present invention is: a microscopic pedestrian simulation stage based on a social force model, wherein the social force model specifically includes: The motion of pedestrian i is described by Newton's second law: ,in, Let i be the mass of pedestrian i; It is the velocity vector; Self-driving force; Repulsive force between pedestrians; Repulsive force from obstacles; self-driving force The formula is: ,in, For the desired velocity scalar, Let be the unit vector pointing to the target location. The current velocity vector, Relaxation time; interpersonal repulsion The formula is: ,in and The interaction strength parameter; Let i be the radius of pedestrian i; Let be the distance between pedestrian i and pedestrian j; Let g be a unit vector pointing from pedestrian j to pedestrian i; g() is an indicator function. k Body compressive stiffness coefficient; obstacle repulsion force The formula is: ,in, The distance from the pedestrian to the nearest obstacle. Let be the unit vector pointing from the obstacle to the pedestrian; And calculate the pedestrian's velocity and position update based on the equation of motion, including: Using each simulation time step Inside, the velocity is updated using the Euler numerical iteration formula of Newton's laws of motion as follows: The position is updated using the Euler numerical integral formula: ,in, It is the velocity vector of the i-th individual at time t. It is the net external force acting on the individual. It is individual quality. It is the position vector of the i-th individual at time t. It is the time step of numerical integration, and the acceleration generated by the net external force. Multiply by the time step to obtain the velocity increment within one time step, thus completing the iterative update of velocity and position.

[0016] A further technical solution of the present invention is: the formula for calculating the Total Evacuation Completion Time (TET) is: ,in, The evacuation trigger time, Let p be the time when passenger p arrives at the safety exit; For all passengers The moment when they finally reach the safe exit.

[0017] According to a second aspect of the present disclosure, a system for predicting and optimizing emergency evacuation of personnel during subway renovation construction is provided, the system comprising: The spatial network modeling module is used to obtain spatial layout information of each construction stage of the subway underground station group renovation project, and to construct a directed graph network model based on the boundary information of the spatial layout. The module for establishing a semantic rule constraint system includes setting two types of triggering conditions for influx pulses: one is a regular passenger flow pulse, and the other is an emergency evacuation pulse. The pedestrian congestion wave prediction and quantitative analysis module is used to identify the queuing state evolution pattern caused by limited exit capacity through the four-stage semantic state transition of S-FSM. It performs quantitative analysis on the two shock waves of the influx pulse. For the bottleneck edge with limited capacity, it calculates the queue formation time and the number of people in the queue based on the number of people entering and the actual capacity of the bottleneck. It calculates the number of people staying based on the integral of the flow rate of people entering the waiting area from the queue formation to the stage release trigger time. It calculates the duration of the release process and the exit evacuation rate based on the ratio of the number of people staying to the actual capacity of the exit. It outputs the spatiotemporal distribution prediction results of the congestion wave driven by S-FSM semantic state transition. The four-stage semantic state of S-FSM includes the initial response and waiting stage, the approach and queuing stage, the passage release and passage stage, and the arrival at the safe area stage. The multi-exit dynamic path allocation optimization module is used to determine all available evacuation exits and their actual capacity under the current working conditions, as well as the total number of people to be evacuated, based on the prediction results of the spatiotemporal distribution of congestion waves. It uses the maximum evacuation completion time of each exit as the optimization variable, establishes equality constraints, and constructs a load balancing optimization model. The optimal solution satisfies that the ratio of the number of people allocated to each exit to its capacity is equal. The passenger compliance rate parameter is introduced, and the corrected actual number of people allocated to each exit is the weighted sum of the number of people allocated by the optimized scheme and the number of people allocated by inertial behavior. The microscopic pedestrian simulation module based on the social force model is used to simulate and extrapolate the evacuation process by assigning pedestrians self-driving force, interaction force between pedestrians, and interaction force between pedestrians and obstacles based on spatial layout information and directed graph network model. The total evacuation completion time assessment and plan output module is used to calculate the total evacuation completion time as the evaluation index, based on the difference between the time when the last passenger arrives at the safe exit and the evacuation trigger time. If the time exceeds the safety standard, the module returns to the multi-exit dynamic path allocation optimization stage for re-optimization. The deviation between the predicted congestion parameters and the actual simulation values ​​is fed back to the parameter library for subsequent stage accuracy optimization. When switching construction stages, the closed area weights and temporary edges of the directed graph network model are updated. The semantic rule constraint system stage is repeatedly established until the total evacuation completion time assessment and plan output stage, forming an evacuation plan system covering the entire renovation cycle.

[0018] According to a third aspect of the present disclosure, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method for predicting and optimizing emergency evacuation of personnel during the subway renovation and construction phase.

[0019] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein computer instructions are stored on the storage medium, and when executed by a processor, the instructions implement the steps of the above-described method for predicting and optimizing emergency evacuation of personnel during the subway renovation and construction phase.

[0020] This invention provides a method, system, and medium for predicting and optimizing emergency evacuation of personnel during subway renovation construction. It characterizes the dynamic evolution of spatial topology through phased semantic network modeling, constructs a spatiotemporal semantic collision analysis framework for influx pulses and phased releases using known passenger flow timetables, quantitatively predicts the spatiotemporal distribution of congestion waves driven by a four-stage, ten-state semantic finite state machine (S-FSM), and uses this to drive the coordinated optimization of multi-exit path equitable allocation and information dissemination guidance strategies. Its beneficial effects include: (1) This invention achieves phased dynamic modeling and adaptive switching of evacuation networks in construction and renovation scenarios. Addressing the dynamic evolution of spatial topology during construction phases, this invention uses the 'construction phase' as an external driving variable throughout the entire evacuation simulation process—the spatial layout data of each construction phase drives the construction of the corresponding network model. During phase switching, only the passage weights of the network edges involved in the closed area need to be adjusted to achieve rapid model switching, without the need for a complete reconstruction of the network model. This method enables the system architecture to iterate in stages following the construction progress, balancing multi-stage continuous simulation capabilities with computational efficiency, ultimately forming a hierarchical emergency evacuation plan system covering the entire construction and renovation cycle.

[0021] (2) A method for predicting the spatiotemporal distribution of congestion waves for construction and renovation scenarios was established. This invention abstracts the flow of people triggered by sudden events into inflow pulses and the batch opening and closing constraints of exits / passages into phased releases, and establishes a spatiotemporal coupling calculation model of the two. This enables quantitative extrapolation of the generation time, spatial location and congestion intensity of congestion waves during evacuation, providing input basis for optimizing the load balancing of exits and realizing a leap from "post-event response" to "prediction-proactive intervention".

[0022] (3) A "prediction-optimization-evaluation" intelligent closed-loop evacuation decision-making framework was constructed. This invention uses the congestion prediction result as the optimization input, and solves the optimal route allocation scheme with the load balancing of each exit as the optimization objective. Passenger compliance rate is introduced to correct the actual allocation effect, and a feedback closed loop is formed through TET simulation evaluation. This method does not rely on real-time sensor data, and can complete modeling and calculation based only on the construction organization plan and passenger flow schedule. It has strong engineering operability, wide applicability, and potential for migration and application to other types of transportation hubs.

[0023] (4) It has good versatility and scalability. The method framework of this invention is not limited to a specific subway underground station group structure or construction plan. By inputting the semantic topology model, passenger flow parameters and bottleneck parameters of each stage, it can be adapted to different engineering scenarios. The supporting Python hybrid simulation engine has been verified in the case of a certain airport T2 terminal and has the potential to be migrated to other types of transportation hubs (high-speed rail stations, subway transfer stations, large sports stadiums, etc.).

[0024] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0025] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0026] Figure 1 This is a flowchart of the method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase in this embodiment of the invention; Figure 2 This is a schematic diagram of the four stages and ten states of a pedestrian's entire life cycle in an embodiment of the present invention; Figure 3 This is a schematic diagram of the system structure for predicting and optimizing emergency evacuation of personnel during the subway renovation and construction phase in an embodiment of the present invention; Figure 4 This is a schematic diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0027] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the present invention are shown in the drawings, not the entire structure.

[0028] Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of these steps can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the steps can be rearranged. The process can be terminated when its operation is complete, but may also have additional steps not included in the figures. The process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0029] This invention aims to solve the technical problem that existing evacuation simulation methods during the renovation and construction of underground subway stations lack the ability to predict congestion and cannot achieve proactive intervention, and overcome the limitations of the traditional "post-event response" evacuation decision-making model in dynamic renovation environments and pulse surge passenger flow conditions.

[0030] To address the challenges of evacuation during construction projects caused by phased changes in spatial topology and the overlap of sudden passenger flows, this invention constructs spatial network models for each construction phase to characterize the evolution of spatial constraints such as construction barriers, temporary passages, and enclosed areas. Based on these models, a pedestrian evacuation behavior model is established. Utilizing known spatial layout parameters and passenger flow temporal distribution patterns from each construction phase, the location and timing of congestion waves during evacuation are predicted. With exit load balancing as the optimization objective, a multi-exit dynamic path allocation strategy is used to proactively intervene in congestion risks. A passenger compliance rate parameter is introduced to correct the actual evacuation effect, and the total evacuation completion time is used as the evaluation index, forming an intelligent closed-loop evacuation decision-making process of "prediction—optimization—evaluation."

[0031] This invention aims to provide an intelligent evacuation simulation method and supporting hybrid simulation platform with congestion prediction capabilities, proactive intervention capabilities, and engineering applicability, providing scientific methods and decision-making tools for the intelligent formulation and safety assessment of passenger evacuation plans during the renovation and construction of subway underground station groups.

[0032] In summary, this invention belongs to the fields of disaster prevention and mitigation in underground engineering, pedestrian evacuation simulation, and semantic digital twin technology. Specifically, it is an intelligent emergency evacuation method and system based on the semantic digital twin (SDT) framework for use during the renovation and construction of subway underground station groups. It is applicable to the intelligent formulation of evacuation plans, the prediction of risks of sudden surges in passenger flow, and the optimization of dynamic guidance strategies for large subway underground station groups with multiple subway lines intersecting under the special working conditions of "operation and renovation at the same time".

[0033] like Figure 1 As shown in the example, this is a method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase, including the following steps: S1. Spatial Network Modeling Stage: Obtain spatial layout information for each construction stage of the subway underground station group renovation project, and construct a directed graph network model based on the boundary information of the spatial layout; S2. Establishing a semantic rule constraint system: This includes setting two types of triggering conditions for influx pulses: one is a regular passenger flow pulse, and the other is an emergency evacuation pulse. S3. Pedestrian Congestion Wave Prediction and Quantitative Analysis Stage: The evolution of queuing states caused by limited exit capacity is identified through the four-stage semantic state transition of S-FSM. The two shock waves of the influx pulse are quantitatively analyzed. For the bottleneck edge with limited capacity, the queue formation time and queue number are calculated based on the number of people entering and the actual capacity of the bottleneck. The number of people staying is calculated based on the integral of the flow rate of people entering the waiting area from the formation of the queue to the trigger time of the phased release. The duration of the release process and the exit evacuation rate are calculated based on the ratio of the number of people staying to the actual capacity of the exit. The spatiotemporal distribution prediction results of the congestion wave driven by S-FSM semantic state transition are output. The four-stage semantic state of S-FSM includes the initial response and waiting stage, the approach and queuing stage, the passage release and passage stage, and the arrival at the safe area stage. S4. Multi-exit dynamic path allocation optimization stage: Based on the spatiotemporal distribution prediction results of congestion waves, the maximum evacuation completion time of each exit is minimized as the optimization variable. All available evacuation exits and their actual capacity under the current working conditions are determined, as well as the total number of people to be evacuated. The number of people allocated to each exit is used as the optimization variable. Equality constraints are established, and a load balancing optimization model is constructed. The optimal solution satisfies that the ratio of the number of people allocated to each exit to its capacity is equal. The passenger compliance rate parameter is introduced. The corrected actual number of people allocated to each exit is the weighted sum of the number of people allocated by the optimized scheme and the number of people allocated by inertial behavior. S5. Microscopic pedestrian simulation stage based on social force model: Based on spatial layout information and directed graph network model, pedestrians are made to conform to social force model by giving them self-driving force, interaction force between pedestrians, and interaction force between pedestrians and obstacles. The speed and position update of pedestrians are calculated according to the motion equation to realize the simulation and deduction of evacuation process. S6. Total Evacuation Completion Time Assessment and Solution Output Stage: The total evacuation completion time is used as the evaluation index, calculated as the difference between the time when the last passenger arrives at the safe exit and the evacuation trigger time; if it exceeds the safety standard, it returns to the multi-exit dynamic path allocation optimization stage for re-optimization; the deviation between the predicted congestion parameters and the actual simulation values ​​is fed back to the parameter library for subsequent stage accuracy optimization; when switching construction stages, the closed area weights and temporary edges of the directed graph network model are updated, and the semantic rule constraint system stage is repeatedly established until the total evacuation completion time assessment and solution output stage, forming an evacuation plan system covering the entire renovation cycle.

[0034] In the specific implementation of S1, spatial layout information of each construction stage of the subway underground station group renovation project is obtained, including the location and size of facilities such as platforms, entrances and exits, ticket gates, waiting areas, and passages, as well as renovation boundary information such as construction fences, temporary partitions, and closed areas. A directed graph network model is constructed based on the boundary information.

[0035] In the specific implementation of S2, two types of triggering conditions for influx pulses are clearly defined: the first is the regular passenger flow pulse, triggered by timetable-driven events such as train arrivals and departures, and transfer connections; the second is the emergency evacuation pulse, driven by mandatory evacuation commands triggered by sudden events such as fires, structural instability, and equipment failures. Both types of pulses are characterized by a certain number of pedestrians entering the target passage or bottleneck area in a short period of time, and a unified influx pulse representation method is adopted in the model. The exits / passages are switched from an impassable state to an accessible state through phased release. In the station renovation scenario, some exits are temporarily closed due to construction and will be opened in batches after the evacuation command is issued or the construction barriers are removed. A four-stage, ten-state semantic finite state machine (S-FSM) full lifecycle stage transition mechanism for individual pedestrians during the evacuation process is also established.

[0036] Specifically, an influx pulse refers to a time-varying passenger flow input caused by concentrated train arrivals, passenger transfers, or emergency evacuation commands, resulting in a certain number of pedestrians concentrating in a target passage or bottleneck area within a short period of time. Emergency evacuation commands are triggered by sudden events such as fires, structural instability, or equipment failures, driving pedestrians to evacuate to exits / passages. Phased release refers to the phased activation events where exits / passages switch from an impassable state to a passable state. In station renovation scenarios, some exits may be temporarily closed due to construction and will be opened in batches after evacuation commands are issued or construction barriers are removed.

[0037] In the specific implementation of S3, such as Figure 2As shown, the S-FSM four-stage semantic state transition—from the initial response and waiting stage, through the approach and queuing stage, to the passage and passage stage, and finally to the safe zone stage—identifies the queuing state evolution pattern caused by limited exit capacity. It quantitatively analyzes the two-stage shock wave of the influx pulse. For the bottleneck edge with limited capacity, it calculates the queue formation time and queue length based on the number of people entering and the actual bottleneck capacity; calculates the number of people remaining based on the integral of the flow rate entering the waiting area from queue formation to the staged release trigger time; and calculates the duration and peak flow of the release wave based on the ratio of the number of people remaining to the actual exit capacity. The result is a prediction of the spatiotemporal distribution of congestion waves driven by S-FSM semantic state transition. Staged release refers to the batch activation events where an exit or passage changes from an impassable state to a passable state.

[0038] In the pedestrian congestion wave prediction quantitative analysis stage, the conditions for the semantic state transition in the four stages of S-FSM are: When pedestrians i If the pedestrian's current location is not within the safe zone or waiting area node, and the safety exit has not yet been opened, i Currently in the initial response and waiting phase; When pedestrians i When the pedestrian's current location is in a safe zone or waiting area node, and the safety exit has been partially opened, i In the approaching and queuing stage; When pedestrians i The current location is within the queuing buffer and the target exit is the currently active exit. Pedestrian i It is currently in the stage of allowing passage and clearing. When pedestrian i's current position is at the target exit position, pedestrian i is in the stage of reaching the safe area.

[0039] In a preferred embodiment, the Four-Phase Ten-State Semantic Finite State Machine (S-FSM) controls pedestrian movement decisions in four phases: Initial Response and Waiting Phase: Pedestrians start from their initial position and move along the shortest path to the designated safe area / waiting area node, or wait in the safe area because the exit has not yet been released, corresponding to states 1 to 3 (movement response, in-place waiting, periphery waiting) in the S-FSM mechanism; Approach and Queuing Phase: Pedestrians trigger an approach intention and approach the evacuation exit bottleneck from the safe area / anchor area, queuing due to limited exit capacity, possibly experiencing queue backtracking, first-level buffering, etc. The states of queuing, secondary buffer queuing, or overflow backlog queuing correspond to states 4 to 8 in the S-FSM mechanism (heading to exit, queue back, primary buffer queuing, secondary buffer queuing, overflow backlog). The passage and clearance phase: After the pedestrian's target exit is activated through phased release, the pedestrian switches to the state of moving towards the target exit, proceeding along the shortest path to the final destination until evacuation is complete, corresponding to state 9 (passage cleared) in the S-FSM mechanism. The safe zone arrival phase: The pedestrian reaches the final safe zone (less than a set threshold from the exit), corresponding to state 10 (reached the safe zone). When a pedestrian is unable to move for an extended period due to extreme deadlock and cannot recover through rerouting, the simulation engine calculates and clears this state, marking it as an abnormal termination, and it is not counted as a successful evacuation. The conditions for pedestrian state transitions are:

[0040] In the formula: P i Let i be the current position of pedestrian i. Location of safe zone / waiting area nodes. Let i be the target exit location. For pedestrian i, the target exit. This is the set of currently activated exits.

[0041] Specifically, the quantitative analysis stage for predicting pedestrian congestion waves: Queue formation time With the number of people in the queue The expression is: , ,in, t This refers to the moment when the inrush pulse reaches the upstream of the bottleneck. q For the number of people injected; The actual traffic capacity at the bottleneck; τ This represents the average travel time for pedestrians from the influx source to the bottleneck. q> At that time, a queue is formed; Calculate the number of residents The expression is: ,in, For the time when the queue is formed, The triggering time for the inrush pulse; The flow rate of people entering the waiting area is expressed in people / second, and the flow rate of people leaving before the phased release trigger is 0. Duration of release process With exit evacuation flow rate The specific expression is: , ,in, To stimulate exports The actual traffic capacity, when Exceeding exports Actual traffic capacity When the product of the waiting time and the waiting time is multiplied, secondary congestion will occur at the exit.

[0042] In the specific implementation of S4, based on the spatiotemporal distribution of congestion waves predicted in S3, minimizing the maximum evacuation completion time of each exit is used as the optimization variable. All available evacuation exits and their actual capacity, as well as the total number of people to be evacuated, are determined under the current operating conditions. Using the number of people allocated to each exit as the optimization variable, equality constraints are established, and a load balancing optimization model is constructed. The optimal solution satisfies that the ratio of the number of people allocated to each exit to its capacity is equal. A passenger compliance rate parameter is introduced, and the corrected actual number of people allocated to each exit is the weighted sum of the number allocated by the optimized scheme and the number allocated by inertial behavior.

[0043] Specifically, the load balancing optimization model includes: Let the set of available exits be O = {o1, o2, ..., o...} m The actual throughput capacity of each exit is: j=1, ..., m, where m is the number of exits; pedestrians are divided into R starting groups according to their injection source, and the number of people to be evacuated in the r-th starting group is... r=1, ...,R, satisfying , Let r be the total number of people to be evacuated from the starting group, and let r be the set of exits that can be reached by the pedestrians in the r-th group. This set is determined by the spatial layout and construction closure status; the reachable exit sets may differ for different starting point groups. Let the number of people in group r assigned to exit j be the optimization variable, satisfying the following condition: (All pedestrians in each group were evacuated.) ≥0, and the actual number of people allocated to each export Limited by its service capacity, it can meet the needs of customers. , Target evacuation time window The optimization objective is to minimize the maximum estimated travel time at each exit. The optimal solution satisfies the condition that the ratio of the total number of people allocated to each exit to its travel capacity is equal. ; Optimal allocation of personnel to each export Calculate as follows: ; The obtained optimal route allocation scheme is transformed into an information dissemination and guidance strategy, introducing a passenger compliance rate parameter λ, 0≤λ≤1, which characterizes the actual response rate of passengers to dynamic guidance information. The value of λ is determined based on actual operational statistics of similar hubs or on-site test calibration. λ=1 indicates complete compliance, and λ=0 indicates complete non-compliance. The proportion of passengers who comply with the guidance is... Evacuation will be carried out according to the optimized plan; the proportion of passengers who do not follow the guidance is [missing information]. Evacuation will be carried out according to the inertial path selection behavior; Revised actual number of personnel allocated to each export for: ,in To optimize the allocation of personnel at exit j in the plan, The number of people allocated to exit j under inertial behavior.

[0044] In the specific implementation of S5, on the existing spatial model, pedestrians are made to conform to the social force model by being given self-driving force, interaction force between pedestrians, and interaction force between pedestrians and obstacles. The speed and position of pedestrians are updated according to the motion equation to realize the simulation and deduction of the evacuation process.

[0045] Furthermore, the social force model specifically includes: The motion of pedestrian i is described by Newton's second law: ,in, Let i be the mass of pedestrian i; It is the velocity vector; Self-driving force; Repulsive force between pedestrians; Repulsive force from obstacles; self-driving force The formula is: ,in, For the desired velocity scalar, Let be the unit vector pointing to the target location. The current velocity vector, Relaxation time; interpersonal repulsion The formula is: ,in and The interaction strength parameter; Let i be the radius of pedestrian i; Let be the distance between pedestrian i and pedestrian j; Let g be the unit vector pointing from pedestrian j to pedestrian i; g() is the indicator function (takes the independent variable if it is greater than 0, otherwise takes 0). kBody compressive stiffness coefficient; obstacle repulsion force The formula is: ,in, The distance from the pedestrian to the nearest obstacle. Let be the unit vector pointing from the obstacle to the pedestrian; And calculate the pedestrian's velocity and position update based on the equation of motion, including: Using each simulation time step Inside, the velocity is updated using the Euler numerical iteration formula of Newton's laws of motion as follows: The position is updated using the Euler numerical integral formula: ,in, It is the velocity vector of the i-th individual at time t. It is the net external force acting on the individual. It is individual quality. It is the position vector of the i-th individual at time t. It is the time step of numerical integration, and the acceleration generated by the net external force. Multiply by the time step to obtain the velocity increment within one time step, thus completing the iterative update of velocity and position.

[0046] In the specific implementation of S6, the total evacuation completion time (TET) is used as the evaluation index, calculated as the difference between the time the last passenger arrives at the safe exit and the evacuation trigger time. If it exceeds the safety standard, it returns to S4 for re-optimization. The deviation between the predicted congestion parameters and the actual simulation values ​​is fed back to the parameter library for subsequent stage accuracy optimization. When switching construction stages, the closed area weights and temporary edges of the network model are updated, and S2 to S6 are repeated to form an evacuation plan system covering the entire renovation cycle, that is, to achieve intelligent evacuation decision-making throughout the entire process of "spatial modeling—S-FSM state modeling—congestion prediction—path optimization—simulation evaluation—stage iteration". Specifically, the formula for calculating the total evacuation completion time (TET) is: ,in, The evacuation trigger time, Let p be the time when passenger p arrives at the safety exit; For all passengers The moment when they finally reach the safe exit.

[0047] Another embodiment provides a system 300 for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase, comprising: The spatial network modeling module 310 is used to obtain spatial layout information of each construction stage of the subway underground station group renovation project, and to construct a directed graph network model based on the boundary information of the spatial layout. The semantic rule constraint system module 320 includes setting two types of triggering conditions for influx pulses: one is the regular passenger flow pulse, and the other is the emergency evacuation pulse. The pedestrian congestion wave prediction and quantitative analysis module 330 is used to identify the queuing state evolution pattern caused by limited exit capacity through the four-stage semantic state transition of S-FSM. It performs quantitative analysis on the two shock waves of the influx pulse. For the bottleneck edge with limited capacity, it calculates the queue formation time and the number of people in the queue based on the number of people entering the waiting area and the actual capacity of the bottleneck. It calculates the number of people staying based on the integral of the flow rate of people entering the waiting area from the queue formation to the stage release trigger time. It calculates the duration of the release process and the exit evacuation rate based on the ratio of the number of people staying to the actual capacity of the exit. It outputs the spatiotemporal distribution prediction results of the congestion wave driven by S-FSM semantic state transition. The four-stage semantic state of S-FSM includes the initial response and waiting stage, the approach and queuing stage, the passage release and passage stage, and the arrival at the safe area stage. The multi-exit dynamic path allocation optimization module 340 is used to determine all available evacuation exits and their actual capacity under the current working conditions, as well as the total number of people to be evacuated, based on the prediction results of the spatiotemporal distribution of congestion waves, minimizing the maximum evacuation completion time of each exit as the optimization variable, using the number of people allocated to each exit as the optimization variable, establishing equality constraints, constructing a load balancing optimization model, and the optimal solution satisfies that the ratio of the number of people allocated to each exit to its capacity is equal. The passenger compliance rate parameter is introduced, and the corrected actual number of people allocated to each exit is the weighted sum of the number of people allocated by the optimized scheme and the number of people allocated by inertial behavior. The microscopic pedestrian simulation module 350 based on the social force model is used to simulate and extrapolate the evacuation process by assigning pedestrians self-driving force, interaction force between pedestrians, and interaction force between pedestrians and obstacles based on spatial layout information and directed graph network model. The Total Evacuation Completion Time Assessment and Solution Output Module 360 ​​uses the total evacuation completion time as the evaluation index, calculated based on the difference between the time the last passenger arrives at the safe exit and the evacuation trigger time. If the time exceeds the safety standard, it returns to the multi-exit dynamic path allocation optimization stage for re-optimization. The deviation between the predicted congestion parameters and the actual simulation values ​​is fed back to the parameter library for subsequent stage accuracy optimization. When switching construction stages, the closed area weights and temporary edges of the directed graph network model are updated, and the semantic rule constraint system stage is repeatedly established until the total evacuation completion time assessment and solution output stage, forming an evacuation plan system covering the entire renovation cycle.

[0048] In addition to the modules mentioned above, the system 300 for predicting and optimizing emergency evacuation of personnel during the subway renovation and construction phase may also include other components. However, since these components are not related to the content of this disclosure, their illustrations and descriptions are omitted here.

[0049] Other specific working processes of the method for predicting and optimizing emergency evacuation of personnel during the subway renovation and construction phase using the aforementioned system 300 for predicting and optimizing emergency evacuation of personnel during the subway renovation and construction phase are described in the above-mentioned embodiment of the method for predicting and optimizing emergency evacuation of personnel during the subway renovation and construction phase, and will not be repeated here.

[0050] Another embodiment illustrates that the system of the present invention can also be achieved by means of... Figure 4 The architecture of the computing device shown is used to implement this. Figure 4 The architecture of the computing device is shown. For example... Figure 4 As shown, it includes a computer system 410, a system bus 430, one or more CPUs 440, input / output 420, and a memory 450. The memory 450 can store various data or files used by the computer for processing and / or communication, as well as program instructions executed by the CPU, including methods for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase. Figure 4 The architecture shown is merely exemplary and should be adjusted according to actual needs when implementing different devices. Figure 4 One or more components are included. The memory 450, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the method for predicting and optimizing personnel emergency evacuation during the subway renovation construction phase described above in this embodiment of the invention. One or more CPUs 440 execute various functional applications and data processing of the system of the present invention by running the software programs, instructions, and modules stored in the memory 450.

[0051] Of course, the server provided in the embodiments of the present invention is not limited to executing the method operations described above, but can also execute related operations in the method for predicting and optimizing personnel emergency evacuation during the subway renovation and construction phase provided in any embodiment of the present invention.

[0052] The memory 450 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on terminal usage. Furthermore, the memory 450 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 450 may further include memory remotely configured relative to one or more CPUs 440, these remote memories being connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0053] Input / output 420 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the device. Input / output 420 may also include a display device such as a display screen.

[0054] This invention also provides a non-transitory computer-readable storage medium storing a computer program. When executed by a processor, this computer program implements the method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase described in the above embodiments. The computer-readable storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0055] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0056] The program code contained on the storage medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0057] Furthermore, other specific working processes of a non-temporary computer-readable storage medium are described in the above-described embodiment of the method for predicting and optimizing personnel emergency evacuation during the subway renovation and construction phase, and will not be repeated here.

[0058] To verify the effectiveness of the method of this invention, the following example, based on an airport renovation and expansion project, provides an implementation example of an intelligent evacuation method for the construction scenario of subway underground station group renovation. This embodiment also demonstrates the dual function of the accompanying Python hybrid simulation engine as both a method implementation carrier and a verification tool.

[0059] (1) Project Overview This embodiment is based on an airport expansion project. Terminal 2 is located in the northern part of the airport, between runways 1 and 2, with a total building area of ​​approximately 400,000 square meters and a designed annual passenger throughput of 31 million. It connects to multiple rail transit lines, including Intercity Line 1, Metro Line 11, Metro Line 20, and Intercity Line 2. This embodiment selects the connection area between Terminal 2 and the rail transit transfer hub as the modeling object.

[0060] Based on the construction organization plan for the expansion project, this embodiment uses the baseline condition of all lines operating simultaneously without closed construction as the initial scenario for evacuation simulation. The semantic topology update in subsequent construction phases can be extended and applied through the phased modeling method of this invention. The evacuation scenario is set as follows: an emergency occurs in the area connecting Terminal 2 and the subway transfer hub, triggering a mandatory evacuation order, requiring the emergency evacuation of all passengers in the area.

[0061] (2) Constraint-aware semantic topology modeling and bottleneck identification Based on the CAD floor plan of Terminal 2 and Tiled software, a constraint-aware semantic graph network model G(V,E) was constructed. The spatial network covers three rail transit lines—Metro Line 11 (11 tracks), Metro Line 20 (10 tracks), and Intercity Railway 1 (9 tracks), totaling 30 tracks.

[0062] Node types include: Injection source nodes (Start_Line11_1~7, Start_Line20_1~7, Start_LineSGS_1~8, a total of 22): correspond to the passenger disembarkation positions after the train arrives at the station on each line.

[0063] Gate nodes (Path_Gate_Line11_1~10, Path_Gate_Line20_1~10, Path_Gate_LineSGS_1~5, a total of 25): correspond to the ticket / security checkpoints for each line leading to the main terminal hall.

[0064] Zone nodes (Zone_Gate_Line11_1~10, Zone_Gate_Line20_1~10, Zone_Gate_LineSGS_1~5, a total of 25): correspond to the waiting buffer zone in front of the gate.

[0065] Endpoint nodes (End_Line11_1 / 2, End_Line20_1, End_LineSGS_1 / 2, a total of 5): correspond to safe evacuation exits.

[0066] Edges are automatically generated through LOS visibility detection between nodes and weighted using Euclidean distance. The wall layer is imported based on the wall and fence boundary lines from the CAD plan.

[0067] Static bottlenecks were identified as follows: ① The Line11 area gate nodes Path_Gate_Line11_1~10 converged passenger flow from 7 injection sources, making it the area with the highest passenger flow density among the three lines; ② The LineSGS area gate nodes Path_Gate_LineSGS_1~5 had relatively limited channel width, restricting their passage capacity. These bottlenecks were then input into the prediction module.

[0068] (3) Semantic rule constraints and S-FSM state modeling Influx Pulse Modeling: The sudden event is triggered at the simulation start time t0=1.0s. Based on the example calculation scale, the scenario is set / scaled proportionally to the predicted passenger flow. Simulation of simultaneous forced evacuation of three lines is performed, with the injection pulse parameters as follows: Line 11: Inject 100 people from Start_Line11_1 to 7, with the target exit being End_Line11_1 / 2 (randomly assigned). Line 20: Inject 100 people from Start_Line20_1 to 7, with the target exit being End_Line20_1. LineSGS: Inject 100 people from Start_LineSGS_1 to 8, with the target exit being End_LineSGS_1 / 2 (randomly assigned). A total of 300 pedestrians need to be evacuated. The expected speed of the pedestrians is uniformly and randomly distributed from 1.2 to 1.8 m / s to simulate the heterogeneity of the crowd, with an average speed of 1.5 m / s.

[0069] Phased release modeling: In this embodiment, all gate nodes (Path_Gate_Line11_1~10, Path_Gate_Line20_1~10, Path_Gate_LineSGS_1~5) are opened after the simulation starts to simulate the simultaneous passage of all lines.

[0070] S-FSM State Initialization: The initial state of each pedestrian instance is set to "Movement Response (State 1)". The path planning algorithm (A*) calculates the complete path from the injection source through the gate node to the destination. The intermediate nodes of the path determine the state transition in sequence according to the four stages of S-FSM.

[0071] (4) Congestion wave prediction based on S-FSM semantic state transition Based on the bottleneck Line11 gate nodes Path_Gate_Line11_1~10 (total throughput capacity) (≈2.5 people / s) and LineSGS gate nodes Path_Gate_LineSGS_1~5 (total throughput capacity) The actual throughput capacity is approximately 1.5 people / second, which is predicted based on the injected pulse parameters. Queue formation time calculation: In Line 11 area, 100 people converge from 7 injection sources to the gate bottleneck via various paths, with an average passage time τ≈30s. (Based on the formula...) Calculation yields t f It takes approximately 100 seconds for a queue to form in front of the turnstiles. In Line 20, 100 people are diverted through 10 turnstiles, and the load on each turnstile is relatively even, so the queue is expected to form relatively late.

[0072] Calculation of the number of residents: by formula Calculations show that the number of people queuing in front of the Line 11 turnstile peaks at t=180s, at which time the number of people queuing is approximately 45.

[0073] Release wave duration and peak flow rate calculation: by formula , Calculate the duration t of the release wave from the Line11 turnstile. r ≈220s, peak flow .

[0074] The prediction result W is output to the optimization module.

[0075] (5) Optimization of dynamic path allocation for multiple exits and correction of compliance rate The available exit set is O={End_Line11_1, End_Line11_2, End_Line20_1, End_LineSGS_1, End_LineSGS_2}, with actual passage capacities of C1=1.2 people / s, C2=1.2 people / s, C3=1.5 people / s, C4=1.0 people / s, and C5=1.0 people / s, totaling 5.9 people / s.

[0076] Total number of people to be evacuated Q total =300 people. (From the formula) Calculate the optimal number of people allocated to each exit: ≈61 people ≈61 people ≈76 people ≈51 people ≈51 people.

[0077] Based on inertia, pedestrians choose exits according to a weighted distance when there is no guidance information. Exits 1 and 2 of Metro Line 11, located near densely populated areas, each attract 80 people; Exit 1 of Metro Line 20, in the center, attracts 60 people; and Exits 1 and 2 of Intercity Railway 1, located in more remote areas, each attract 40 people. =80 people =80 people =60 people =40 people =40 people.

[0078] The passenger compliance rate λ is set at 0.7, which physically means that 70% of passengers evacuate according to the recommended route after receiving dynamic guidance information, while the remaining 30% still follow their habitual route selection behavior. This value is based on operational experience and literature from similar hubs. (From the formula...) The actual number of personnel allocated to each export after correction: 1 ≈ 67 people 2 ≈ 67 people 3 ≈ 71 people 4 ≈ 48 people 5 ≈ 48 people.

[0079] The revised route allocation scheme will be transformed into an information dissemination and guidance strategy: the terminal's public address system will announce the directions of each exit, and electronic displays will show the real-time traffic status of each exit.

[0080] (6) Hybrid simulation engine micro-deduction and TET evaluation The corrected allocation results are input into a Python hybrid simulation engine for microscopic inference. The simulation runs the social force model on the semantic topological model G, with a time step d. t =0.1s, physical substep 0.025s. Pedestrian movement is driven by the combined force of self-driving force, repulsive force between pedestrians, and repulsive force from obstacles.

[0081] The simulation records the queue length at each bottleneck node, the flow rate at each exit, and the arrival time (tp) of each passenger at the destination. The simulation calculates the total evacuation completion time (TET) to be approximately 220 seconds. The simulation also outputs pedestrian trajectory data (evacuation_data_log.csv) at each time step, containing the location coordinates and status of each pedestrian at each time step, for posterior analysis and visualization verification.

[0082] (7) Solution output and stage iteration The optimal solution is output as a structured evacuation plan. During the construction phase transition, the weights of the closed areas and temporary edges of the semantic network model are updated, and the above steps are repeated to form an evacuation plan system covering the entire lifecycle of the T2 terminal area expansion.

[0083] This embodiment fully demonstrates the application of the present invention in the expansion project of Terminal 2 of an airport: from importing CAD floor plans, constraint-aware semantic topology modeling, S-FSM four-stage ten-state initialization, influx pulse and phased release configuration, congestion wave prediction, multi-exit load balancing optimization, to micro-deduction and TET evaluation using a hybrid simulation engine. The accompanying Python hybrid simulation engine completed the entire calculation process from spatial modeling and pedestrian motion simulation to trajectory data output, verifying the effectiveness and engineering feasibility of this method in a real large-scale hub project.

[0084] Based on the above embodiments and verification examples, (1) this invention proposes a phased dynamic network modeling and construction phase adaptive update method for the characteristics of spatial topology dynamic evolution under the special working condition of "operation and renovation at the same time" during the renovation and construction of subway underground station groups. The method pre-establishes spatial network models corresponding to each construction phase according to the construction organization plan, and accurately represents the phased changes of spatial constraints such as construction enclosures, temporary passages, and closed areas with a three-layer structure of nodes-edges-regions; when the construction phase is switched, the effective switching of different phase models is achieved by adjusting the passage weight of the network edges involved in the closed area, without the need to rebuild the network model as a whole, taking into account the multi-stage continuous simulation capability and computational efficiency, forming a hierarchical emergency evacuation plan system covering the entire renovation and construction cycle, and making up for the shortcomings of the existing static evacuation simulation technology in adapting to the dynamic spatial evolution of construction. (2) This invention proposes a method for predicting the spatiotemporal distribution of congestion waves in construction and renovation scenarios. By abstracting the flow of people triggered by sudden events into inflow pulses and the batch opening and closing constraints of exits / passages into phased releases, a spatiotemporal coupling calculation model of the two is established to realize the quantitative prediction of the generation time, spatial location and congestion intensity of congestion waves during evacuation. Based on the bottleneck capacity and pulse parameters, this prediction method quantitatively calculates the queue formation time, queue length, number of people staying and peak flow of release waves, and outputs the spatiotemporal distribution prediction results of congestion waves, providing input basis for subsequent exit load balancing optimization, and realizing the transformation from traditional passive post-event handling to risk prediction. (3) This invention constructs a set of "prediction-optimization-evaluation" intelligent closed-loop evacuation decision-making method. Based on the congestion wave prediction results, it drives the dynamic path allocation optimization of multiple exits, solves the optimal allocation scheme with the load balancing of each exit as the optimization objective, introduces the passenger compliance rate parameter to correct the actual allocation effect, and uses the total evacuation completion time as the evaluation index to feed the simulation evaluation results back to the optimization module to form a closed-loop iteration. When the construction phase changes, the method returns to the spatial network modeling step to update the model and repeats the entire process until it covers all stages of the renovation construction. In the end, it forms a hierarchical emergency evacuation plan system that covers the entire renovation cycle, providing decision support for the formulation of passenger evacuation plans and safety assessments during the renovation construction of the station group.

[0085] In this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a step or method that comprises a list of elements includes not only those elements but also other elements not expressly listed or inherent to such a step or method.

[0086] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A method for predicting and optimizing emergency evacuation of personnel during the construction phase of subway renovation, characterized in that, Includes the following steps: Spatial network modeling stage: Obtain spatial layout information of each construction stage of the subway underground station group renovation project, and construct a directed graph network model based on the boundary information of the spatial layout; The stage of establishing a semantic rule constraint system includes setting two types of triggering conditions for influx pulses: one is the regular passenger flow pulse, and the other is the emergency evacuation pulse. Pedestrian congestion wave prediction quantitative analysis stage: The evolution pattern of queuing state caused by limited exit capacity is identified through the four-stage semantic state transition of S-FSM. The two shock waves of the influx pulse are quantitatively analyzed. For the bottleneck edge with limited capacity, the queue formation time and queue number are calculated based on the number of people entering and the actual capacity of the bottleneck. The number of people staying is calculated based on the integral of the flow rate of people entering the waiting area from the queue formation to the stage release trigger time. The duration of the release process and the exit evacuation rate are calculated based on the ratio of the number of people staying to the actual capacity of the exit. The spatiotemporal distribution prediction results of the congestion wave driven by S-FSM semantic state transition are output. The four-stage semantic state of S-FSM includes the initial response and waiting stage, the approach and queuing stage, the passage release and passage stage, and the arrival at the safe area stage. Multi-exit dynamic path allocation optimization stage: Based on the spatiotemporal distribution prediction results of congestion waves, the maximum evacuation completion time of each exit is minimized as the optimization variable. All available evacuation exits and their actual capacity under the current working conditions are determined, as well as the total number of people to be evacuated. The number of people allocated to each exit is used as the optimization variable. Equality constraints are established, and a load balancing optimization model is constructed. The optimal solution satisfies that the ratio of the number of people allocated to each exit to its capacity is equal. The passenger compliance rate parameter is introduced. The corrected actual number of people allocated to each exit is the weighted sum of the number of people allocated by the optimized scheme and the number of people allocated by inertial behavior. The microscopic pedestrian simulation stage based on the social force model: Based on spatial layout information and a directed graph network model, pedestrians are made to conform to the social force model by being given self-driving force, interaction force between pedestrians, and interaction force between pedestrians and obstacles. The speed and position of pedestrians are updated according to the motion equation to realize the simulation and deduction of the evacuation process. Total evacuation completion time assessment and plan output phase: The total evacuation completion time is used as the evaluation index, calculated as the difference between the time when the last passenger arrives at the safe exit and the evacuation trigger time; If the safety standard is exceeded, return to the multi-exit dynamic path allocation optimization stage for re-optimization; feed back the deviation between the predicted congestion parameters and the actual simulation values ​​to the parameter library for subsequent stage accuracy optimization; when switching construction stages, update the closed area weights and temporary edges of the directed graph network model, repeat the semantic rule constraint system stage until the total evacuation completion time assessment and scheme output stage, forming an evacuation plan system covering the entire transformation cycle.

2. The method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase according to claim 1, characterized in that, In the pedestrian congestion wave prediction and quantitative analysis phase, phased release refers to the batch activation events that switch the exit or passage from an impassable state to a passable state.

3. The method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase according to claim 1, characterized in that, In the pedestrian congestion wave prediction quantitative analysis stage, the conditions for the semantic state transition in the four stages of S-FSM are: When pedestrians i If the pedestrian's current location is not within the safe zone or waiting area node, and the safety exit has not yet been opened, i Currently in the initial response and waiting phase; When pedestrians i When the pedestrian's current location is in a safe zone or waiting area node, and the safety exit has been partially opened, i In the approaching and queuing stage; When pedestrians i The current location is within the queuing buffer and the target exit is the currently active exit. Pedestrian i It is currently in the stage of allowing passage and clearing. When pedestrian i's current position is at the target exit position, pedestrian i is in the stage of reaching the safe area.

4. The method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase according to claim 1, characterized in that, Pedestrian congestion wave prediction quantitative analysis stage: Queue formation time With the number of people in the queue The expression is: , ,in, t This refers to the moment when the inrush pulse reaches the upstream of the bottleneck. q For the number of people injected; The actual traffic capacity at the bottleneck; τ The average travel time for pedestrians from the influx source to the bottleneck; when q> At that time, a queue is formed; Calculate the number of residents The expression is: ,in, For the time when the queue is formed, The triggering time for the inrush pulse; The flow rate of people entering the waiting area is expressed in people / second, and the flow rate of people leaving before the phased release trigger is 0. Duration of release process With exit evacuation flow rate The specific expression is: , ,in, To stimulate exports The actual traffic capacity, when Exceeding exports Actual traffic capacity When the product of the waiting time and the waiting time is multiplied, secondary congestion will occur at the exit.

5. The method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase according to claim 1, characterized in that, The load balancing optimization model specifically includes: Let the set of available exits be O = {o1, o2, ..., o...} m The actual throughput capacity of each exit is: j=1, ..., m, where m is the number of exits; pedestrians are divided into R starting groups according to their injection source, and the number of people to be evacuated in the r-th starting group is... r=1, ...,R, satisfying , Let r be the total number of people to be evacuated from the starting group, and let r be the set of exits that can be reached by the pedestrians in the r-th group. ;set up Let the number of people in group r assigned to exit j be the optimization variable, satisfying the following condition: , ≥0, and the actual number of people allocated to each export Limited by its service capacity, it can meet the needs of customers. , For the target evacuation time window, The optimization objective is to minimize the maximum estimated travel time at each exit. The optimal solution satisfies the condition that the ratio of the total number of people allocated to each exit to its travel capacity is equal. ; Optimal allocation of personnel to each export Calculate as follows: ; The obtained optimal route allocation scheme is transformed into an information dissemination and guidance strategy, introducing a passenger compliance rate parameter λ, 0≤λ≤1, which characterizes the actual response rate of passengers to dynamic guidance information. The value of λ is determined based on actual operational statistics of similar hubs or on-site test calibration. λ=1 indicates complete compliance, and λ=0 indicates complete non-compliance. The proportion of passengers who comply with the guidance is... Evacuation will be carried out according to the optimized plan; the proportion of passengers who do not follow the guidance is [missing information]. Evacuation will be carried out according to the inertial path selection behavior; Revised actual number of personnel allocated to each export for: ,in To optimize the allocation of personnel at exit j in the plan, The number of people allocated to exit j under inertial behavior.

6. The method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase according to claim 1, characterized in that, The micro-level pedestrian simulation stage based on the social force model, which specifically includes: The motion of pedestrian i is described by Newton's second law: ,in, Let i be the mass of pedestrian i; It is the velocity vector; Self-driving force; Repulsive force between pedestrians; Repulsive force from obstacles; self-driving force The formula is: ,in, For the desired velocity scalar, Let be the unit vector pointing to the target location. The current velocity vector, Relaxation time; interpersonal repulsion The formula is: ,in and The interaction strength parameter; Let i be the radius of pedestrian i; Let be the distance between pedestrian i and pedestrian j; Let g be a unit vector pointing from pedestrian j to pedestrian i; g() is an indicator function. k Body compressive stiffness coefficient; obstacle repulsion force The formula is: ,in, The distance from the pedestrian to the nearest obstacle. Let be the unit vector pointing from the obstacle to the pedestrian; And calculate the pedestrian's velocity and position update based on the equation of motion, including: Using each simulation time step Inside, the velocity is updated using the Euler numerical iteration formula of Newton's laws of motion as follows: The position is updated using the Euler numerical integral formula: ,in, It is the velocity vector of the i-th individual at time t. It is the net external force acting on the individual. It is individual quality. It is the position vector of the i-th individual at time t. It is the time step of numerical integration, and the acceleration generated by the net external force. Multiply by the time step to obtain the velocity increment within one time step, thus completing the iterative update of velocity and position.

7. The method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase according to claim 1, characterized in that, The formula for calculating the Total Evacuation Completion Time (TET) is: ,in, The evacuation trigger time, Let p be the time when passenger p arrives at the safety exit; For all passengers The moment when they finally reach the safe exit.

8. A system for predicting and optimizing emergency evacuation of personnel during subway renovation construction, characterized in that, The system includes: The spatial network modeling module is used to obtain spatial layout information of each construction stage of the subway underground station group renovation project, and to construct a directed graph network model based on the boundary information of the spatial layout. The module for establishing a semantic rule constraint system includes setting two types of triggering conditions for influx pulses: one is a regular passenger flow pulse, and the other is an emergency evacuation pulse. The pedestrian congestion wave prediction and quantitative analysis module is used to identify the queuing state evolution pattern caused by limited exit capacity through the four-stage semantic state transition of S-FSM. It performs quantitative analysis on the two shock waves of the influx pulse. For the bottleneck edge with limited capacity, it calculates the queue formation time and the number of people in the queue based on the number of people entering and the actual capacity of the bottleneck. It calculates the number of people staying based on the integral of the flow rate of people entering the waiting area from the queue formation to the stage release trigger time. It calculates the duration of the release process and the exit evacuation rate based on the ratio of the number of people staying to the actual capacity of the exit. It outputs the spatiotemporal distribution prediction results of the congestion wave driven by S-FSM semantic state transition. The four-stage semantic state of S-FSM includes the initial response and waiting stage, the approach and queuing stage, the passage release and passage stage, and the arrival at the safe area stage. The multi-exit dynamic path allocation optimization module is used to determine all available evacuation exits and their actual capacity under the current working conditions, as well as the total number of people to be evacuated, based on the prediction results of the spatiotemporal distribution of congestion waves. It uses the maximum evacuation completion time of each exit as the optimization variable, establishes equality constraints, and constructs a load balancing optimization model. The optimal solution satisfies that the ratio of the number of people allocated to each exit to its capacity is equal. The passenger compliance rate parameter is introduced, and the corrected actual number of people allocated to each exit is the weighted sum of the number of people allocated by the optimized scheme and the number of people allocated by inertial behavior. The microscopic pedestrian simulation module based on the social force model is used to simulate and extrapolate the evacuation process by assigning pedestrians self-driving force, interaction force between pedestrians, and interaction force between pedestrians and obstacles based on spatial layout information and directed graph network model. The total evacuation completion time assessment and plan output module is used to calculate the total evacuation completion time as the evaluation index, based on the difference between the time when the last passenger arrives at the safe exit and the evacuation trigger time. If the time exceeds the safety standard, the module returns to the multi-exit dynamic path allocation optimization stage for re-optimization. The deviation between the predicted congestion parameters and the actual simulation values ​​is fed back to the parameter library for subsequent stage accuracy optimization. When switching construction stages, the closed area weights and temporary edges of the directed graph network model are updated. The semantic rule constraint system stage is repeatedly established until the total evacuation completion time assessment and plan output stage, forming an evacuation plan system covering the entire renovation cycle.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, wherein computer instructions are stored on the storage medium, characterized in that, When the instruction is executed by the processor, it implements the steps of the method for predicting and optimizing emergency evacuation of personnel during the subway renovation construction phase as described in any one of claims 1 to 7.

Citation Information

Patent Citations

  • Simulation path planning method and system for evacuation of people in subway station under multiple hazard sources

    CN114298438A

  • Subway evacuation dynamic decision simulation system based on virtual reality and artificial intelligence

    CN120874364A