Fire emergency evacuation path dynamic planning and intelligent guiding system and method

Abstract

The application relates to the technical field of wisdom fire fighting and public safety emergency, and discloses a fire emergency evacuation path dynamic planning and intelligent guiding system and method, which comprises a multi-source heterogeneous environment perception network, a personnel real-time positioning and state perception network, an edge computing and convergence node, a central decision and path planning platform and an intelligent evacuation guiding execution network. Through the construction of a full-link closed-loop system architecture of multi-source perception, edge computing, central decision and intelligent execution, the application realizes real-time, comprehensive and accurate perception of the evacuation environment and personnel situation, and solves the core problems of information perception lag, lack of data support for decision-making and insufficient linkage of various subsystems.

Description

Technical Field

[0001] This invention relates to the fields of smart fire protection and public safety emergency technology, specifically to a dynamic planning and intelligent guidance system and method for fire emergency evacuation routes. Background Technology

[0002] Fire is one of the major disasters threatening public safety, and efficient emergency evacuation is crucial to reducing casualties. Traditional fire evacuation guidance mainly relies on pre-set fixed evacuation signs (such as constantly lit or flashing "safety exit" indicator lights) and emergency broadcasts. However, this static and passive guidance method reveals serious shortcomings in the complex and ever-changing real-world fire scenarios: The guidance strategies are rigid and unable to adapt to dynamic disaster situations: the paths indicated by fixed signs are preset and cannot be changed according to real-time changes in the disaster situation, such as the location of the fire source, the spread of smoke, and blockages in passageways. A fire may cause the preset "safe" path to become dangerous, but static signs will still guide people to dangerous areas, creating "misleading" situations and exacerbating casualties.

[0003] The lack of a holistic perspective can easily lead to localized congestion and conflicts: traditional methods cannot perceive the real-time distribution and flow of people within a building. Large numbers of people may blindly rush towards the nearest exit, causing severe congestion at key points such as evacuation staircases and entrances, creating new risk points, and even triggering stampedes, greatly reducing overall evacuation efficiency.

[0004] Ignoring individual differences and lacking personalized guidance: All personnel receive the same instructions, failing to consider the differences in mobility (such as the elderly, children, and disabled people), familiarity with the building, and real-time physiological conditions of different individuals, and failing to provide personalized paths for those with mobility impairments to avoid obstacles such as stairs.

[0005] Information perception is lagging and decision-making lacks data support: Emergency commanders have difficulty quickly, comprehensively and accurately grasping key situational information such as the fire point, the spread trend, and the location of trapped personnel. Decisions are mostly based on experience, making it difficult to formulate scientific and efficient overall evacuation strategies, let alone intervene and optimize the evacuation process in real time.

[0006] The system suffers from low intelligence and insufficient coordination among subsystems: fire alarm, video surveillance, emergency lighting, broadcasting, and access control systems often operate independently, lacking information sharing and forming "information silos." There is a lack of a "smart brain" capable of integrating multi-source information, performing intelligent analysis, and collaboratively controlling various guidance terminals to execute unified dynamic strategies.

[0007] In view of this, we propose a dynamic planning and intelligent guidance system and method for fire emergency evacuation routes. Summary of the Invention

[0008] The purpose of this invention is to provide a dynamic planning and intelligent guidance system and method for fire emergency evacuation routes, so as to solve the problems existing in the above-mentioned background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A dynamic planning and intelligent guidance system for fire emergency evacuation routes, the system comprising: A multi-source heterogeneous environmental sensing network is used to collect fire-related environmental parameters and on-site situation information in real time. A real-time personnel positioning and status awareness network is used to acquire the three-dimensional spatial location, movement speed, density distribution, and physiological status information of personnel in a building in real time, as well as feedback from user terminals. Edge computing and aggregation nodes are used to aggregate, preprocess, and fuse local real-time data from the environmental perception network and personnel perception network, perform preliminary risk assessment and data compression, and report structured regional situation summaries to the central decision-making platform. The central decision-making and route planning platform receives and integrates global situational data from various edge computing nodes. Based on the built-in building information model, dynamic risk assessment model, multi-constraint route planning algorithm and evacuation strategy knowledge base, it generates and dynamically updates personalized optimal evacuation route plans for each affected evacuation unit in real time. The intelligent evacuation guidance execution network is used to receive and execute guidance instructions from the central decision-making platform, convert dynamically planned evacuation routes into visible, audible, and perceptible physical guidance signals, and provide real-time navigation to personnel.

[0010] Preferably, the dynamic risk assessment model running in the central decision-making and path planning platform is used to quantify the real-time hazard level of each spatial unit within the building during the fire development process. Its core is to calculate the comprehensive risk value of each grid unit at time t. This value is a weighted composite of three factors: fire threat, evacuation obstruction, and structural risk. ; in, Fire threat factors are calculated based on smoke, heat, flame, and CO data and fire spread models, representing the immediate threat level of heat, smoke, and toxic substances. To mitigate evacuation obstacles, the unit's personnel density, movement speed, passageway capacity, and lighting conditions were taken into account. As a structural risk factor, the fire resistance limit of building components, possible collapse risk, and availability of evacuation facilities were considered. , , These are the dynamic weighting coefficients for each factor, which are adjusted according to the stage and situation of the fire.

[0011] Preferably, the multi-constraint path planning algorithm running in the central decision-making and path planning platform aims to find a feasible path with the minimum overall cost within a finite time for each person or group to be evacuated in a dynamic risk field, from the starting point to the set of safe exits. The overall cost C of this path is jointly determined by the path's physical length, cumulative risk exposure, expected travel time, and impact on overall evacuation efficiency. ; in, Let be the geometric length of the path element. This is a real-time risk value. The predicted travel delay is based on the current pedestrian density. For path Other personnel paths The costs of interaction effects (such as conflict, congestion).

[0012] Preferably, the multi-constraint path planning algorithm adopts a hierarchical planning strategy that integrates the potential field method and conflict resolution: First, a "virtual potential field" is constructed based on the dynamic risk field R, where the fire source and high-risk area are high repulsive potentials, and the safety exit is a high gravitational potential, generating an initial individual gravitational path for each simulation entity; then, at the group level, through a "microscopic traffic flow simulation and conflict prediction module," the congestion and conflict at the intersection points that may occur in the spacetime of all individual paths are simulated; finally, a collaborative strategy based on time window reservation is adopted to guide the speed, stagger the peak scheduling, or replan the individual paths that may conflict, generating a final conflict-free collaborative evacuation path scheme.

[0013] Preferably, the variable information flags in the intelligent evacuation guidance execution network and the guidance information received by the mobile user terminal application are highly contextualized, and their content generation follows the following rules: a) When the risk value R is lower than the threshold within a certain distance ahead of the planned path, the normal directional guidance and distance information are displayed; b) When a sudden increase in risk is detected ahead of the path (such as flame blockade), immediately update the display of alternative paths and alert with a flashing highlight and a warning sound; c) When the population density on a planned evacuation route exceeds the safety threshold, a message “Congestion ahead, it is recommended to slow down or wait” is sent to the terminals of people who have not yet entered the section, and diversion routes are recommended to people behind. d) For individuals with mobility difficulties (such as the elderly or injured) identified through location information or physiological feedback, the path received by their terminal will automatically avoid stairs, prioritize elevators or refuge floors, and trigger manual assistance notifications at key nodes.

[0014] Preferably, the system also includes a digital twin simulation and deduction module, which is connected to the central decision-making platform. After the fire alarm is confirmed and before the large-scale evacuation is initiated, the module initializes the simulation environment using the current real-time data, quickly deduces the execution effect of various preset or dynamically generated evacuation plans in the next few minutes, predicts possible bottlenecks and risks, and feeds back key indicators in the deduction results (such as expected evacuation time, maximum congestion point, and number of high-risk exposures) to the decision-making platform to assist it in selecting the comprehensive and optimal global evacuation strategy from multiple feasible options.

[0015] Preferably, the digital twin simulation and deduction module adopts an agent-based micro-simulation model, and the behavior rules of each simulation entity are driven by a basic behavior library and a real-time guidance strategy. The deduction process supports "if-then" analysis, allowing commanders to manually set or modify key variables such as fire development parameters and the opening and closing status of evacuation doors, and observe the changes in the deduction results in real time, thereby evaluating the effectiveness of different intervention measures and providing decision support for emergency command.

[0016] Preferably, the edge computing and aggregation node has local autonomous guidance and decision-making capabilities. When communication with the central decision-making platform is interrupted, it can independently control the intelligent evacuation guidance equipment in its local area based on the locally aggregated environmental and personnel data and according to the preset offline decision-making logic, execute the optimal evacuation guidance for the area, and perform status synchronization and strategy calibration with the central platform after communication is restored.

[0017] Preferably, the system supports precise search and rescue assistance functions in the later stages of evacuation or after a disaster. Specifically, the system records and analyzes the last known personnel location information, dynamically planned path instructions, and a list of personnel who have not reached the safe area. Combined with building information modeling, it generates a heat map of possible trapped areas and search and rescue path suggestions, and guides search and rescue personnel through mobile terminals. At the same time, it can send vibration, strong light, and other signals to smart devices in potentially trapped areas to assist in positioning.

[0018] A method for dynamic planning and intelligent guidance of fire emergency evacuation routes, characterized by the following steps: S1. Real-time collection of fire-related environmental parameters and on-site situation information through various types of sensors; S2. Real-time acquisition of the three-dimensional spatial location, movement speed, density distribution, and physiological status information of people in the building through indoor positioning infrastructure and mobile user terminals; S3. Aggregate, preprocess, and fuse local real-time data from the environmental perception network and personnel perception network, perform preliminary risk assessment and data compression, and report a structured regional situation summary to the central decision-making platform; S4. Receive and integrate global situational data from each edge computing node, and based on the built-in building information model, dynamic risk assessment model, multi-constraint path planning algorithm and evacuation strategy knowledge base, generate and dynamically update personalized optimal evacuation path schemes for each affected evacuation unit in real time. S5. Receive and execute guidance instructions from the central decision-making platform, convert the dynamically planned evacuation routes into visible, audible, and perceptible physical guidance signals, and provide real-time navigation to personnel.

[0019] By employing the above technical solution, this invention provides a dynamic planning and intelligent guidance system and method for fire emergency evacuation routes. It possesses at least the following beneficial effects: This invention constructs a closed-loop system architecture encompassing multi-source sensing, edge computing, central decision-making, and intelligent execution. This architecture enables real-time, comprehensive, and accurate perception of the evacuation environment and personnel situation, addressing core issues such as lagging information perception, lack of data support for decision-making, and insufficient coordination among subsystems. By innovatively designing and applying a dynamic risk assessment model, it quantifies abstract fire threats, evacuation obstacles, and structural risks into a spatiotemporally continuous three-dimensional risk field, providing precise and dynamic decision-making basis for path planning and solving the problem of rigid guidance strategies unable to adapt to dynamic disaster situations. Furthermore, it develops a hierarchical planning strategy that integrates the potential field method and conflict resolution. The corresponding multi-constraint path planning algorithm realizes a paradigm shift from a single shortest path to globally coordinated optimal evacuation, effectively solving the bottleneck problem of lacking a global perspective and easily causing local congestion and conflict. Through contextualized intelligent guidance execution network and support for personalized path dynamic planning, it achieves a leap from indiscriminate broadcasting to precise contextualized and personalized guidance, solving the problem of ignoring individual differences and lacking personalized guidance. By integrating digital twin simulation and inference modules, it provides a powerful decision support tool for emergency command, realizing an upgrade from experience-based decision-making to simulation-assisted decision-making, further strengthening the system's ability to cope with complex situations. Attached Figure Description

[0020] The accompanying drawings, which are provided to further illustrate the invention, constitute a part of this application: Figure 1 This is a schematic diagram of the structure of the dynamic planning and intelligent guidance system for fire emergency evacuation routes of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] See Figure 1 This invention provides a dynamic planning and intelligent guidance system and method for fire emergency evacuation routes, including: A multi-source heterogeneous environmental sensing network 101 is used to collect fire-related environmental parameters and on-site situation information in real time. It should be noted that the multi-source heterogeneous environmental sensing network consists of various types of sensor nodes deployed in different areas of the building. The sensor nodes include at least smoke sensors, temperature sensors, flame detectors, carbon monoxide sensors, video image acquisition devices, and smart door magnets and indicator light status sensors used to monitor the status of evacuation routes. This network is a sensory system for dynamically sensing fire conditions. Its multi-source nature is reflected in the integration of detectors based on various physical principles, including smoke, heat, flame, and carbon monoxide (CO), enabling cross-verification of fire conditions from different dimensions (particulate matter, heat, light radiation, specific gases). Heterogeneity refers to the fact that these sensors may use different communication protocols (such as Modbus, BACnet, LoRa) and data formats. Integrated video image acquisition devices can be used for visual confirmation of fire points, smoke diffusion direction, and preliminary estimation of personnel distribution. Intelligent door and indicator light status sensors are used to monitor the physical accessibility of evacuation routes (e.g., whether fire doors have been illegally opened or malfunctioning, and whether emergency indicator lights are intact), ensuring that planned routes are feasible in practice. The deployment of this network must cover all fire compartments, evacuation routes, and key risk areas, forming a comprehensive monitoring system without blind spots.

[0023] The real-time personnel positioning and status awareness network 102 is used to acquire the three-dimensional spatial position, movement speed, density distribution and physiological status information of personnel in the building in real time, as well as the feedback from the user terminal. It should be noted that this network forms the foundation for personalized guidance and global crowd control, and consists of indoor positioning infrastructure and mobile user terminals deployed within buildings. The indoor positioning infrastructure can be built using technologies such as Wi-Fi, Bluetooth beacons (iBeacon), UWB (Ultra-Wideband), or visual analytics, providing users with real-time location data at the meter or even sub-meter level. In addition to receiving positioning signals, mobile user terminal applications (such as mobile apps and smart bracelets) can proactively report users' physiological status information, such as using accelerometers and heart rate sensors to determine if a user has fallen, is moving slowly, or is in a state of stress. Combining the location information of all users, the system can calculate the real-time distribution of crowd density and average movement speed in each area, which is a crucial input for assessing channel congestion risk and implementing crowd control strategies.

[0024] Edge computing and aggregation node 103 is used to aggregate, preprocess and fuse local real-time data from the environmental perception network and personnel perception network, perform preliminary risk assessment and data compression, and report a structured regional situation summary to the central decision-making platform. It's important to note that this node serves as the system's regional nerve center, distributed across floors and fire compartments, bearing the crucial responsibility of data reduction and real-time response. It first preprocesses raw data from various local sensors through protocol parsing, filtering, noise reduction, and spatiotemporal alignment. Next, it performs preliminary risk assessments, such as quickly determining the presence and approximate location of a local fire based on data from several adjacent smoke detectors in the area. More importantly, it performs data fusion and compression, refining massive amounts of raw readings (such as sensor data occurring several times per second and continuous video streams) into a higher-level, structured regional situation summary, such as "East corridor of Zone A: smoke concentration continues to rise, temperature is normal, current personnel density is moderate, and movement speed is normal." Reporting only this summary, rather than all the raw data, to the central platform significantly reduces network bandwidth and central computing pressure, and improves system response speed.

[0025] The central decision-making and route planning platform 104 receives and integrates global situational data from various edge computing nodes. Based on the built-in building information model, dynamic risk assessment model, multi-constraint route planning algorithm and evacuation strategy knowledge base, it generates and dynamically updates personalized optimal evacuation route plans for each affected evacuation unit in real time. It is important to note that this platform serves as the system's intelligent brain and decision-making core. The Building Information Model (BIM) is its static knowledge base, providing precise spatial information such as the building's 3D structure, material properties, evacuation routes, safety exits, and the location of fire-fighting facilities. The platform integrates summaries reported by all edge nodes to form a global, unified situational view. Based on this view, the platform runs a dynamic risk assessment model, generating a 3D dynamic risk field covering the entire building. Subsequently, it invokes a multi-constraint path planning algorithm to calculate, in real time, a safe path to a designated exit with the minimum overall cost for each identified person to be evacuated (or an evacuation group as a whole), starting from their current location and constrained by the dynamic risk field, personnel distribution, and building structure. The evacuation strategy knowledge base stores expert experience and contingency plan templates for different building types (such as shopping malls, hospitals, and high-rise residential buildings) and fire scenarios, guiding the formulation of global evacuation strategies (such as zoned evacuation and layered evacuation). All calculations and planning results need to be completed within seconds or even sub-seconds to cope with the rapid changes in the disaster situation.

[0026] The intelligent evacuation guidance execution network 105 is used to receive and execute guidance instructions from the central decision-making platform, convert dynamically planned evacuation routes into visible, audible, and perceptible physical guidance signals, and provide real-time navigation to personnel. It should be noted that this network serves as the executor of system decisions, consisting of controlled intelligent evacuation indicator lights, variable message signs, directional audible and visual alarms, access control systems, and mobile user terminal applications. It is responsible for translating virtual route plans into physically perceptible guidance for personnel. It comprises multiple types of controlled terminals: intelligent evacuation indicator lights can dynamically change arrow direction to indicate new safe routes; variable message signs can display text, graphics, and guidance information and warnings; directional audible and visual alarms can emit highly directional voice prompts and flashing lights in specific areas; access control systems can automatically unlock safety exits or close fire doors in hazardous areas upon command. Most importantly, the mobile user terminal application provides each user with one-to-one, augmented reality (AR) or map-based real-time navigation, accurately indicating turns, distances, and even road conditions ahead (congestion, danger). The synergy of multiple guidance methods ensures that guidance information is effectively conveyed under varying visibility, noise levels, and user conditions.

[0027] The dynamic risk assessment model running in the central decision-making and path planning platform is used to quantify the real-time hazard level of each spatial unit within a building during the fire development process. Its core is to calculate the comprehensive risk value of each grid unit at time t. This value is a weighted composite of three factors: fire threat, evacuation obstruction, and structural risk. ; in, Fire threat factors are calculated based on smoke, heat, flame, and CO data and fire spread models, representing the immediate threat level of heat, smoke, and toxic substances. To mitigate evacuation obstacles, the unit's personnel density, movement speed, passageway capacity, and lighting conditions were taken into account. As a structural risk factor, the fire resistance limit of building components, possible collapse risk, and availability of evacuation facilities were considered. , , These are the dynamic weighting coefficients for each factor, which are adjusted according to the stage and situation of the fire. It should be noted that this model is the scientific basis for path planning. It discretizes the building space into a grid, assigning each grid a comprehensive risk value R that varies over time. The fire threat factor F is the core; its calculation not only relies on real-time sensor data interpolation but also incorporates fire dynamics simulation to predict the spread of smoke and high temperatures, giving the risk field a certain degree of foresight. The evacuation obstruction factor O incorporates the influence of "people"; high density and low speed mean low passage efficiency and the risk of stampedes, while lighting failure exacerbates panic and the risk of falls. The structural risk factor S considers the "reliability" of the building; for example, load-bearing components near fire sources may lose strength under high temperatures, and some "safety exits" may fail to open due to shutter malfunctions. The weights of the three factors are... , , It is not fixed; for example, in the early stages of a fire. High weighting; when large-scale evacuation begins and bottlenecks form, The weights should be increased dynamically to guide the flow of traffic. This multidimensional risk field is the foundation for achieving dynamic and adaptive path planning.

[0028] The multi-constraint path planning algorithm running in the central decision-making and path planning platform aims to find a feasible path with the minimum overall cost within a finite time for each person or group to be evacuated in a dynamic risk field, from the starting point to the set of safe exits. The overall cost C of this path is determined by the path's physical length, cumulative risk exposure, expected travel time, and impact on overall evacuation efficiency. ; in, Let be the geometric length of the path element. This is a real-time risk value. The predicted travel delay is based on the current pedestrian density. For path Other personnel paths The costs of interactive impacts (such as conflicts and congestion); It should be noted that this algorithm is the core of this invention's global optimization guidance. It extends the traditional shortest path problem into a complex multi-objective, dynamic, collaborative optimization problem. The integral term of the objective function C represents the "quality" of the path itself: minimizing the geometric length ( Minimize the cumulative risk in the areas traversed ( Minimize the expected delay caused by congestion. The key lies in the last item. It introduces "social cost" or "system cost" to quantify the negative impact of an individual's path on the paths of all other people in the system. For example, if many people choose the same shortest path, it can lead to severe congestion (increased I value). The algorithm can reduce the global sum of I values ​​by adjusting the weights of α, β, γ, and λ, or by directly planning slightly longer but smoother paths for some people, thereby avoiding the "fallacy of composition" and achieving optimal evacuation efficiency at the system level. The algorithm needs to solve this high-dimensional optimization problem in real time and quickly, often employing heuristic algorithms (such as the improved A* algorithm combined with the potential field method) or distributed cooperative algorithms.

[0029] The multi-constraint path planning algorithm employs a hierarchical planning strategy that integrates the potential field method and conflict resolution: First, a "virtual potential field" is constructed based on the dynamic risk field R, where the fire source and high-risk area are high repulsive potentials, and the safety exit is a high gravitational potential, generating an initial individual gravitational path for each simulation entity; then, at the group level, a "microscopic traffic flow simulation and conflict prediction module" is used to pre-simulate the congestion and conflicts that may occur at the intersection points of all individual paths in time and space; finally, a collaborative strategy based on time window reservation is adopted to guide the speed, stagger peak scheduling, or locally replan for individual paths that may conflict, generating a final conflict-free collaborative evacuation path scheme; It's important to note that this layered strategy cleverly balances computational efficiency and planning quality. The first layer (potential field method) quickly generates an initial path for each person that is roughly oriented correctly and avoids static / dynamic risk zones, reacting rapidly. The second layer (conflict prediction and resolution) is crucial to ensuring the feasibility of the solution. The system utilizes a microscopic traffic flow simulation model, treating everyone as an intelligent agent, using their initial path and typical movement speed (considering individual differences) as input to quickly extrapolate the movement of the crowd in the next few tens of seconds to minutes, accurately predicting which stairwell or lobby will experience intersections and congestion. The third layer (coordination strategy) then implements scheduling for the predicted conflict points: for example, assigning different passage "time windows" to groups A and B who are about to reach the intersection, prompting group A to "speed up" and group B to "wait a moment" via their terminal apps; or planning an alternative route for some people. This is similar to designing a sophisticated "air traffic control" plan for the entire evacuation process, thereby fundamentally avoiding congestion and stampedes.

[0030] The variable information flags in the intelligent evacuation guidance execution network and the guidance information received by the mobile user terminal application are highly contextualized, and their content generation follows the rules below: a) When the risk value R is lower than the threshold within a certain distance ahead of the planned path, the normal directional guidance and distance information are displayed; b) When a sudden increase in risk is detected ahead of the path (such as flame blockade), immediately update the display of alternative paths and alert with a flashing highlight and a warning sound; c) When the population density on a planned evacuation route exceeds the safety threshold, a message “Congestion ahead, it is recommended to slow down or wait” is sent to the terminals of people who have not yet entered the section, and diversion routes are recommended to people behind. d) For individuals with mobility difficulties (such as the elderly or injured) identified through location information or physiological feedback, the path received by their terminal will automatically avoid stairs, prioritize elevators or refuge floors, and trigger manual assistance notifications at key nodes. It should be noted that this rule set defines "how the system speaks," reflecting the initiative, precision, and humanistic care of intelligent guidance. It is no longer simply broadcasting "evacuate to the safety exit," but rather providing "action prescriptions" based on the real-time situation faced by each individual. Rule a) provides routine guidance. Rules b) and c) embody dynamic risk avoidance and load balancing: b) addresses sudden environmental risks, and c) addresses risks caused by individuals themselves (congestion). The system needs to continuously conduct proactive monitoring, intervening in advance for those affected once a section of road is predicted to become dangerous or congested. Rule d) is the core of differentiated services. The system identifies special groups through data analysis (slow, irregular movement trajectories) or device feedback (fall detection, abnormal heart rate), not only planning accessible routes but also sending assistance requests to nearby security personnel or volunteer terminals through the system backend, achieving "human-machine collaborative" rescue.

[0031] The system also includes a digital twin simulation and deduction module, which is connected to the central decision-making platform. After a fire alarm is confirmed and before a large-scale evacuation is initiated, the simulation environment is initialized using the current real-time data. The system can quickly deduce the execution effects of various preset or dynamically generated evacuation plans in the next few minutes, predict potential bottlenecks and risks, and feed back key indicators from the deduction results (such as estimated evacuation time, maximum congestion point, and number of people exposed to high risk) to the decision-making platform to assist it in selecting the comprehensive and optimal global evacuation strategy from multiple feasible options. It's important to note that this module serves as a "sandbox" and "oracle" for advanced command and decision-making. In critical emergency situations, commanders often find it difficult to judge the merits of prioritizing evacuation based solely on intuition. The digital twin module, based on real-world data of the current fire situation and personnel distribution, rapidly clones a virtual evacuation environment. Within tens of seconds, it can simulate the execution of different evacuation strategies (such as different exit allocation schemes and evacuation initiation sequences), and output quantitative projection reports, such as: "Using scheme A, the total evacuation time is estimated at 8 minutes, but the No. 3 stairwell entrance will experience congestion exceeding 200% of the safe density after 5 minutes; using scheme B, the total evacuation time is 9 minutes, but there is no congestion at critical points, and the number of high-risk exposures is reduced by 60%." This provides commanders with objective and quantitative decision-making basis, enabling them to choose the "better" or even "optimal" option from multiple "feasible" schemes.

[0032] The digital twin simulation and deduction module adopts an agent-based micro-simulation model. The behavior rules of each simulation entity are driven by a basic behavior library and a real-time guidance strategy. The deduction process supports "what-if" analysis, allowing commanders to manually set or modify key variables such as fire development parameters and the opening and closing status of evacuation doors, and observe the changes in the deduction results in real time, thereby evaluating the effectiveness of different intervention measures and providing decision support for emergency command. It's important to note that the reliability of the simulation depends on the model. The agent-based micro-simulation model models each evacuee as an agent with autonomous perception, decision-making, and movement capabilities. Its basic behavior library defines general behavioral logic, such as obstacle avoidance, conformity, and finding exit signs. Real-time guidance strategies are injected into the dynamic paths and contextualized instructions calculated by the system for each agent. The combination of these two aspects allows the simulation to reflect both the natural reactions of people in emergency situations (such as panic and blindness) and the intervention effect of the system's intelligent guidance. The "what if-then" analysis function further enhances its practicality. Commanders can interactively ask questions such as, "What if the fire spreads 20% faster?" or "What if we manually force open the backup fire escape?" The module can immediately re-demonstrate and display the results. This is equivalent to providing commanders with a "decision-making laboratory" that can be repeatedly tested and refined, greatly improving the foresight and scientific nature of emergency decision-making.

[0033] The edge computing and aggregation node has local autonomous guidance and decision-making capabilities. When communication with the central decision-making platform is interrupted, it can independently control the intelligent evacuation guidance equipment in its local area based on the locally aggregated environmental and personnel data and according to the preset offline decision-making logic, execute the optimal evacuation guidance for the area, and synchronize its status and calibrate its strategy with the central platform after communication is restored. It's important to note that a fire may damage network equipment and cause line interruptions, resulting in edge nodes losing contact with the central control center. In this situation, edge nodes with local autonomous decision-making capabilities become the "surviving" command center. Utilizing local sensor data and personnel location information, they run a simplified yet effective local risk assessment and path planning algorithm, independently controlling indicator lights, broadcasts, and other equipment in their area to guide personnel to exits or refuge rooms deemed safe locally. This ensures that, even in the worst-case scenario, the system still provides basic dynamic guidance functions, preventing a complete degradation into a static system due to central control failure. Once communication is restored, the node uploads all operational logs and decision data from the local period to the central platform for status synchronization. The central platform can then assess the rationality of local decisions and perform necessary policy calibrations to maintain global consistency.

[0034] The system supports precise search and rescue assistance functions in the later stages of evacuation or after a disaster. Specifically, the system records and analyzes the last known personnel location information, dynamically planned path instructions, and a list of personnel who have not reached the safe area. Combined with building information model, it generates a heat map of possible trapped areas and search and rescue route suggestions. It guides search and rescue personnel through mobile terminals and can send vibration, strong light and other signals to smart devices in possible trapped areas to assist in positioning. It is worth noting that by extending the system's value from "evacuation guidance" to "life search and rescue," a complete closed loop of emergency response is formed. The system continuously records the spatiotemporal trajectory of each person. Once the evacuation is basically completed, by comparing the "should-be" and "actually present" lists, combined with each person's last received route instructions and last location, the system can perform intelligent analysis: for example, if someone was last seen near the east stairwell, and the system had guided them to evacuate to that stairwell, the "probability of being trapped" in that stairwell and its adjacent areas will be significantly increased. Based on this, a heat map of the trapped area can indicate the priority search direction for the search and rescue team. The search and rescue route suggestions planned by the system for the search and rescue team will comprehensively consider the safety of the route and the efficiency of reaching the target area. In addition, by sending wake-up signals (bright flashing lights, specific frequency vibrations) to the remaining smart devices (such as smart lights and sensors) in the suspected area through the Internet of Things, it may provide a means of calling for help or locating trapped but conscious persons, or provide visual guidance for search and rescue personnel, greatly improving the targeting and efficiency of the search and rescue.

[0035] A method for dynamic planning and intelligent guidance of fire emergency evacuation routes, comprising: S1. Real-time collection of fire-related environmental parameters and on-site situation information through various types of sensors; S2. Real-time acquisition of the three-dimensional spatial location, movement speed, density distribution, and physiological status information of people in the building through indoor positioning infrastructure and mobile user terminals; S3. Aggregate, preprocess, and fuse local real-time data from the environmental perception network and personnel perception network, perform preliminary risk assessment and data compression, and report a structured regional situation summary to the central decision-making platform; S4. Receive and integrate global situational data from each edge computing node, and based on the built-in building information model, dynamic risk assessment model, multi-constraint path planning algorithm and evacuation strategy knowledge base, generate and dynamically update personalized optimal evacuation path schemes for each affected evacuation unit in real time. S5. Receive and execute guidance instructions from the central decision-making platform, convert the dynamically planned evacuation routes into visible, audible, and perceptible physical guidance signals, and provide real-time navigation to personnel. It is worth noting that data acquisition is continuous, edge processing and central decision-making are iteratively periodic (e.g., several times per second), and guidance execution is instantaneous. The entire process forms a dynamic feedback control loop of "monitoring -> analysis -> planning -> guidance -> re-monitoring (based on personnel movement and situational changes -> re-analysis / adjustment..."). It is this rapidly operating intelligent closed loop that enables the system to act like a commander with a "panoramic view" and "supercomputing power," continuously guiding hundreds or even thousands of personnel to evacuate along constantly optimized safe routes in a rapidly changing fire scene, achieving a leap from static plans to dynamic intelligence.

[0036] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A dynamic planning and intelligent guidance system for fire emergency evacuation routes, characterized in that, The system includes: A multi-source heterogeneous environmental sensing network is used to collect fire-related environmental parameters and on-site situation information in real time. A real-time personnel positioning and status awareness network is used to acquire the three-dimensional spatial location, movement speed, density distribution, and physiological status information of personnel in a building in real time, as well as feedback from user terminals. Edge computing and aggregation nodes are used to aggregate, preprocess, and fuse local real-time data from the environmental perception network and personnel perception network, perform preliminary risk assessment and data compression, and report structured regional situation summaries to the central decision-making platform. The central decision-making and route planning platform receives and integrates global situational data from various edge computing nodes. Based on the built-in building information model, dynamic risk assessment model, multi-constraint route planning algorithm and evacuation strategy knowledge base, it generates and dynamically updates personalized optimal evacuation route plans for each affected evacuation unit in real time. The intelligent evacuation guidance execution network is used to receive and execute guidance instructions from the central decision-making platform, convert dynamically planned evacuation routes into visible, audible, and perceptible physical guidance signals, and provide real-time navigation to personnel.

2. The dynamic planning and intelligent guidance system for fire emergency evacuation routes according to claim 1, characterized in that, The dynamic risk assessment model running in the central decision-making and path planning platform is used to quantify the real-time hazard level of each spatial unit within a building during the fire development process. Its core is to calculate the comprehensive risk value of each grid unit at time t. This value is a weighted composite of three factors: fire threat, evacuation obstruction, and structural risk. ； in, Fire threat factors are calculated based on smoke, heat, flame, and CO data and fire spread models, representing the immediate threat level of heat, smoke, and toxic substances. To mitigate evacuation obstacles, the unit's personnel density, movement speed, passageway capacity, and lighting conditions were taken into account. As a structural risk factor, the fire resistance limit of building components, possible collapse risk, and availability of evacuation facilities were considered. , , These are the dynamic weighting coefficients for each factor, which are adjusted according to the stage and situation of the fire.

3. The dynamic planning and intelligent guidance system for fire emergency evacuation routes according to claim 1, characterized in that, The multi-constraint path planning algorithm running in the central decision-making and path planning platform aims to find a feasible path with the minimum overall cost within a finite time for each person or group to be evacuated in a dynamic risk field, from the starting point to the set of safe exits. The overall cost C of this path is determined by the path's physical length, cumulative risk exposure, expected travel time, and impact on overall evacuation efficiency. ； in, Let be the geometric length of the path element. This is a real-time risk value. The predicted travel delay is based on the current pedestrian density. For path Other personnel paths The costs of interaction effects (such as conflict, congestion).

4. The dynamic planning and intelligent guidance system for fire emergency evacuation routes according to claim 3, characterized in that, The multi-constraint path planning algorithm employs a hierarchical planning strategy that integrates the potential field method and conflict resolution: First, a "virtual potential field" is constructed based on the dynamic risk field R, where the fire source and high-risk area are high repulsive potentials, and the safety exit is a high gravitational potential, generating an initial individual gravitational path for each simulation entity; then, at the group level, a "microscopic traffic flow simulation and conflict prediction module" is used to pre-simulate the congestion and conflicts that may occur at the intersection points of all individual paths in time and space; finally, a collaborative strategy based on time window reservation is adopted to guide the speed, stagger peak scheduling, or locally replan for individual paths that may conflict, generating a final conflict-free collaborative evacuation path scheme.

5. The dynamic planning and intelligent guidance system for fire emergency evacuation routes according to claim 1, characterized in that, The variable information flags in the intelligent evacuation guidance execution network and the guidance information received by the mobile user terminal application are highly contextualized, and their content generation follows the rules below: a) When the risk value R is lower than the threshold within a certain distance ahead of the planned path, the normal directional guidance and distance information are displayed; b) When a sudden increase in risk is detected ahead of the path (such as flame blockade), immediately update the display of alternative paths and alert with a flashing highlight and a warning sound; c) When the population density on a planned evacuation route exceeds the safety threshold, a message "Congestion ahead, it is recommended to slow down or wait" is sent to the terminals of people who have not yet entered the section, and diversion routes are recommended to people behind. d) For individuals with mobility difficulties (such as the elderly or injured) identified through location information or physiological feedback, the path received by their terminal will automatically avoid stairs, prioritize elevators or refuge floors, and trigger manual assistance notifications at key nodes.

6. The dynamic planning and intelligent guidance system for fire emergency evacuation routes according to claim 1, characterized in that, The system also includes a digital twin simulation and deduction module, which is connected to the central decision-making platform. After a fire alarm is confirmed and before a large-scale evacuation is initiated, the system uses real-time data to initialize the simulation environment, quickly deduces the execution effects of various preset or dynamically generated evacuation plans in the next few minutes, predicts potential bottlenecks and risks, and feeds back key indicators from the deduction results (such as estimated evacuation time, maximum congestion point, and number of people exposed to high risk) to the decision-making platform to assist it in selecting the comprehensive and optimal global evacuation strategy from multiple feasible options.

7. The dynamic planning and intelligent guidance system for fire emergency evacuation routes according to claim 1, characterized in that, The digital twin simulation and deduction module adopts an agent-based micro-simulation model. The behavior rules of each simulation entity are driven by a basic behavior library and a real-time guidance strategy. The deduction process supports "what-if" analysis, allowing commanders to manually set or modify key variables such as fire development parameters and the opening and closing status of evacuation doors, and observe the changes in the deduction results in real time, thereby evaluating the effectiveness of different intervention measures and providing decision support for emergency command.

8. The dynamic planning and intelligent guidance system for fire emergency evacuation routes according to claim 1, characterized in that, The edge computing and aggregation nodes have local autonomous guidance and decision-making capabilities. When communication with the central decision-making platform is interrupted, they can independently control the intelligent evacuation guidance equipment in their area based on the locally aggregated environmental and personnel data and according to the preset offline decision-making logic, execute the optimal evacuation guidance for the area, and synchronize the status and calibrate the strategy with the central platform after communication is restored.

9. The dynamic planning and intelligent guidance system for fire emergency evacuation routes according to claim 1, characterized in that, The system supports precise search and rescue assistance functions in the later stages of evacuation or after a disaster. Specifically, the system records and analyzes the last known personnel location information, dynamically planned path instructions, and a list of personnel who have not reached the safe area. Combined with building information modeling, it generates heat maps of possible trapped areas and search and rescue route suggestions, and guides search and rescue personnel through mobile terminals. At the same time, it can send vibration, strong light, and other signals to smart devices in potentially trapped areas to assist in positioning.

10. A method for dynamic planning and intelligent guidance of fire emergency evacuation routes operating in a system as described in any one of claims 1-9, characterized in that, Includes the following steps: S1. Real-time collection of fire-related environmental parameters and on-site situation information through various types of sensors; S2. Real-time acquisition of the three-dimensional spatial location, movement speed, density distribution, and physiological status information of people in the building through indoor positioning infrastructure and mobile user terminals; S3. Aggregate, preprocess, and fuse local real-time data from the environmental perception network and personnel perception network, perform preliminary risk assessment and data compression, and report a structured regional situation summary to the central decision-making platform; S4. Receive and integrate global situational data from each edge computing node, and based on the built-in building information model, dynamic risk assessment model, multi-constraint path planning algorithm and evacuation strategy knowledge base, generate and dynamically update personalized optimal evacuation path schemes for each affected evacuation unit in real time. S5. Receive and execute guidance instructions from the central decision-making platform, convert the dynamically planned evacuation routes into visible, audible, and perceptible physical guidance signals, and provide real-time navigation to personnel.

Images