Multi-disaster-oriented real-time optimization and collaborative linkage method for dynamic evacuation path of subway three-dimensional imaging

Abstract

The application discloses a subway three-dimensional imaging dynamic evacuation path real-time optimization and collaborative linkage method for multiple disasters, and belongs to the technical field of rail transit emergency treatment. Multiple source heterogeneous factors such as disaster influence, crowd density, channel basic conditions and visibility are quantified into dynamic weight coefficients, so that path calculation can respond to disaster development, crowd flow and environmental changes in real time, and traditional static or single rule guidance is upgraded to a real dynamic adaptive decision. Secondly, an improved Dijkstra algorithm is used to globally optimize the shortest total escape time, so that the generated evacuation path can simultaneously consider safety and timeliness under complex risk environment. On this basis, the optimized path decision is converted into three-dimensional guidance information and multi-system collaborative control strategy, and a complete closed loop from perception fusion, dynamic modeling, optimal solution to three-dimensional presentation and system linkage is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of rail transit emergency handling, specifically a real-time optimization and collaborative linkage method for three-dimensional imaging dynamic evacuation paths of subways facing multiple disasters. Background Technique

[0002] As a closed underground space, the emergency evacuation of subways faces multiple challenges such as dynamic risks, complex structures, and crowded crowds, which pose extremely high requirements for the timeliness and accuracy of evacuation guidance methods. However, there is a fundamental technical bottleneck in the current mainstream evacuation guidance technology at the methodological level: the lack of an intelligent control logic that can quantify multi-source dynamic environment information into optimal evacuation decisions in real time.

[0003] Specifically, there are three interrelated defects in the data processing logic of existing methods: First, they lack a unified multi-source heterogeneous information fusion mechanism and cannot jointly model and weight and quantify dynamic parameters with different physical meanings and dimensions such as disaster distance, gas concentration, crowd density, and visibility, resulting in path generation decisions relying only on single rules rather than real-time comprehensive risk assessment of the scene; Second, a closed-loop iterative logic triggered by real-time perception data is not built in their algorithm architecture, and the generated guidance directions cannot be updated synchronously during the spread of fire, diffusion of smoke, or migration of gas, resulting in evacuation instructions lagging behind the development trend of disasters and making it difficult to ensure the continuous optimality of the path; In addition, due to the lack of standardized data interfaces and collaborative control strategies between systems, the evacuation guidance module and devices such as smoke exhaust, broadcasting, and access control are in independent data islands and cannot generate unified linkage control instructions, resulting in fragmented responses of on-site emergency resources and low overall disposal efficiency. It is precisely these core methodological defects that make it difficult for existing technologies to achieve scientific and accurate real-time evacuation decisions in dynamic risk environments. Summary of the Invention

[0004] In order to solve the problems mentioned in the above background technique, the present invention provides a real-time optimization and collaborative linkage method for three-dimensional imaging dynamic evacuation paths of subways facing multiple disasters. The technical solutions adopted by the present invention are as follows:

[0005] A real-time optimization and collaborative linkage method for three-dimensional imaging dynamic evacuation paths of subways facing multiple disasters, characterized by including the steps of:

[0006] S10: Real-time collect disaster type information, environmental status information, and crowd density information through various types of perception devices deployed in the subway station, and summarize and generate environmental perception information for describing the current emergency scene;

[0007] S20: Construct an evacuation network model containing nodes and edges based on the acquired subway spatial structure information, and optimize the evacuation network model in real time based on environmental perception information, thereby generating dynamic network model information;

[0008] S30: Based on the dynamic network model information, the dynamic weight coefficients of each side are generated, the road segment escape information of each side is calculated, and the optimal evacuation route information is generated with the goal of minimizing the total escape time of the path.

[0009] S40: Convert the optimal evacuation path information into a set of control instructions for the three-dimensional imaging display lights and transmit them to each display light node to generate three-dimensional guidance information for presenting three-dimensional guidance;

[0010] S50: Generate a collaborative control strategy based on the optimal evacuation route information, and dynamically update the collaborative control strategy according to real-time monitoring information.

[0011] By adopting the above technical solution, various types of sensing devices deployed in subway stations are used to collect disaster type information, environmental status information, and crowd density information in real time. These multi-source data are aggregated and integrated to generate environmental sensing information that can fully describe the current emergency scenario. Secondly, based on the acquired subway spatial structure information, an evacuation network model containing nodes and edges is constructed, and the model is optimized and updated in real time according to the environmental sensing information to generate network model information that can reflect the dynamic changes of the current scenario. Then, based on the dynamic network model information, the dynamic weight coefficients of each edge are calculated, and the escape information of each edge is calculated. The optimal evacuation path information from each node to the safe exit is generated with the goal of minimizing the total escape time. Next, the optimal evacuation path information is converted into a control instruction set for three-dimensional imaging display lights and transmitted to each display light node through the communication network to generate display control information for driving the display lights to present three-dimensional guidance. Finally, a collaborative control strategy that links with the existing subway system is generated based on the optimal evacuation path information, and the strategy is dynamically updated according to real-time monitoring information. This constructs a closed-loop control logic that extends from multi-source data perception, dynamic optimization of network models, real-time calculation of optimal paths, generation of three-dimensional guidance, to multi-system collaborative linkage, providing a systematic methodological framework for dynamic evacuation decision-making in multi-hazard scenarios.

[0012] In a preferred embodiment, this application can be further configured such that: multiple types of sensing devices include a smoke sensor, a temperature sensor, a toxic gas sensor, a visibility meter, and an AI camera; step S10 includes the following steps:

[0013] S101: By deploying multiple types of sensing devices on subway platform pillars, transfer passage corners and stair entrances, smoke concentration information, temperature information, toxic gas concentration information, visibility information and crowd density information are collected simultaneously, and the original sensing information from multiple sources is obtained.

[0014] S102: The original sensing information is transmitted to the edge computing node, and the random forest algorithm is used to perform data fusion and recognition to generate the environmental sensing information.

[0015] By adopting the above technical solution, firstly, multiple types of sensing devices, such as smoke sensors, temperature sensors, toxic gas sensors, visibility meters, and AI cameras, deployed on subway platform pillars, transfer passage corners, and stairwell entrances, simultaneously collect smoke concentration information, temperature information, toxic gas concentration information, visibility information, and crowd density information, summarizing multi-source raw sensing information. Then, this raw sensing information is transmitted to edge computing nodes, and a random forest algorithm is used to fuse and identify multi-sensor data, generating environmental sensing information that includes disaster type, disaster level, environmental status, and crowd distribution information. In this way, heterogeneous multi-source raw data is transformed into structured information that uniformly describes the emergency scenario through intelligent fusion algorithms, providing an accurate and reliable data foundation for subsequent network model optimization and path calculation.

[0016] In a preferred embodiment, this application can be further configured such that step S20 includes the following steps:

[0017] S201: Based on the spatial structure information of the subway, the subway entrances and exits, passage intersections, stairwells and elevator entrances are abstracted as nodes, and the passages, stairs and elevators connecting the nodes are abstracted as edges to construct an evacuation network model.

[0018] S202: Based on the environmental perception information, update the real-time pedestrian flow density attribute, disaster impact range attribute, and real-time visibility attribute of each node and edge to generate the dynamic network model information.

[0019] By adopting the above technical solution, firstly, based on the acquired subway spatial structure information, subway entrances and exits, passageway intersections, stairwells, and elevator entrances are abstracted as network nodes, and the passageways, staircases, and elevators connecting the nodes are abstracted as network edges, thus constructing an initial evacuation network model. Then, based on environmental perception information, the real-time pedestrian flow density, disaster impact range, and real-time visibility attributes of each node and edge are updated in real time, generating network model information that reflects the dynamic changes of the current scene. The purpose is to transform the static subway physical space into a dynamically updatable networked mathematical model, enabling subsequent path calculations to be based on real-time perception data, ensuring that evacuation decisions remain synchronized with the actual situation on site.

[0020] In a preferred embodiment, this application can be further configured such that step S30 includes the following steps:

[0021] S301: Based on the dynamic network model information, calculate the disaster impact coefficient, population density coefficient, channel basic coefficient and visibility coefficient of each edge respectively, and perform fusion calculation to generate the dynamic weight coefficient of each edge;

[0022] S302: Based on the dynamic weight coefficient, the physical length information of each side and the basic traffic speed information, generate the road segment escape information corresponding to each side according to the preset road segment escape time calculation formula;

[0023] S303: The improved Dijkstra algorithm is adopted to traverse all nodes in the evacuation network model with the goal of minimizing the total escape time of the path, solve the optimal path from each node to the safe exit, and generate the optimal evacuation path information.

[0024] By adopting the above technical solution, firstly, based on the dynamic network model information, the disaster impact coefficient, crowd density coefficient, channel basic coefficient, and visibility coefficient of each edge are calculated separately, and these four coefficients are fused to generate the dynamic weight coefficient of each edge. Then, based on the dynamic weight coefficient, the physical length information of each edge, and the basic passage speed information, the road segment escape information corresponding to each edge is generated according to the preset road segment escape time calculation formula. Finally, an improved Dijkstra algorithm is used to traverse all nodes in the evacuation network model with the goal of minimizing the total escape time of the path, solving for the optimal path from each node to the safe exit, and generating the optimal evacuation path information. Furthermore, the multi-dimensional dynamic influencing factors are quantified into a unified weight system, and based on this, the optimal path is accurately solved, ensuring that the generated evacuation path has the shortest escape time under multiple constraints such as disaster risk, crowd congestion, channel capacity, and visibility conditions.

[0025] In a preferred embodiment, this application can be further configured such that step S301 includes the following steps:

[0026] S3011: Obtain the straight-line distance information from the center point of each road segment to the disaster center point, real-time toxic gas concentration information, and disaster type correction coefficient; generate the disaster impact coefficient according to the preset calculation formula; the specific formula is as follows:

[0027]

[0028] in, The straight-line distance from the center point of the road segment to the center of the disaster. The maximum distance affected by the fire source is set at 100 meters. This refers to the real-time concentration of toxic gas within the road section. The human body's tolerance threshold for toxic gas concentration is set at 5 mg / m³. The following are the disaster type correction factors: fire = 1.2, earthquake = 1.5, toxic gas leak = 2.0, no disaster = 1.0; =1.0、 =1.2、 =0.5 is the calibration parameter; The value range is from 1.0 to 6.0;

[0029] S3012: Based on the ratio of real-time crowd density information of the road segment to a preset density threshold, the crowd density coefficient is generated according to a preset crowd density coefficient calculation formula; the specific formula is as follows:

[0030]

[0031] in, This represents the real-time population density of the road segment. With a density of 3 people per square meter, β = 1.0 when the density does not exceed the threshold, and increases with density after exceeding the threshold.

[0032] S3013: Based on the actual width and slope information of the passage, the passage basic coefficient is generated according to the preset passage basic coefficient calculation formula; the specific formula is as follows:

[0033]

[0034] in, The actual width of the channel. Set to 3 meters; The slope of the passage. Set to 0.02; The value range is controlled between 0.8 and 1.2, and the larger the value, the stronger the passage capability;

[0035] S3014: Based on the ratio of real-time visibility information of the road segment to a preset visibility standard value, the visibility coefficient is generated according to a preset visibility coefficient calculation formula; the specific formula is as follows:

[0036]

[0037] in, Real-time visibility of the road segment. The standard value for normal visibility is set at 50 meters. When visibility is ≥ 50 meters... =1.0, which decreases as visibility decreases below 50 meters, with a minimum retention of 0.3.

[0038] By adopting the above technical solution, step S3011 obtains the straight-line distance information from the center point of each road segment to the center point of the disaster, real-time toxic gas concentration information, and disaster type correction coefficient, and generates a disaster impact coefficient according to a preset calculation formula. This coefficient is used to quantify the degree of obstruction of road segment passage by the disaster. Step S3012 generates a crowd density coefficient based on the ratio of real-time crowd density information of the road segment to a preset density threshold, according to a preset calculation formula. This coefficient is used to quantify the impact of crowd congestion on passage speed. Step S3013 generates a basic channel coefficient based on the actual width and slope information of the channel, according to a preset calculation formula. This coefficient is used to quantify the passage capacity of the channel itself. Step S3014 generates a visibility coefficient based on the ratio of real-time visibility information of the road segment to a preset visibility standard value, according to a preset calculation formula. This coefficient is used to quantify the impact of visibility on passage efficiency. The purpose of this step is to quantify the four heterogeneous factors—disaster, crowd, channel physical attributes, and environmental conditions—into coefficients with clear physical meaning, providing standardized input parameters for subsequent dynamic weight fusion.

[0039] In a preferred embodiment, this application can be further configured such that, in step S302, the preset formula for calculating the escape time of the road segment is as follows:

[0040]

[0041] in, Let be the physical length of the i-th channel segment. Set to 1.2 m / s, The bigger, The smaller the section, the longer the escape time.

[0042] The total escape time T for the route is calculated by summing the escape times of all segments included in the path. The specific calculation formula is as follows:

[0043]

[0044] Where n is the total number of road segments included in the optimal path, the algorithm aims to find the minimum value of T from each evacuation node to the safe exit, while avoiding disaster nodes and congested road segments.

[0045] By adopting the above technical solution, firstly, based on dynamic weight coefficients, physical length information of each side, and basic traffic speed information, escape information for each side is generated according to a preset formula for calculating escape time. This formula uses the product of the disaster impact coefficient and the population density coefficient as a risk amplification factor, and the product of the channel basic coefficient and the visibility coefficient as a traffic capacity attenuation factor, both of which affect the basic travel time. Then, by accumulating the escape times of each side of the path, the total escape time of the path is calculated, and the optimal path is solved with the goal of minimizing this total escape time. At the same time, areas marked as disaster nodes or congested road sections are automatically avoided during the solution process. The purpose of this step is to transform multi-dimensional dynamic weights into quantifiable time costs, so that the path optimization objective of "shortest total escape time" can truly reflect the comprehensive risk and traffic efficiency of the scenario, ensuring that the generated path is both safe and efficient.

[0046] In a preferred embodiment, this application can be further configured such that step S303 includes the following steps:

[0047] S3031: Initialize the total escape time of all evacuation nodes to infinity, set the total escape time of only the safe exit nodes to 0, and create a priority queue based on a binary heap to add all nodes to the queue.

[0048] S3032: Take the node with the shortest total escape time from the priority queue as the current node, and traverse all adjacent nodes of the current node;

[0049] S3033: For each of the adjacent nodes, a candidate total escape time is generated based on the sum of the current node's total escape time and the escape time of the connecting edge. If the candidate total escape time is less than the current total escape time of the adjacent node, the total escape time and predecessor node information of the adjacent node are updated, and the updated node is added back to the priority queue.

[0050] S3034: Repeat steps S3032 to S3033 until the priority queue is empty or all safety exit nodes have been processed. Backtrack from each evacuation node to the safety exit node through the predecessor node information to generate the optimal evacuation path information.

[0051] By adopting the above technical solution, firstly, the total escape time of all evacuation nodes is initialized to infinity, while the total escape time of the safety exit nodes is set to 0. A priority queue based on a binary heap is created and all nodes are added to the queue. Then, the node with the smallest current total escape time is taken from the priority queue as the current node, and all adjacent nodes of this node are traversed. Secondly, for each adjacent node, a candidate total escape time is generated based on the sum of the current node's total escape time and the escape time of the connecting edge. If the candidate time is less than the current recorded total escape time of the adjacent node, the total escape time and predecessor node information of the adjacent node are updated, and the updated node is added back to the priority queue. Finally, the above steps are repeated until the priority queue is empty or all safety exit nodes have been processed. Finally, the evacuation path information is generated by backtracking from each evacuation node to the safety exit node through the predecessor node information, which includes the complete path node sequence and the estimated total evacuation time. The purpose of this step is to quickly solve for the optimal path from all nodes to the safe exit in a dynamic network model using an efficient search algorithm optimized by a binary heap, ensuring that the real-time performance of the path calculation meets the timeliness requirements of emergency response.

[0052] In a preferred embodiment, this application can be further configured such that step S40 includes the following steps:

[0053] S401: Convert the optimal evacuation route information into a set of control instructions containing equipment identification information, guide color parameter information, arrow direction encoding information, remaining distance value information and effective duration information, and transmit it to each three-dimensional imaging display light;

[0054] S402: After receiving and parsing the control instruction set, each 3D imaging display lamp generates display control information for driving the display lamp to present the 3D arrow shape, color encoding status, and remaining distance projection.

[0055] By adopting the above technical solution, the optimal evacuation route information is first converted into a control command set containing equipment identification information, directional color parameter information, arrow direction encoding information, remaining distance numerical information, and effective duration information. This set is then transmitted to each 3D imaging display light via a communication network. Each 3D imaging display light receives and parses the control command set, generating display control information to drive the display light to present the 3D arrow shape, color encoding status, and remaining distance projection. Thus, the optimal route decision calculated by the algorithm layer is sent to the terminal display device through a standardized command format and transformed into stereoscopic visual guidance information that personnel can directly perceive, achieving the crucial transformation from "decision-making" to "presentation."

[0056] In a preferred embodiment, this application can be further configured such that step S50 includes the following steps:

[0057] S501: Generate a collaborative control strategy based on the optimal evacuation path information, which includes the control parameters and action timing of each collaborative system device, and send it to the existing subway system;

[0058] S502: Real-time collection of crowd density and personnel behavior information in the passage through AI cameras; when congestion or reverse behavior is detected, automatic path recalculation is triggered, and the optimal evacuation path information and the collaborative control strategy are updated based on the recalculation results.

[0059] S503: After the sensor continuously monitors and detects that the environmental parameters have returned to normal and meet the preset disaster relief conditions, it automatically determines that the disaster has been relieved and generates system recovery instructions and indicator light reset control information.

[0060] By adopting the above technical solution, step S501 generates a collaborative control strategy based on the optimal evacuation route information, which includes control parameters and action sequences of each collaborative system device, and sends it to the existing subway system through a standardized interface; step S502 collects real-time information on crowd density and personnel behavior in the passageway using AI cameras, and automatically triggers route recalculation when congestion or reverse movement is detected, and updates the optimal evacuation route information and collaborative control strategy based on the recalculation results; step S503, after sensors continuously monitor that environmental parameters have returned to normal and meet the preset disaster relief conditions, automatically determines that the disaster has been relieved and generates system recovery instructions and indicator light reset control information. The purpose of this step is to extend the evacuation route decision-making to collaborative linkage with the existing subway system, and to form a dynamically adjusted closed-loop mechanism through real-time monitoring data, ensuring that the entire evacuation process can adapt to scene changes and ultimately achieve orderly recovery.

[0061] The beneficial effects of this invention on the real-time optimization and collaborative linkage method for dynamic evacuation paths in subway 3D imaging for multiple disasters are as follows:

[0062] By unifying and quantifying multi-source heterogeneous factors such as disaster impact, crowd density, basic passage conditions, and visibility into dynamic weight coefficients, path calculation can respond in real time to disaster development, crowd flow, and environmental changes, elevating traditional static or single-rule guidance to truly dynamic adaptive decision-making. Secondly, an improved Dijkstra algorithm is employed to achieve global optimization with the goal of minimizing the total escape time, ensuring that the generated evacuation paths balance safety and timeliness in complex and risky environments. Based on this, the optimized path decisions are transformed into three-dimensional guidance information and multi-system collaborative control strategies, achieving a complete closed loop from perception fusion, dynamic modeling, optimal solution to three-dimensional presentation and system linkage. This method, with dynamic weight fusion as its core control logic, systematically solves the problem of accurate evacuation under multiple constraints of "dynamic risk + complex structure + crowd congestion" in multi-hazard scenarios, significantly improving the scientific rigor and reliability of subway emergency response. Attached Figure Description

[0063] Figure 1 This is a flowchart of an embodiment of the method for real-time optimization and collaborative linkage of dynamic evacuation paths in subway 3D imaging for multiple disasters in this application;

[0064] Figure 2 This is a flowchart of step S30 in the embodiment of the method for real-time optimization and collaborative linkage of dynamic evacuation paths in subway 3D imaging for multiple disasters in this application.

[0065] Figure 3 This is a flowchart illustrating step S303 in the embodiment of the method for real-time optimization and collaborative linkage of dynamic evacuation paths in subway 3D imaging for multiple disasters in this application. Detailed Implementation

[0066] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0067] In one embodiment, such as Figure 1 As shown, this application discloses a method for real-time optimization and collaborative linkage of dynamic evacuation paths based on 3D imaging in subway systems for multiple disasters, specifically including the following steps:

[0068] S10: Collect disaster type information, environmental status information and crowd density information in real time through various types of sensing devices deployed in subway stations, and summarize them to generate environmental sensing information to describe the current emergency scenario;

[0069] In this embodiment, "multi-type sensing devices" refers to a general term for various sensors and monitoring devices deployed in subway stations to collect various emergency-related data; "disaster type information" refers to the types of disasters currently occurring, such as fires, earthquakes, and toxic gas leaks, identified by the sensing devices; "environmental status information" refers to data describing the environmental conditions within the subway station, including temperature, smoke concentration, visibility, and toxic gas concentration; "crowd density information" is quantitative data describing the density of people in various areas of the subway station; and "environmental sensing information" is a unified structured information that can completely describe the current emergency scenario, generated by aggregating and integrating information collected from multiple sources.

[0070] Specifically, by deploying various types of sensing devices in subway stations, information on disaster type, environmental status, and population density in the current emergency scenario is collected in real time. After summarizing and integrating these multi-source data, environmental sensing information that can fully describe the current scenario is generated.

[0071] S20: Construct an evacuation network model containing nodes and edges based on the acquired subway spatial structure information, and optimize the evacuation network model in real time based on environmental perception information, thereby generating dynamic network model information;

[0072] In this embodiment, subway spatial structure information refers to data describing the physical spatial layout within a subway station, including passage location, length, width, slope, number of stair steps, etc.; nodes refer to point-like elements that abstract key locations such as subway entrances / exits, passage intersections, stairwells, and elevator entrances in the subway spatial network model; edges refer to line-like elements that abstract passageways, stairs, elevators, and other travel paths connecting nodes in the subway spatial network model; the evacuation network model is a graph structure model composed of nodes and edges constructed based on subway spatial structure information, used to represent the network of paths accessible to personnel; dynamic network model information refers to network model data that reflects the dynamic changes of the current scene after updating the attributes of each node and edge based on real-time environmental perception information on the basis of the initial evacuation network model.

[0073] Specifically, based on the acquired subway spatial structure information, an evacuation network model consisting of nodes and edges is constructed, and the model is optimized and updated in real time according to environmental perception information to generate dynamic network model information that can reflect the current dynamic changes.

[0074] S30: Based on the dynamic network model information, the dynamic weight coefficients of each side are generated, the road segment escape information of each side is calculated, and the optimal evacuation route information is generated with the goal of minimizing the total escape time of the path.

[0075] In this embodiment, the dynamic weight coefficient refers to the quantitative weight calculated for each edge, which comprehensively reflects multiple factors such as disaster impact, crowd congestion, physical conditions of the passage, and visibility, and is used to evaluate the passage cost of that edge; the road segment escape information refers to the quantitative data calculated based on the dynamic weight coefficient and the physical properties of the edge, used to characterize the time required to pass through a certain edge; the total escape time of the path refers to the total time required to reach the safe exit from a certain evacuation node along a path, which is obtained by summing the escape times of each edge segment that constitutes the path; the optimal evacuation path information refers to the optimal path data from each evacuation node to the safe exit obtained with the goal of minimizing the total escape time of the path, including the path sequence and the estimated time.

[0076] Specifically, the dynamic weight coefficients of each side are calculated based on the information of the dynamic network model, and then the road segment escape information of each side is calculated. The optimal path from each node to the safe exit is solved with the goal of minimizing the total escape time of the path, and the optimal evacuation path information is generated.

[0077] S40: Convert the optimal evacuation path information into a set of control instructions for the three-dimensional imaging display lights and transmit them to each display light node to generate three-dimensional guidance information for presenting three-dimensional guidance;

[0078] In this embodiment, the control instruction set is a standardized instruction data set that converts the optimal evacuation path information into a set of instructions that can be recognized and executed by the 3D imaging display lights; the display light node refers to each 3D imaging display light device unit deployed in the subway passage; the 3D guidance information refers to the visual guidance signal presented by the 3D imaging display lights, which includes the direction of the 3D arrow, the color-coded status, and the remaining distance projection.

[0079] Specifically, the optimal evacuation route information is converted into a set of control instructions that can be recognized by the 3D imaging display lights, and transmitted to each display light node through a communication network to generate 3D guidance information that presents 3D arrow directions, color-coded status, and remaining distance projections to evacuees.

[0080] S50: Generate a collaborative control strategy based on the optimal evacuation route information, and dynamically update the collaborative control strategy according to real-time monitoring information.

[0081] In this embodiment, the collaborative control strategy refers to the control parameters and timing arrangement generated based on the optimal evacuation path information to guide the collaborative actions of the existing subway system; the real-time monitoring information refers to the latest sensing data continuously collected during the evacuation process, used to dynamically assess scene changes and trigger strategy updates.

[0082] Specifically, a collaborative control strategy is generated based on the optimal evacuation route information and linked with the existing subway system. This strategy is then dynamically updated based on real-time monitoring information to ensure that the entire evacuation process can adapt to changes in the scenario.

[0083] In one embodiment, the multi-type sensing devices include a smoke sensor, a temperature sensor, a toxic gas sensor, a visibility meter, and an AI camera. Step S10 includes the following steps:

[0084] S101: By deploying multiple types of sensing devices on subway platform pillars, transfer passage corners and stair entrances, smoke concentration information, temperature information, toxic gas concentration information, visibility information and crowd density information are collected simultaneously, and the original sensing information from multiple sources is obtained.

[0085] S102: The original sensing information is transmitted to the edge computing node, and the random forest algorithm is used to perform data fusion and recognition to generate the environmental sensing information.

[0086] In this embodiment, smoke concentration information refers to quantitative data collected by a smoke sensor describing the amount of smoke in the air; temperature information refers to quantitative data collected by a temperature sensor describing the ambient temperature; toxic gas concentration information refers to quantitative data collected by a toxic gas sensor describing the amount of toxic gases in the air; visibility information refers to quantitative data measured by a visibility meter describing the maximum distance at which a person can clearly see objects in front of them; crowd density information refers to quantitative data obtained by an AI camera through image analysis describing the number of people per unit area; raw perception information is the raw data set collected synchronously by various sensing devices without fusion processing; edge computing nodes are computing devices deployed near the data source location, i.e., within the subway station, used to process and fuse the raw perception information locally to reduce transmission latency.

[0087] Specifically, firstly, various types of sensing devices, including smoke sensors, temperature sensors, toxic gas sensors, visibility meters, and AI cameras, are deployed at key locations such as subway platform pillars, transfer passage corners, and stairwell entrances to simultaneously collect current smoke concentration, temperature, toxic gas concentration, visibility, and crowd density information. This data from different sensors is then aggregated to obtain multi-source raw sensing information. Subsequently, this raw sensing information is transmitted to edge computing nodes deployed within the station. At these nodes, a random forest algorithm is used to fuse and identify the multi-source data, transforming the originally isolated and heterogeneous raw data into environmental sensing information that comprehensively describes the current emergency scenario, including disaster type, disaster level, environmental state, and crowd distribution information. The purpose of this step is to transform scattered raw sensing data into unified and reliable structured information through edge computing and intelligent fusion algorithms, providing high-quality input data for subsequent network model optimization and path calculation.

[0088] In one embodiment, step S20 includes the following steps:

[0089] S201: Based on the spatial structure information of the subway, the subway entrances and exits, passage intersections, stairwells and elevator entrances are abstracted as nodes, and the passages, stairs and elevators connecting the nodes are abstracted as edges to construct an evacuation network model.

[0090] S202: Based on the environmental perception information, update the real-time pedestrian flow density attribute, disaster impact range attribute, and real-time visibility attribute of each node and edge to generate the dynamic network model information.

[0091] In this embodiment, the real-time pedestrian density attribute refers to the dynamic attribute value assigned to each node or edge in the evacuation network model, reflecting the current population density at that location; the disaster impact range attribute is the dynamic attribute value assigned to each edge in the evacuation network model, reflecting the degree of disaster impact on that road segment, and is usually calculated based on the disaster type, location, and direction of spread; the real-time visibility attribute refers to the dynamic attribute value assigned to each edge in the network model, reflecting the current visibility conditions of that road segment.

[0092] Specifically, based on the acquired subway spatial structure information, key locations such as subway entrances / exits, passageway intersections, stairwells, and elevator entrances are abstracted as network nodes, and the passageways, staircases, and elevators connecting these nodes are abstracted as network edges, thus constructing an initial evacuation network model. This model, in graph structure form, fully describes the network of accessible paths for people within the subway station. Subsequently, based on environmental perception information, the attributes of each node and edge in the model are updated in real time: the real-time pedestrian flow density attribute of each node and edge is updated according to crowd density information; the disaster impact range attribute of each edge is updated according to disaster type and location information; and the real-time visibility attribute of each edge is updated according to visibility information. After attribute updates, a dynamic network model information that reflects the dynamic changes of the current scene is generated. The purpose of this step is to transform the static subway physical space into a dynamically updatable networked mathematical model, enabling subsequent path calculations to be based on real-time attribute data that is synchronized with the actual situation on site, ensuring the timeliness and accuracy of evacuation decisions.

[0093] like Figure 2 As shown, in one embodiment, step S30 includes the following steps:

[0094] S301: Based on the dynamic network model information, calculate the disaster impact coefficient, population density coefficient, channel basic coefficient and visibility coefficient of each edge respectively, and perform fusion calculation to generate the dynamic weight coefficient of each edge;

[0095] S302: Based on the dynamic weight coefficient, the physical length information of each side and the basic traffic speed information, generate the road segment escape information corresponding to each side according to the preset road segment escape time calculation formula;

[0096] S303: The improved Dijkstra algorithm is adopted to traverse all nodes in the evacuation network model with the goal of minimizing the total escape time of the path, solve the optimal path from each node to the safe exit, and generate the optimal evacuation path information.

[0097] In this embodiment, the disaster impact coefficient refers to a coefficient calculated for each edge that quantifies the degree of disaster impact on the road segment. The larger the value, the more severe the obstruction to traffic on the road segment. The crowd density coefficient refers to a coefficient calculated for each edge that quantifies the impact of crowd congestion on the traffic speed of the road segment. The larger the value, the more significant the decrease in traffic efficiency caused by congestion. The channel foundation coefficient refers to a coefficient calculated for each edge that quantifies the impact of the physical conditions of the road segment, such as width and slope, on the traffic capacity. The larger the value, the better the traffic conditions of the channel itself. The visibility coefficient refers to a coefficient calculated for each edge that quantifies the impact of visibility conditions on the traffic efficiency of the road segment. The smaller the value, the worse the visibility, and the more severe the obstruction to traffic. The dynamic weight coefficient is a single factor that comprehensively reflects the traffic cost of the road segment after the disaster impact coefficient, crowd density coefficient, channel foundation coefficient, and visibility coefficient are integrated and calculated. Quantified weights; physical length information refers to quantified data describing the actual distance of the passage corresponding to each edge; basic passage speed information refers to the average moving speed of people on a straight passage under ideal conditions without any interference; road segment escape time calculation formula: refers to a pre-set mathematical expression used to calculate the time required to cross a certain edge based on dynamic weight coefficients, physical length, and basic passage speed; road segment escape information refers to the specific numerical value used to characterize the time required to cross a certain edge, calculated based on the formula; improved Dijkstra algorithm refers to the shortest path solution algorithm optimized by using a binary heap priority queue based on the classic Dijkstra algorithm, used to efficiently search for the path with the minimum total weight in a weighted graph; optimal evacuation route information refers to the data set generated after the algorithm is solved, containing the optimal path sequence from each node to the safe exit and the estimated total time.

[0098] Specifically, based on the attribute data of each edge contained in the dynamic network model information, the disaster impact coefficient, population density coefficient, channel basic coefficient, and visibility coefficient of each edge are calculated separately. Then, these four coefficients, each reflecting different dimensions of influencing factors, are fused to generate a dynamic weight coefficient for each edge. This coefficient comprehensively represents the passage cost of that road segment. Next, based on the dynamic weight coefficient, the physical length information of each edge, and the basic passage speed information, the escape information corresponding to each edge is calculated according to the preset road segment escape time calculation formula, i.e., the time required to pass through that road segment. Finally, using an improved Dijkstra algorithm, with the goal of minimizing the total escape time, all nodes in the evacuation network model are traversed to solve for the optimal path from each node to the safe exit, ultimately generating optimal evacuation path information containing a complete path sequence and the estimated total evacuation time. The purpose of this step is to quantify multi-dimensional dynamic influencing factors into a unified passage time cost and to quickly solve for the optimal path in the dynamic network using an efficient algorithm, providing accurate decision-making basis for subsequent three-dimensional guidance and system coordination.

[0099] In one embodiment, step S301 includes the following steps:

[0100] S3011: Obtain the straight-line distance information from the center point of each road segment to the disaster center point, real-time toxic gas concentration information, and disaster type correction coefficient; generate the disaster impact coefficient according to the preset calculation formula; the specific formula is as follows:

[0101]

[0102] in, The straight-line distance from the center point of the road segment to the center of the disaster. The maximum distance affected by the fire source is set at 100 meters. This refers to the real-time concentration of toxic gas within the road section. The human body's tolerance threshold for toxic gas concentration is set at 5 mg / m³. The following are the disaster type correction factors: fire = 1.2, earthquake = 1.5, toxic gas leak = 2.0, no disaster = 1.0; =1.0、 =1.2、 =0.5 is the calibration parameter; The value range is from 1.0 to 6.0;

[0103] S3012: Based on the ratio of real-time crowd density information of the road segment to a preset density threshold, the crowd density coefficient is generated according to a preset crowd density coefficient calculation formula; the specific formula is as follows:

[0104]

[0105] in, This represents the real-time population density of the road segment. With a density of 3 people per square meter, β = 1.0 when the density does not exceed the threshold, and increases with density after exceeding the threshold.

[0106] S3013: Based on the actual width and slope information of the passage, the passage basic coefficient is generated according to the preset passage basic coefficient calculation formula; the specific formula is as follows:

[0107]

[0108] in, The actual width of the channel. Set to 3 meters; The slope of the passage. Set to 0.02; The value range is controlled between 0.8 and 1.2, and the larger the value, the stronger the passage capability;

[0109] S3014: Based on the ratio of real-time visibility information of the road segment to a preset visibility standard value, the visibility coefficient is generated according to a preset visibility coefficient calculation formula; the specific formula is as follows:

[0110]

[0111] in, Real-time visibility of the road segment. The standard value for normal visibility is set at 50 meters. When visibility is ≥ 50 meters... =1.0, which decreases as visibility decreases below 50 meters, with a minimum retention of 0.3.

[0112] In this embodiment, the road segment center point refers to the geometric center of the passage corresponding to each side, used to calculate the distance between the road segment and the disaster center; the disaster center point refers to the source location of the disaster, such as the ignition point of a fire, the source of a toxic gas leak, etc.; the straight-line distance information refers to the spatial straight-line distance between the road segment center point and the disaster center point, used to quantify the proximity of the road segment to the disaster; the real-time toxic gas concentration information refers to the toxic gas content data of the area where the road segment is located, collected by a toxic gas sensor at the current moment; the disaster type correction coefficient refers to the adjustment coefficient set according to the different degrees of traffic obstruction caused by different disaster types, such as fire, earthquake, and toxic gas leak. The preset density threshold refers to the pre-set critical density value used to determine whether the road segment begins to affect traffic efficiency due to crowding; the width information refers to the physical width measurement value of the passage corresponding to each side; the slope information refers to the tilt angle or slope ratio measurement value of the passage corresponding to each side; the visibility standard value refers to the pre-set critical visibility distance value used to determine whether visibility begins to affect traffic.

[0113] Specifically, the system acquires information on the straight-line distance from the center point of each road segment to the center point of the disaster, the real-time toxic gas concentration of the road segment, and a disaster type correction coefficient determined based on the disaster type. These parameters are then substituted into a preset disaster impact coefficient calculation formula to generate a disaster impact coefficient that reflects the degree of disaster impact on the road segment. This coefficient comprehensively considers the distance from the disaster, the hazard of the toxic gas concentration, and the characteristics of different disaster types. Next, based on the ratio of the real-time crowd density information of the road segment to a preset density threshold, a crowd density coefficient is generated according to a preset crowd density coefficient calculation formula. This coefficient reflects the impact of crowd congestion on the traffic speed of the road segment. When the density is below the threshold, the coefficient is 1, indicating no impact; above the threshold, it increases with increasing density. Finally, based on the actual width and slope information of the passageway for the road segment, a passageway basic coefficient is generated according to a preset passageway basic coefficient calculation formula. This coefficient comprehensively considers width and slope factors; a larger value indicates better traffic conditions. Finally, based on the ratio of the real-time visibility information of the road segment to the preset visibility standard value, a visibility coefficient reflecting the impact of visibility conditions on traffic efficiency is generated according to the preset visibility coefficient calculation formula. This coefficient takes a value of 1 when visibility is good and decreases as visibility decreases. This quantifies four heterogeneous factors—disasters, crowds, physical conditions, and environmental visibility—into coefficients with clear physical meanings, providing standardized input parameters for subsequent dynamic weight fusion.

[0114] In one embodiment, the preset formula for calculating the escape time of a road segment in step S302 is as follows:

[0115]

[0116] in, Let be the physical length of the i-th channel segment. Set to 1.2 m / s, The bigger, The smaller the section, the longer the escape time.

[0117] The total escape time T for the route is calculated by summing the escape times of all segments included in the path. The specific calculation formula is as follows:

[0118]

[0119] Where n is the total number of road segments included in the optimal path, the algorithm aims to find the minimum value of T from each evacuation node to the safe exit, while avoiding disaster nodes and congested road segments.

[0120] In this embodiment, the road segment escape information refers to the calculated road segment escape time value for each edge; the total path escape time refers to the sum of the road segment escape times of all edges that constitute a path; and the total number of road segments included in the path refers to the number of edges traversed in a path from the starting point to the ending point.

[0121] Specifically, firstly, based on the dynamic weighting coefficients generated in step S301, combined with the physical length information and basic traffic speed information of each edge, the escape time is calculated according to the preset formula. This formula applies the dynamic weighting coefficients to the basic travel time, where the product of the disaster impact coefficient and the population density coefficient serves as a risk amplification factor, and the product of the channel basic coefficient and the visibility coefficient serves as a traffic capacity attenuation factor, jointly determining the actual time required to pass through that edge. After the calculation is completed, the escape information for each edge is obtained. Subsequently, for any path composed of multiple edges, the escape times of all its edges are summed to obtain the total escape time for that path. The improved Dijkstra algorithm searches for the minimum total escape time from each evacuation node to the safe exit, and automatically avoids areas marked as disaster nodes or congested sections during the solution process, ensuring that the generated path avoids high-risk areas while maintaining traffic efficiency. The purpose of this step is to transform multidimensional dynamic weights into quantifiable time costs, so that the path optimization objective of "shortest total escape time" can truly reflect the comprehensive risks and traffic efficiency of the scenario.

[0122] like Figure 3 As shown, in one embodiment, step S303 includes the following steps:

[0123] S3031: Initialize the total escape time of all evacuation nodes to infinity, set the total escape time of only the safe exit nodes to 0, and create a priority queue based on a binary heap to add all nodes to the queue.

[0124] S3032: Take the node with the shortest total escape time from the priority queue as the current node, and traverse all adjacent nodes of the current node;

[0125] S3033: For each of the adjacent nodes, a candidate total escape time is generated based on the sum of the current node's total escape time and the escape time of the connecting edge. If the candidate total escape time is less than the current total escape time of the adjacent node, the total escape time and predecessor node information of the adjacent node are updated, and the updated node is added back to the priority queue.

[0126] S3034: Repeat steps S3032 to S3033 until the priority queue is empty or all safety exit nodes have been processed. Backtrack from each evacuation node to the safety exit node through the predecessor node information to generate the optimal evacuation path information.

[0127] In this embodiment, evacuation nodes refer to all nodes in the evacuation network model that represent the possible locations of people to be evacuated; total escape time refers to the shortest known time from a node to a safe exit; total escape time infinity refers to a maximum value assigned to unexplored nodes during the algorithm initialization phase, indicating that the time to reach the safe exit is currently unknown; safe exit nodes refer to nodes in the evacuation network model that represent those that can reach the ground safety area; priority queue refers to a queue sorted according to the current total escape time of nodes, from which the node with the shortest time is taken for expansion each time; candidate total escape time refers to the possible time required to reach an adjacent node through the current node, which is obtained by adding the total escape time of the current node to the escape time of the connecting edge segment; predecessor node information refers to the information recorded during the path search process about which node each node is extended from, used for the final reverse backtracking to generate a complete path; reverse backtracking refers to the process of starting from the safe exit node and tracing back along the predecessor node information to the starting node to obtain the complete path sequence.

[0128] Specifically, the algorithm is first initialized by setting the total escape time of all evacuation nodes to infinity to indicate that no path has been found, while setting the total escape time of only safety exit nodes to 0. A priority queue based on a binary heap data structure is created, and all nodes are added to the queue, which is sorted according to their total escape time. Then, iteration begins. The node with the smallest total escape time is selected from the priority queue as the current node, and all its adjacent nodes are traversed. Next, for each adjacent node, based on the sum of the current node's total escape time and the escape time of the connecting edge, a candidate total escape time to reach that adjacent node via the current node is calculated. If this candidate time is less than the currently recorded total escape time of the adjacent node, it means a better path has been found. Therefore, the total escape time and predecessor node information of the adjacent node are updated, and the updated node is re-added to the priority queue for subsequent expansion. Finally, repeat the above iterative process until the priority queue is empty or all safe exit nodes have been processed. At this point, the shortest path for all nodes has been determined. Finally, starting from each evacuation node, backtrack to the safe exit node using the recorded predecessor node information to generate the optimal evacuation path information, which includes the complete path node sequence and the estimated total evacuation time. The purpose of this step is to quickly solve for the optimal path from all nodes to the safe exit in a dynamic network model using an efficient search algorithm optimized by a binary heap, ensuring that the real-time performance of path calculation meets the timeliness requirements of emergency response.

[0129] In one embodiment, step S40 includes the following steps:

[0130] S401: Convert the optimal evacuation route information into a set of control instructions containing equipment identification information, guide color parameter information, arrow direction encoding information, remaining distance value information and effective duration information, and transmit it to each three-dimensional imaging display light;

[0131] S402: After receiving and parsing the control instruction set, each 3D imaging display lamp generates display control information for driving the display lamp to present the 3D arrow shape, color encoding status, and remaining distance projection.

[0132] In this embodiment, device identification information refers to information used to uniquely identify each 3D imaging display light device, ensuring that control commands can be accurately delivered to the target device; guidance color parameter information refers to control parameters used to specify the color that the display light should display, such as green representing a safe path, yellow representing a slow path, and red representing a danger zone; arrow direction encoding information refers to encoded data used to specify the direction of the 3D arrow that the display light should display, such as horizontal straight, left turn, right turn, upstairs, downstairs; remaining distance value information refers to a value used to indicate how far the current position is from the safety exit, which is displayed by the display light through laser projection technology; effective duration information specifies the effective duration of the current control command, after which the display light can automatically resume or wait for a new command; control command set refers to a standardized command data set that encapsulates the above types of information and can be parsed and executed by the display light; display control information refers to the control signal data generated by the display light after parsing the command, used to drive the hardware to present specific display effects.

[0133] Specifically, the optimal evacuation path information generated in step S30 is first converted, and control parameters related to each display light node are extracted. These parameters include device identification information for identifying target equipment, guidance color parameters for determining display colors, arrow direction encoding information for determining the direction of the arrow, numerical information for indicating the remaining distance, and the effective duration of the command. These parameters are encapsulated into a standardized control command set, which is then transmitted to each 3D imaging display light deployed in the subway passage via a communication network. Upon receiving the control command set, each 3D imaging display light parses the commands, extracts its required display parameters, and generates display control information based on these parameters to drive the hardware to present the corresponding display effects: driving the LED ring array to cooperate with Fresnel lenses to generate a 3D arrow shape in a specified direction, controlling the LED beads to present the corresponding color-coded state, and projecting the remaining distance to the safety exit onto the ground or wall in real time using laser projection technology. The purpose of this step is to transmit the optimal path decision calculated by the algorithm layer to the terminal display device through a standardized command format, transforming it into 3D visual guidance information that personnel can directly perceive, thus achieving the key transformation from "decision-making" to "presentation."

[0134] In one embodiment, step S50 includes the following steps:

[0135] S501: Generate a collaborative control strategy based on the optimal evacuation path information, which includes the control parameters and action timing of each collaborative system device, and send it to the existing subway system;

[0136] S502: Real-time collection of crowd density and personnel behavior information in the passage through AI cameras; when congestion or reverse behavior is detected, automatic path recalculation is triggered, and the optimal evacuation path information and the collaborative control strategy are updated based on the recalculation results.

[0137] S503: After the sensor continuously monitors and detects that the environmental parameters have returned to normal and meet the preset disaster relief conditions, it automatically determines that the disaster has been relieved and generates system recovery instructions and indicator light reset control information.

[0138] In this embodiment, the collaborative control strategy refers to the complete strategy data that integrates the control parameters and action timing information of the collaborative system equipment and is used to uniformly guide the linkage of multiple systems; the channel crowd density information refers to the data obtained in real time through AI camera analysis that describes the density of people in each channel; the personnel behavior information refers to the data information obtained in real time through AI camera analysis that describes abnormal personnel behaviors such as going against the flow or lingering; the system recovery command refers to the command data generated after the disaster is resolved and used to guide each system to restore normal operation mode; and the indicator light reset control information refers to the reset command used to control the 3D imaging indicator light to switch from emergency guidance mode back to normal operation mode.

[0139] Specifically, firstly, based on the optimal evacuation route information generated in step S30, a collaborative control strategy is generated, including control parameters and action sequences for each collaborative system device. This strategy is then sent to the subway's existing automatic fire alarm system, environmental and equipment monitoring system, broadcasting system, and access control system via a standardized interface, enabling coordinated responses from smoke extraction, lighting, broadcasting, and access control devices. Then, during the evacuation process, AI cameras collect real-time information on crowd density and personnel behavior within the passageways. When congestion is detected in a passageway or reverse movement is observed, the system automatically determines that the current route may no longer be optimal, triggering a route recalculation process. Step S30 is re-executed to generate new optimal evacuation route information, and the collaborative control strategy is updated synchronously based on the recalculation results, achieving dynamic intervention and adjustment of the evacuation process. Finally, throughout the evacuation process, various sensors continuously monitor environmental parameters. When smoke concentration, toxic gas concentration, vibration, and other parameters return to normal and meet preset disaster clearance conditions, the system automatically determines that the disaster has been cleared. It then generates a system recovery command to guide each collaborative system back to normal operation mode, and simultaneously generates indicator light reset control information to control the 3D imaging indicator lights to switch from emergency guidance mode back to normal operation mode. The purpose of this step is to extend the decision-making of evacuation routes to the coordination and linkage with the existing subway system, and to form a closed-loop mechanism for dynamic adjustment through real-time monitoring data, so as to ensure that the entire evacuation process can adapt to changes in the scene and ultimately achieve orderly recovery.

[0140] The present invention and its embodiments have been described above. This description is not restrictive. The accompanying drawings are only one embodiment of the present invention. The actual structure is not limited to this. In short, if a person skilled in the art is inspired by this description and designs a similar structure and embodiment without departing from the spirit of the present invention, such design should fall within the protection scope of the present invention.

Claims

1. A method for real-time optimization and collaborative linkage of dynamic evacuation paths in subway 3D imaging for multiple disasters, characterized by: Including the following steps: S10: Collect disaster type information, environmental status information and crowd density information in real time through various types of sensing devices deployed in subway stations, and summarize them to generate environmental sensing information to describe the current emergency scenario; S20: Construct an evacuation network model containing nodes and edges based on the acquired subway spatial structure information, and optimize the evacuation network model in real time based on environmental perception information, thereby generating dynamic network model information; S30: Based on the dynamic network model information, the dynamic weight coefficients of each side are generated, the road segment escape information of each side is calculated, and the optimal evacuation route information is generated with the goal of minimizing the total escape time of the path. S40: Convert the optimal evacuation path information into a set of control instructions for the three-dimensional imaging display lights and transmit them to each display light node to generate three-dimensional guidance information for presenting three-dimensional guidance; S50: Generate a collaborative control strategy based on the optimal evacuation route information, and dynamically update the collaborative control strategy according to real-time monitoring information.

2. The method for real-time optimization and collaborative linkage of dynamic evacuation paths for subway three-dimensional imaging oriented to multiple disasters as described in claim 1, characterized in that: The various types of sensing devices include smoke sensors, temperature sensors, toxic gas sensors, visibility meters, and AI cameras. Step S10 includes the following steps: S101: By deploying multiple types of sensing devices on subway platform pillars, transfer passage corners and stair entrances, smoke concentration information, temperature information, toxic gas concentration information, visibility information and crowd density information are collected simultaneously, and the original sensing information from multiple sources is obtained. S102: The original sensing information is transmitted to the edge computing node, and the random forest algorithm is used to perform data fusion and recognition to generate the environmental sensing information.

3. The method for real-time optimization and collaborative linkage of dynamic evacuation paths for subway three-dimensional imaging oriented to multiple disasters as described in claim 1, characterized in that: Step S20 includes the following steps: S201: Based on the spatial structure information of the subway, the subway entrances and exits, passage intersections, stairwells and elevator entrances are abstracted as nodes, and the passages, stairs and elevators connecting the nodes are abstracted as edges to construct an evacuation network model. S202: Based on the environmental perception information, update the real-time pedestrian flow density attribute, disaster impact range attribute, and real-time visibility attribute of each node and edge to generate the dynamic network model information.

4. The method for real-time optimization and collaborative linkage of dynamic evacuation paths for subway three-dimensional imaging oriented towards multiple disasters as described in claim 1, characterized in that: Step S30 includes the following steps: S301: Based on the dynamic network model information, calculate the disaster impact coefficient, population density coefficient, channel basic coefficient and visibility coefficient of each edge respectively, and perform fusion calculation to generate the dynamic weight coefficient of each edge; S302: Based on the dynamic weight coefficient, the physical length information of each side and the basic traffic speed information, generate the road segment escape information corresponding to each side according to the preset road segment escape time calculation formula; S303: The improved Dijkstra algorithm is adopted to traverse all nodes in the evacuation network model with the goal of minimizing the total escape time of the path, solve the optimal path from each node to the safe exit, and generate the optimal evacuation path information.

5. The method for real-time optimization and collaborative linkage of dynamic evacuation paths for subway three-dimensional imaging oriented towards multiple disasters as described in claim 4, characterized in that: Step S301 includes the following steps: S3011: Obtain the straight-line distance information from the center point of each road segment to the disaster center point, real-time toxic gas concentration information, and disaster type correction coefficient; generate the disaster impact coefficient according to the preset calculation formula; the specific formula is as follows: in, The straight-line distance from the center point of the road segment to the center of the disaster. The maximum distance affected by the fire source is set at 100 meters. This refers to the real-time concentration of toxic gas within the road section. The human body's tolerance threshold for toxic gas concentration is set at 5 mg / m³. The following are the disaster type correction factors: fire = 1.2, earthquake = 1.5, toxic gas leak = 2.0, no disaster = 1.0; =1.0、 =1.2、 =0.5 is the calibration parameter; The value range is from 1.0 to 6.0; S3012: Based on the ratio of real-time crowd density information of the road segment to a preset density threshold, the crowd density coefficient is generated according to a preset crowd density coefficient calculation formula; the specific formula is as follows: in, This represents the real-time population density of the road segment. With a density of 3 people per square meter, β = 1.0 when the density does not exceed the threshold, and increases with density after exceeding the threshold. S3013: Based on the actual width and slope information of the passage, the passage basic coefficient is generated according to the preset passage basic coefficient calculation formula; the specific formula is as follows: in, The actual width of the channel. Set to 3 meters; The slope of the passage. Set to 0.02; The value range is controlled between 0.8 and 1.2, and the larger the value, the stronger the passage capability; S3014: Based on the ratio of real-time visibility information of the road segment to a preset visibility standard value, the visibility coefficient is generated according to a preset visibility coefficient calculation formula; the specific formula is as follows: in, Real-time visibility of the road segment. The standard value for normal visibility is set at 50 meters. When visibility is ≥ 50 meters... =1.0, which decreases as visibility decreases below 50 meters, with a minimum retention of 0.

3.

6. The method for real-time optimization and collaborative linkage of dynamic evacuation paths for subway three-dimensional imaging oriented to multiple disasters, as described in any one of claims 4-5, is characterized in that: In step S302, the preset formula for calculating the escape time of the road segment is as follows: in, Let be the physical length of the i-th channel segment. Set to 1.2 m / s, The bigger, The smaller the section, the longer the escape time. The total escape time T for the route is calculated by summing the escape times of all segments included in the path. The specific calculation formula is as follows: Where n is the total number of road segments included in the optimal path, the algorithm aims to find the minimum value of T from each evacuation node to the safe exit, while avoiding disaster nodes and congested road segments.

7. The method for real-time optimization and collaborative linkage of dynamic evacuation paths for subway three-dimensional imaging oriented to multiple disasters as described in claim 4, characterized in that: Step S303 includes the following steps: S3031: Initialize the total escape time of all evacuation nodes to infinity, set the total escape time of only the safe exit nodes to 0, and create a priority queue based on a binary heap to add all nodes to the queue. S3032: Take the node with the shortest total escape time from the priority queue as the current node, and traverse all adjacent nodes of the current node; S3033: For each of the adjacent nodes, a candidate total escape time is generated based on the sum of the current node's total escape time and the escape time of the connecting edge. If the candidate total escape time is less than the current total escape time of the adjacent node, the total escape time and predecessor node information of the adjacent node are updated, and the updated node is added back to the priority queue. S3034: Repeat steps S3032 to S3033 until the priority queue is empty or all safety exit nodes have been processed. Backtrack from each evacuation node to the safety exit node through the predecessor node information to generate the optimal evacuation path information.

8. The method for real-time optimization and collaborative linkage of dynamic evacuation paths for subway three-dimensional imaging oriented to multiple disasters as described in claim 1, characterized in that: Step S40 includes the following steps: S401: Convert the optimal evacuation route information into a set of control instructions containing equipment identification information, guide color parameter information, arrow direction encoding information, remaining distance value information and effective duration information, and transmit it to each three-dimensional imaging display light; S402: After receiving and parsing the control instruction set, each 3D imaging display lamp generates display control information for driving the display lamp to present the 3D arrow shape, color encoding status, and remaining distance projection.

9. The method for real-time optimization and collaborative linkage of dynamic evacuation paths for subway three-dimensional imaging oriented to multiple disasters as described in claim 1, characterized in that: Step S50 includes the following steps: S501: Generate a collaborative control strategy based on the optimal evacuation path information, which includes the control parameters and action timing of each collaborative system device, and send it to the existing subway system; S502: Real-time collection of crowd density and personnel behavior information in the passage through AI cameras; when congestion or reverse behavior is detected, automatic path recalculation is triggered, and the optimal evacuation path information and the collaborative control strategy are updated based on the recalculation results. S503: After the sensor continuously monitors and detects that the environmental parameters have returned to normal and meet the preset disaster relief conditions, it automatically determines that the disaster has been relieved and generates system recovery instructions and indicator light reset control information.

Images