Fire-fighting escape guiding method based on sensor

Abstract

The invention provides a fire-fighting escape guiding method based on a sensor, and the method comprises the steps: calculating a safe evacuation path and evacuation facilities through a graph theory algorithm according to a building structure model and the position information of escapers in combination with a fire spreading trend, a fire spreading speed and a fire spreading direction; according to the fire development situation and a preset trigger threshold value, the updating frequency of fire induction and evacuation alarms is dynamically adjusted, and it is ensured that real-time alarms transmitted to escapers have perspectiveness and accuracy; fire induction information and evacuation alarms are dynamically pushed to escapers at different positions through electronic display equipment or sound prompts, and evacuation paths and facilities which should not go to or be used at present are informed; according to the real-time position of the escaper and the fire development condition, the induction direction and the evacuation path are continuously optimized, and the accuracy and the real-time performance of user safety induction and evacuation alarm are ensured.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a fire escape guidance method based on a sensor. Background Art

[0002] The core technical challenge faced by the fire emergency evacuation system is how to accurately predict the development of the fire and intelligently guide the evacuation of personnel in a complex and changeable fire environment. Specifically, the system needs to dynamically simulate the spread trend, speed and direction of the fire based on limited sensor data, combined with the three-dimensional model of the building, through methods such as computational fluid dynamics. This process involves real-time processing of massive data and rapid calculation of complex models, which places extremely high demands on the system's computing power and algorithm efficiency. At the same time, the fire situation changes rapidly, and how to ensure the accuracy and timeliness of the prediction results is also a major problem. On this basis, the system also needs to quickly calculate the optimal evacuation path based on the development trend of the fire, combined with the building structure and personnel distribution. This process not only takes into account the static building layout, but also takes into account the dynamically changing fire and personnel flow. It is a complex optimization problem with multiple objectives and multiple constraints. In addition, how to convey the evacuation instructions to each escapee in a timely and accurate manner and make dynamic adjustments based on actual conditions is also an important challenge faced by the system. This not only involves the timeliness and reliability of information transmission, but also requires considering the psychological and behavioral characteristics of people in emergency situations to ensure that the instructions can be effectively executed. In general, the fire emergency evacuation system needs to make accurate judgments on complex and changing fire situations based on limited and uncertain information in a very short period of time, and provide personalized evacuation guidance for a large number of dispersed individuals. This places extremely high demands on the system's intelligence level and real-time response capabilities. Summary of the invention

[0003] The present invention provides a sensor-based fire escape guidance method, which mainly includes:

[0004] Based on the ambient temperature information, smoke concentration information and combustible gas content information collected by the sensor, combined with the three-dimensional space information inside the building, the heat transfer simulation and smoke flow simulation are constructed through the computational fluid dynamics model to obtain the fire spread trend, fire spread speed and fire expansion direction;

[0005] According to the fire spread trend, fire spread speed and fire expansion direction, the area most seriously affected by the fire is determined, and the location of the fire point, the location of the heat source and the subsequent locations that may induce risks are dynamically marked in the system;

[0006] According to the building structure model and the location information of the escapees, combined with the fire spread trend, fire spread speed and fire expansion direction, the safe evacuation path and evacuation facilities are calculated through graph theory algorithms;

[0007] Dynamically adjust the update frequency of fire induction and evacuation alarms according to the fire development and preset trigger thresholds to ensure that the real-time alarms delivered to escapees are forward-looking and accurate;

[0008] If the fire spread trend, fire spread speed or fire expansion direction changes, the fire-affected area, fire starting point location, heat source location and subsequent risk locations will be recalculated, and the evacuation routes and evacuation facilities will be updated;

[0009] Dynamically push fire guidance information and evacuation alarms to escapees at different locations through electronic display devices or sound prompts, informing them of the evacuation routes and facilities that should not be taken or used at present;

[0010] According to the real-time location of the escapees and the development of the fire, the guidance direction and evacuation path are continuously optimized to ensure the accuracy and real-time nature of user safety guidance and evacuation alarms.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0012] The present invention discloses an intelligent fire evacuation system based on multi-source information. The system collects environmental data through sensors, combines building three-dimensional information, and uses computational fluid dynamics models to simulate the fire spread trend. According to the simulation results, the present invention dynamically determines the most seriously affected areas and risk points, and marks them in the system. At the same time, combined with the building structure and personnel position, a graph theory algorithm is applied to calculate the safe evacuation path. The present invention can adjust the alarm frequency in real time according to the changes in the fire situation to ensure the foresight and accuracy of the information. When the fire development situation changes, the system will re-evaluate the risk area and update the evacuation plan. Through electronic equipment, the present invention pushes personalized fire induction information and evacuation alarms to personnel in different locations. The system will also continuously optimize the evacuation path according to the real-time location of the personnel to ensure the safety and timeliness of the guidance. The present invention realizes accurate evacuation guidance in fire situations by integrating multi-dimensional data, dynamic simulation and intelligent algorithms, and effectively improves the level of building fire safety management. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The present invention is a flow chart of a sensor-based fire escape guidance method.

[0014] Figure 2 It is a schematic diagram of a sensor-based fire escape guidance method of the present invention.

[0015] Figure 3 It is another schematic diagram of a sensor-based fire escape guidance method of the present invention. DETAILED DESCRIPTION

[0016] The technical scheme of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] like Figure 1-3 In this embodiment, a sensor-based fire escape guidance method may specifically include:

[0018] S101. Based on the ambient temperature information, smoke concentration information and combustible gas content information collected by the sensor, combined with the three-dimensional space information inside the building, a heat transfer simulation and a smoke flow simulation are constructed through a computational fluid dynamics model to obtain the fire spread trend, fire spread speed and fire expansion direction.

[0019] The sensor continuously monitors the ambient temperature information data in the building. If the temperature of any point in the ambient temperature information data is higher than the preset temperature threshold, the location information of the point and the smoke concentration data information and combustible gas content data information of the surrounding environment of the point are collected. According to the location information of the point whose temperature is higher than the preset temperature threshold and the smoke concentration and combustible gas content data information collected around the location, the parameters of the fire spread calculation model in the computational fluid dynamics model are constructed. The computational fluid dynamics model generates a three-dimensional space calculation grid of the building based on the geometric structure of the building, and performs differential calculation based on the three-dimensional space calculation grid to obtain a simulated flow field. A three-dimensional space is constructed based on the structural data information of the building, and a heat transfer calculation model is constructed by combining the material structure distribution information and thermal conductivity calculation data information inside the building with the thermal radiation calculation model and the thermal convection calculation model. The temperature field in the simulated flow field is calculated to obtain the heat data of each grid in the three-dimensional space. The heat data output by the heat transfer model is obtained, and the smoke concentration and combustible gas data information are integrated to perform differential calculation to construct a smoke flow model, and the spatial distribution of smoke particles in each grid is obtained. The smoke flow model includes a physical model of air entrainment and diffusion behavior, and simulates the state of smoke particles in each grid in the three-dimensional space at different times. According to the physical model in the smoke flow model and the state information of smoke particles at different times in each grid, a smoke concentration calculation model is established to predict the development trend of the fire, and the burning rate information and fire stage information of the fire in each grid are obtained, so as to make a real-time judgment on the state of the fire area in the building. If the fire in any grid is in the smoldering stage and the ambient temperature of the grid rises to the first threshold, the state of the smoke particles is discretized by constructing an adaptive hybrid difference format, and the burning rate of the surrounding grids is obtained as the combustion reaction rate. The extreme point is searched in the burning rate using the gradient descent method to obtain the fire spread speed at different extreme points. The burning rate is determined according to the gradient descent method, and the fire spread speed information of the extreme point is combined. The differential model and the finite element model are used to fuse the three-dimensional space geometric dimensions to construct a three-dimensional geometric fire simulation model. The temperature field and smoke field space are simulated by the computational fluid dynamics model, so as to perform numerical simulation to obtain the expansion direction of the fire in three-dimensional space.

[0020] Specifically, the monitoring of ambient temperature in buildings is the key to fire early warning. The sensor network can collect temperature data in real time. Once it exceeds a preset threshold, such as 60°C, further data collection is triggered. For example, when the temperature in the corridor on the 3rd floor of an office building reaches 65°C, the system immediately records the coordinates of the location and collects the surrounding smoke concentration and combustible gas content. These data provide basic parameters for building a fire spread model. The computational fluid dynamics model is the core of simulating the development of fire. Taking the office building as an example, a three-dimensional grid is generated according to the building structure diagram, and each grid unit represents a space of 0.5m×0.5m×0.5m. Using the finite difference method, parameters such as temperature and airflow velocity in each grid can be calculated to form a simulated flow field. The heat transfer model combines the properties of building materials, such as the thermal conductivity of the wall, to calculate the propagation of heat in space. Assuming that the wall of the corridor on the 3rd floor is gypsum board with a thermal conductivity of 0.17W / (m·K), the diffusion rate of heat to the adjacent area can be estimated based on this. The smoke flow model simulates the movement of smoke particles, taking into account the air entrainment and diffusion effects. For example, in windless conditions, smoke may rise at a speed of 0.3m / s. Predicting the development trend of fire is the key to prevention and control. By analyzing the temperature change rate, smoke concentration and other parameters of each grid, it can be determined which stage the fire is in. For example, if the temperature of a grid slowly rises to 200°C and the smoke concentration gradually increases, it may be in the smoldering stage. If the temperature suddenly rises rapidly to above 600°C, it may have entered the stage of full combustion. The adaptive mixed difference format can improve the simulation accuracy. In areas with large temperature gradients, such as near the fire source, a high-order difference method can be used; in areas with gentle temperature changes, a low-order method is used to ensure accuracy and improve computational efficiency. The gradient descent method is used to find the extreme points of the burning rate, which helps predict the direction of fire spread. For example, if the burning rate is found to reach a maximum value of 0.1kg / (m 2 ·s), which is likely to be the front of the fire spread. Finally, by integrating the outputs of each sub-model, a complete three-dimensional fire simulation model can be constructed. This allows firefighters to visualize the fire development process, predict dangerous areas, and optimize evacuation routes. For example, the simulation results show that the smoke concentration in a certain area on the 4th floor may exceed the standard in 10 minutes, so corresponding measures can be taken in advance. This fire simulation system based on multi-source data and complex models provides strong technical support for building fire safety.

[0021] According to the ambient temperature, smoke concentration and combustible gas content information, combined with the three-dimensional space information inside the building, the heat transfer and smoke flow simulation is constructed through the computational fluid dynamics model to obtain the fire spread trend, spread speed and expansion direction.

[0022] Obtain the three-dimensional spatial information inside the building, including room layout, door and window locations, vent locations, etc., and build a three-dimensional model of the building. Arrange temperature sensors, smoke concentration sensors, and combustible gas sensors inside the building to collect data on ambient temperature, smoke concentration, and combustible gas content in real time. Based on the three-dimensional model of the building and the data collected by the sensors, build a computational fluid dynamics model to simulate the heat transfer and smoke flow inside the building. Through the computational fluid dynamics model, calculate the temperature distribution, smoke concentration distribution, and combustible gas concentration distribution at different time points and different locations. According to the changes in temperature distribution, smoke concentration distribution, and combustible gas concentration distribution, determine the location and size of the fire, and determine the spread trend and expansion direction of the fire. Use machine learning algorithms, such as support vector machines or neural networks, to predict the spread speed of the fire and the spread range in the future based on historical fire data and current fire parameters. Combine the prediction information such as the fire spread trend, spread speed, and expansion direction with the three-dimensional model of the building to generate an intuitive fire spread visualization effect diagram, providing auxiliary decision-making basis for fire rescue.

[0023] Specifically, the construction of a three-dimensional model of a building is the basis for fire simulation. Taking a 10-story office building as an example, laser scanning technology can be used to obtain accurate spatial information inside the building. The scanner can capture structural features such as walls, doors, windows, and stairs with millimeter-level accuracy and generate point cloud data. Through point cloud processing software, the raw data can be converted into a three-dimensional mesh model to accurately restore the size and layout of each room. This high-precision model is crucial for subsequent fire simulation because it determines the path of airflow and heat transfer. The layout of the sensor network needs to consider coverage and data accuracy. In this office building, 20 temperature sensors, 10 smoke concentration sensors, and 5 combustible gas sensors can be installed on each floor. The temperature sensors should be evenly distributed, focusing on covering corridors and large office areas. Smoke sensors are arranged near the ceiling, and combustible gas sensors are installed around potential leak sources. All sensors transmit data in real time through a wireless network, and the sampling frequency can be set to once per second to ensure that small changes in the early stages of a fire are captured. Computational fluid dynamics (CFD) models are the core of simulating the development of fires. For this office building, the finite volume method can be used to discretize space into millions of tiny units. The mass, momentum and energy conservation equations are applied in each unit to simulate the movement of thermal airflow. Boundary conditions include wall heat conduction, airflow exchange at door and window openings, etc. The initial conditions are set based on the data collected by the sensor in real time. Through iterative solution, the temperature, velocity and pressure field distribution at any time and any position can be obtained. The calculation of temperature distribution takes into account three heat transfer modes: convection, conduction and radiation. For example, if an abnormally high temperature is found in an office on the 5th floor, the CFD model can simulate how the heat is conducted to the adjacent room through the wall and how the thermal airflow diffuses through the corridor. The distribution of smoke concentration needs to consider the generation, transportation and deposition of smoke particles. The simulation of the diffusion of combustible gas is similar, but the effect of chemical reactions on concentration needs to be considered. The judgment of the location and size of the fire takes into account multiple indicators. Temperature is the most direct indicator. For example, a temperature exceeding 200°C at a certain place may mean that an open flame has appeared. The rapid rise in smoke concentration is also an important signal at the beginning of a fire. The concentration of combustible gas exceeds the lower explosion limit, which indicates a major safety hazard. By analyzing the spatiotemporal distribution characteristics of these parameters, the location of the fire source and the development trend of the fire can be inferred. Machine learning algorithms play an important role in fire prediction. A model based on a support vector machine (SVM) can be constructed, with inputs including parameters such as current temperature, smoke density, combustible gas concentration, and building structural characteristics. By learning from historical fire data, the model can predict the spread speed and range of the fire in the next 30 minutes. For example, the model may predict that the fire will spread to the north at a speed of 0.5 meters per minute and break through the fire partition after 15 minutes. Visualization of fire spread is key to decision support. Combining the prediction results with a three-dimensional building model can generate a dynamic fire development map. For example, red is used to represent high temperature areas, yellow is used to represent areas with high smoke concentrations, and blue arrows indicate the direction of fire spread.This intuitive visualization enables fire commanders to quickly assess the situation and develop the best firefighting and evacuation strategies. The implementation process of the entire system reflects the characteristics of multidisciplinary integration. Building Information Modeling (BIM) technology provides the basis for spatial information collection. Internet of Things technology ensures the reliable transmission of real-time data. High-performance computing technology supports the rapid solution of complex CFD models. Artificial intelligence algorithms enhance the system's predictive capabilities. This comprehensive application not only improves the accuracy of fire warnings, but also points the way for the development of smart fire protection.

[0024] S102. Determine the area most seriously affected by the fire based on the fire spread trend, fire spread speed and fire expansion direction, and dynamically mark the location of the fire point, heat source and subsequent locations that may induce risks in the system.

[0025] The heat source distribution image of the fire scene is obtained by infrared thermal imaging. According to the location, area and temperature change of the heat source in the image, the fire starting point and the main heat source location are determined. The computer vision algorithm is used to analyze the continuously acquired heat source distribution images, and the spread direction and speed of the fire are determined by the shape, size and position changes of the heat source area. According to the spread speed and direction of the fire, combined with the internal structure diagram of the building and the distribution of combustibles, the spread trend of the fire in the future and the most likely affected areas are predicted. The fire spread trend is correlated with the distribution information of people in the building to determine the areas most affected by the fire and need to be evacuated first, and dynamically marked with warning colors in the system. According to the internal structure and material characteristics of the building, the support vector machine algorithm is used to identify the risk locations that may cause secondary disasters and mark them in the system. Taking into account the fire spread trend, speed, direction and distribution of risk locations, the ant colony optimization algorithm is used to plan the evacuation route, and the route information is dynamically pushed to the on-site commander and evacuation guide. Continuously track the development of fires, dynamically update fire spread prediction results, affected area divisions and evacuation route planning based on real-time on-site information, and provide accurate decision-making basis for fire fighting and personnel evacuation.

[0026] Specifically, infrared thermal imaging technology plays a key role in fire monitoring. Through infrared thermal imagers, heat source distribution images of the fire scene can be obtained, which can clearly show the location, area and temperature changes of the heat source. For example, when a fire occurs in a multi-story office building, the thermal imager may show that there is a high-temperature area in the southwest corner of the third floor, with a temperature of 800°C and an area of ​​about 10 square meters. This is likely to be the starting point of the fire. Computer vision algorithms can analyze the continuously acquired heat source distribution images and determine the direction and speed of fire spread by tracking the shape, size and position changes of the heat source area. For example, if the heat source area expands 2 meters to the east within 5 minutes, it can be inferred that the fire spreads eastward at a speed of 0.4 meters per minute. This dynamic analysis is crucial for predicting the development trend of the fire. Combined with the internal structure diagram of the building and the distribution of combustibles, the spread trend of the fire in the future and the most likely affected areas can be predicted. Assuming that the fire is spreading to a laboratory storing flammable chemicals, this area will be marked as a high-risk area and violent combustion is expected to occur in a short period of time. By correlating the fire spread trend with the distribution of people in the building, it is possible to identify the areas most affected by the fire and requiring priority evacuation. For example, if a meeting with 50 people is taking place in the conference room on the fourth floor and the fire is expected to affect the area in 20 minutes, the system will mark the conference room as a red warning zone to remind rescuers to evacuate first. Support vector machine algorithms can be used to identify risky locations that may cause secondary disasters. By inputting information such as material properties and stored items in various areas of the building, the algorithm can learn and identify potential danger points. For example, it may mark a storage room where compressed gas is stored as a high-risk area because high temperatures may cause the gas cylinder to explode. Ant colony optimization algorithms perform well in planning evacuation routes. The algorithm simulates the behavior of ants looking for food and can find the optimal evacuation path in a complex building structure. For example, considering the direction of fire spread, smoke diffusion, and the distribution of people, the algorithm may recommend using the east staircase as the main evacuation route while avoiding areas where collapse may occur. The system's dynamic update capability ensures real-time and accurate decision-making. If a new heat source is suddenly discovered or the speed of fire spread changes, the system will immediately recalculate and update the prediction results. This real-time adjustment allows rescuers to keep up to date and make the best decisions. For example, if a previously safe evacuation route is suddenly blocked by thick smoke, the system will immediately re-plan the route and notify on-site personnel. This comprehensive system provides accurate decision-making support for fire fighting and evacuation by integrating multiple advanced technologies. It can not only predict the development of fires, but also identify potential dangers and optimize evacuation strategies, greatly improving the efficiency and safety of firefighting and rescue.In complex fire scenarios, such intelligent assistance systems can become a powerful assistant to fire commanders, helping them make correct judgments in emergency situations and protecting life and property to the greatest extent possible.

[0027] S103. According to the building structure model and the location information of the escapees, combined with the fire spread trend, fire spread speed and fire expansion direction, a safe evacuation route and evacuation facilities are calculated through a graph theory algorithm.

[0028] Obtain the building structure model information, including the building's floor plan, number of floors, room distribution, etc., and construct the building's topological structure diagram. Obtain personnel location information, determine the room or area where each person is located, and map the personnel location to the building topological structure diagram. Obtain fire spread information, including the location of the fire, spread trend, spread speed and spread direction, and map the fire spread information to the building topological structure diagram. According to the building topological structure diagram, personnel location and fire spread information, construct a spatiotemporal weight graph, in which nodes represent rooms or areas, edges represent the connectivity between adjacent rooms or areas, and the weight of the edge represents the impact of the fire spread. Use the shortest path algorithm, such as the Dijkstra algorithm or the A algorithm, to calculate the shortest evacuation path from each personnel location to the safe exit on the spatiotemporal weight graph to obtain a personalized evacuation path. According to the building topological structure diagram and the evacuation path, determine the evacuation facilities that need to be opened, such as emergency lighting, signs, fire doors, etc., and control the status of these facilities. Monitor the fire spread and personnel evacuation progress in real time, dynamically adjust the evacuation path and evacuation facilities according to the latest situation, and ensure the safety and effectiveness of the evacuation.

[0029] Specifically, the building structure model information is the basis of the intelligent evacuation system. Taking a ten-story office building as an example, the system first obtains information such as its floor plan, number of floors, and room distribution. Through 3D modeling technology, an accurate topological structure diagram containing key elements such as corridors, stairs, and elevators is constructed. This digital model not only reflects the geometric characteristics of the physical space, but also contains the attribute information of important facilities such as doors, windows, and firewalls. Personnel positioning technology is the key to achieving personalized evacuation. Indoor positioning systems can use Bluetooth beacons, wireless LAN or ultra-wideband technology to accurately locate each person's location. For example, a meeting with 20 participants is taking place in the conference room on the fourth floor. The system can track the location of each participant in real time and map this information to the building topology diagram. The acquisition and analysis of fire spread information is the core of formulating evacuation strategies. Through infrared thermal imaging and a smoke detector network, the system can quickly locate the source of the fire and predict the development of the fire. Assuming that the fire starts in the storage room in the southwest corner of the third floor, the system will calculate that the fire spreads to the northeast at a speed of 0.5 meters per minute based on the intensity of the heat source, the smoke concentration, and the characteristics of the building materials. This dynamic information is updated in real time to the building topology diagram. The construction of the spatiotemporal weight graph is a key step in optimizing the evacuation path. The system abstracts each area of ​​the building as a node, and the channels between adjacent areas are represented as edges. The weight of the edge not only considers the physical distance, but also includes factors such as the impact of the fire and the smoke concentration. For example, although the corridor on the third floor is the shortest path to the safe exit, its weight value will increase sharply over time due to the spread of the fire, making the algorithm tend to choose a safer alternative route. The calculation of the personalized evacuation path adopts the improved A algorithm. Based on the traditional Dijkstra algorithm, this algorithm introduces a heuristic function to find the optimal path faster. For the 20 people in the conference room on the fourth floor, the system will calculate the safest and most efficient evacuation route according to each person's specific location, physical condition (such as people with limited mobility) and the current fire situation. Intelligent control of evacuation facilities is an important link to ensure smooth evacuation. According to the calculated evacuation path, the system automatically turns on the emergency lighting and signs in the corresponding area. For example, the lighting brightness of the stairwell on the east side of the fourth floor is increased to the maximum, and the ground luminous indicator strip leading to the staircase is activated at the same time. In addition, the system will control the opening and closing status of the fire doors according to the development of the fire to prevent the spread of the fire and create a safe passage. Real-time monitoring and dynamic adjustment mechanisms ensure the adaptability of the evacuation process. If the original evacuation route is suddenly blocked by thick smoke, the system will immediately recalculate the route and send updated instructions to relevant personnel through mobile devices. At the same time, according to the flow of personnel, the system can adjust the status of evacuation facilities, such as temporarily opening emergency exits that are usually not open to the public to relieve congestion pressure. This comprehensive system realizes intelligent management of complex fire scenes by integrating multiple advanced technologies.It can not only quickly respond to the changing fire situation, but also provide the most suitable personalized evacuation plan for each person, greatly improving the efficiency and safety of evacuation. In practical applications, this system can become a powerful support tool for fire command, helping decision makers make more accurate and timely judgments in emergency situations.

[0030] According to the building structure model and the location information of the escapees, combined with the fire spread trend, spread speed and spread direction, the safety factor, accessibility and congestion of the evacuation path are calculated through graph theory algorithm, so as to obtain the optimal safe evacuation path and evacuation facilities.

[0031] According to the building structure model, the layout information of the building is obtained, including the location coordinates and connectivity of rooms, corridors, stairs, etc., and the building topology map is constructed. The location of the fire and the location of the personnel are obtained, and the fire source node and the personnel node are marked in the building topology map. According to the fire spread trend, spread speed and spread direction, the cellular automaton model is used to simulate the dynamic spread process of the fire in the building, and the fire impact range at different times is obtained. In the building topology map, the nodes and edges affected by the fire are removed, and the building topology structure is updated. At the same time, according to the fire spread trend, the nodes and edges that may be affected in the future are predicted to reduce their safety factors. The Dijkstra shortest path algorithm is used to calculate the evacuation path with the personnel node as the starting point and the evacuation exit as the end point. In the calculation process, the safety factor, length and congestion of the path are comprehensively considered to obtain the optimal evacuation path with the minimum comprehensive cost. If the congestion of the path exceeds the preset threshold, the path is judged to be a congested path. According to the personnel density and evacuation flow, the cellular automaton model is used to simulate the evacuation process, predict the evacuation time of the congested path, and use it as the weighted item of the path cost. According to the distribution of evacuation paths in the building topology map, determine the safe exits and evacuation passages that need to be opened, and determine the configuration of evacuation facilities based on the flow of people on the path, such as safety signs, emergency lighting and evacuation guidance devices, so as to achieve efficient and orderly evacuation.

[0032] Specifically, the building topology is the basis for fire evacuation route planning. It obtains the layout information of the building, abstracts the spatial elements such as rooms, corridors, and stairs as nodes, and represents the connectivity between them as edges to form a graph structure. For example, a three-story office building may contain multiple offices, conference rooms, corridors, and stairs. These spatial elements are represented as nodes in the topology, and the doors and passages between them are represented as edges. The location of the fire and the location of the personnel are key information for evacuation route planning. In the building topology, the fire source node represents the location of the fire, and the personnel node represents the location of the personnel who need to be evacuated. This information can be obtained through the fire alarm system and the personnel positioning system. For example, a fire broke out in a conference room on the second floor of an office building, and there are employees distributed in various offices. This information will be marked on the corresponding nodes of the topology. The cellular automaton model is an effective tool for simulating the dynamic spread of fire. It divides the building space into several grid units, each unit represents a cell, and defines the cell state transition rules according to the physical characteristics of the fire spread and the surrounding environmental conditions. Through iterative calculation, the fire impact range at different times can be obtained. For example, assuming that a fire starts in a conference room, the spread of the fire in 10 minutes and 20 minutes can be predicted based on factors such as the room material and ventilation conditions. When updating the building topology, it is necessary to remove the nodes and edges affected by the fire and reduce the safety factor of the potentially affected areas. This ensures that the evacuation path avoids dangerous areas and improves the safety of evacuation. For example, if the fire has spread to a corridor, all edges connected to the corridor will be removed, and the safety factor of the adjacent room nodes will be reduced accordingly. The Dijkstra shortest path algorithm is an effective method for calculating the optimal evacuation path. It finds the path with the lowest comprehensive cost from the starting point to the end point through iterative calculation. In the fire evacuation scenario, the cost of the path includes not only the physical distance, but also the safety factor and congestion. For example, a shorter path close to the fire source may be abandoned due to its low safety factor, and replaced by a longer but safer path. For crowded paths, the cellular automaton model can more accurately simulate the evacuation process. Each person is regarded as a cell, and the rules for personnel movement are defined based on the surrounding environment and the location of other people. Through simulation, the actual evacuation time of crowded paths can be predicted, so that the path cost can be evaluated more reasonably. For example, a seemingly short corridor may actually take longer to evacuate due to high density of people, and other alternative paths may need to be selected. Finally, based on the calculated evacuation path, the safety exits and evacuation passages that need to be opened are determined, and the corresponding evacuation facilities are configured. For example, for paths with a large expected flow of people, the density of safety signs can be increased, emergency lighting can be strengthened, and evacuation guides can be deployed. These measures can effectively improve evacuation efficiency, reduce confusion and congestion, and ultimately achieve safe and orderly fire evacuation.

[0033] S104. Dynamically adjust the update frequency of fire induction and evacuation alarms according to the fire development situation and the preset trigger threshold to ensure that the real-time alarms delivered to escapees are forward-looking and accurate.

[0034] Obtain the current fire data of each location sent back by the sensing devices in each location of the building structure, and collect the evacuation positions sent back by the equipment on the personnel. Analyze the current fire data of each location through the constructed deep learning network to obtain the current fire degree of all locations in the building; set different fire induction trigger thresholds and personnel evacuation trigger thresholds according to the differences in personnel density and fire degree change trends. Set the calculation formula for the trigger threshold of each location, which needs to include variables such as location and space size. The induction threshold is used to induce personnel to evacuate to a certain area, usually taking a smaller value. The evacuation threshold is used to trigger personnel to escape from danger, usually taking a larger value. If the initial density of personnel is below the average density of the building, increase the induction threshold and reduce the evacuation threshold. Determine the threshold based on personnel and fire information; obtain a set of dynamic thresholds. According to the current location information of the personnel and the dynamic threshold, if the current location of the personnel is greater than the induction threshold, send the induction information to them. If it is greater than the evacuation threshold, send an evacuation alarm. Each alarm carries the target evacuation direction, and each alarm is sent at a preset time interval, which is represented by T. T is affected by the number of people around. The calculation formula of T is as follows: T = k 1 / (sum(D_{i})). D represents the distance, and D_{i} represents the distance between the person and other person i. i refers to each person that can be found, and the integers 1, 2, and 3 are used to represent the person in ascending order. The sum symbol ∑ sums each item D_{i} from i = 1 to i = m, where m represents the total number of people, and m changes in real time according to the current collected personnel positions. k is the proportional coefficient, which adjusts the degree of influence of the crowd density around the person on the sending time interval T, and is taken as 100. Calculate the real-time update frequency of the alarm sent at each current location. The moving average method is used to fit the fire data of all locations, and a fire change time series prediction model corresponding to each location is constructed; a model is constructed for each location; the model outputs the fire prediction for the next two time intervals T, and the model also outputs the residual of this time series data; the distribution range of the residual is determined by the random forest method, and a confidence interval of the fire prediction data is determined. Through this model, the possible fire level at each location in the future is determined, and combined with the judgment logic in step 3, the frequency of future alarm updates is estimated. If the estimated frequency is lower than the frequency threshold, the frequency is increased, otherwise it is reduced. According to the locations of all evacuees, the preset maximum number of people at each location, and the trend of fire severity at each location, the current optimal evacuation route for personnel is obtained. The optimal route is output as a real-time alarm, and each route has a number. Personnel choose the path according to the instructions of this number. Each person gets different route information, and the route number changes in real time, and the number changes continuously according to the fire situation and personnel location changes. According to the path number connected to each location, the number is formed into real-time induction information and sent down to determine the optimal route for the next time. Through the long short-term memory network combined with the fire information, the route changes in the future period are determined.Obtain the historical speed distribution information of evacuees on different evacuation routes under the model, obtain the age and physical fitness indicators of the current evacuees, and obtain the indication information of acceleration or deceleration of the evacuees based on the speed ratio between the evacuees and other personnel on the same route; if the age is lower than the average age, increase the evacuation speed indication value, otherwise reduce the speed indication value. If the density of people around a certain location is higher than a certain threshold, reduce the evacuation speed of people in this area. Collect the alarm history information returned by all devices with alarm functions, and use the gradient boosting decision tree method to obtain the rule base based on the actual route selection information of the personnel in the information and the corresponding fire information, select the optimal evacuation route according to different fire levels and different building structures, and determine the optimal route for sending real-time alarms. Automatically update the rule base every other day, and calculate the matching degree between the results obtained by this method and the results obtained by the fifth step method. Use a solution with a higher matching degree for output to form a route alarm that is finally sent to all personnel.

[0035] S105. If the fire spread trend, fire spread speed or fire expansion direction changes, the fire-affected area, fire point location, heat source location and subsequent risk locations are recalculated, and the evacuation routes and evacuation facilities are updated.

[0036] The fire monitoring system is used to obtain dynamic information such as the fire spread trend, spread speed and spread direction in real time. Based on the acquired fire dynamic information, the fire spread model is used to calculate the scope of the affected area. In the calculated affected area, the fire point and heat source location are determined by the heat source positioning algorithm. According to the fire spread trend and speed, combined with the layout of the building, the subsequent risk location of the fire is predicted. Based on the information of the affected area, the fire point, the heat source location and the risk location, the evacuation route is planned using the shortest path algorithm. According to the planned evacuation route, the guidance direction of the evacuation indication facility is dynamically adjusted and updated. If the fire spread changes, return to step 1 to re-acquire the fire dynamic information, iterate the subsequent steps, and realize the real-time update of the evacuation route and facilities.

[0037] Specifically, the fire monitoring system collects data in real time through devices such as temperature sensors, smoke detectors and infrared cameras to analyze the trend, speed and direction of fire spread. For example, in a multi-story office building, the system detected a sharp rise in temperature and an increase in smoke density in the northwest corner of the third floor, which was determined to be the initial point of the fire. Through continuous monitoring, it was found that the fire spread to the southeast at a speed of about 2 meters per minute. The fire spread model calculates the affected area based on factors such as building structure, materials and ventilation conditions combined with real-time data. In the above office building case, the model predicts that the fire will affect the entire third floor within 15 minutes and may spread upward to the fourth floor through the stairwell. The heat source location algorithm uses multi-point temperature data and image analysis to accurately determine the location of the fire, such as the electrical room in the northwest corner. At the same time, other high-temperature areas are identified, such as the adjacent file room. When predicting the subsequent risk locations of the fire, the system considers the building layout and the distribution of flammable materials. For example, it is predicted that the storage room adjacent to the fire point may become the next focus of fire spread because it stores a large number of paper documents. In addition, the elevator shaft located in the middle of the floor is also identified as a potential risk area, which may accelerate the upward spread of the fire. The shortest path algorithm comprehensively considers the affected area, the fire point, the location of the heat source and the risk location to plan the safest and most efficient evacuation route. In this case, the algorithm planned a path for people in the southeast area of ​​the third floor to avoid the fire: move east to the safe staircase and then evacuate downstairs. At the same time, a preventive evacuation route was formulated for people on the fourth floor to prevent them from using the central staircase that may be affected by the fire. According to the planned evacuation path, the system dynamically adjusts the evacuation indication facilities. For example, the LED sign in the corridor on the third floor was changed from pointing to the central staircase to the safe staircase on the east side to ensure that people are away from the fire. At the same time, the broadcasting system issues targeted instructions to guide people in different areas to take corresponding actions. The system continuously monitors changes in the fire situation. Once a change in the fire spread pattern is found, such as a new fire source on the fourth floor, the fire dynamic information is immediately re-acquired and the evacuation strategy is updated. This iterative mechanism ensures the real-time and accuracy of the evacuation path and facility guidance, greatly improving the efficiency of safe evacuation of people in fire situations. Through this dynamic adjustment, the system can cope with complex and changing fire situations and maximize the safety of people.

[0038] S106. Dynamically push fire guidance information and evacuation alarms to escapees at different locations through electronic display devices or sound prompts, informing them of evacuation routes and facilities that should not be traveled to or used at present.

[0039] Get the current location information of the escapee and determine the distribution of the escapee through the preset sensor network. If the location information of the escapee is successfully obtained, proceed to the next step, otherwise re-collect the location information. According to the real-time data of the fire monitoring system, determine the location of the fire and the spread trend, and generate fire dynamic information. If the fire information changes, update the fire dynamic information database to ensure the real-time information. Combine the fire dynamic information and the preset evacuation path planning to determine the safety of each evacuation path and generate a list of disabled paths. If a path is determined to be unsafe, add it to the disabled path list, and push the disabled path alarm to the escapee through the display device or sound device. Synchronously monitor the status of evacuation facilities, such as elevators, safety doors, etc., to determine the list of available and disabled facilities. If a facility is determined to be unavailable, add it to the disabled facility list, and push the disabled facility alarm to the escapee through the display device or sound device. Integrate the disabled path and disabled facility information to generate comprehensive guidance information. According to the current location of the escapee and the fire dynamic information, dynamically adjust the guidance information content to ensure that the information received by the escapee is targeted and real-time. Dynamically push comprehensive guidance information and evacuation alarms to escapees through electronic display devices or sound prompt devices. If the escapee approaches a prohibited path or facility, the alarm intensity is increased to prompt the escapee to adjust the evacuation direction. Continuously monitor the change of the escapee's position and the dynamic information of the fire, and update the guidance information and alarm content in real time. If the escapee's position or the fire situation changes, re-judge the safety of the path and facilities, update the guidance information and alarm, and ensure that the escapee always receives the latest evacuation guidance.

[0040] Specifically, obtaining the current location information of the escapee is a key step in the evacuation system. Through the preset sensor network, such as Wi-Fi positioning, RFID tags or infrared sensors, the distribution of the escapee can be accurately determined. For example, in an office building, multiple sensors are installed on each floor, and the smart devices carried by the escapee communicate with the sensors, and the system obtains their location information in real time. If the sensor on a certain floor fails to successfully obtain information, the system will automatically re-collect the location information to ensure the accuracy of the data. According to the real-time data of the fire monitoring system, such as temperature, smoke concentration and flame sensor, the location of the fire and the spread trend are determined. Assuming that the temperature sensor on a certain floor shows abnormally high temperature and the smoke sensor detects thick smoke, the system immediately generates fire dynamic information, showing that the fire originated from an office on the floor and quickly spread to the surrounding areas. If the fire information changes, such as the spread speed increases or the direction changes, the system immediately updates the fire dynamic information database to ensure the real-time nature of the information. Combining the fire dynamic information and the preset evacuation path planning, the system determines the safety of each evacuation path. For example, if a path is close to the fire source or the smoke concentration is too high, the system determines it as unsafe and adds it to the disabled path list. Push disabled path alarms to escapees through electronic display boards or sound devices, such as "Please note that the left corridor has been blocked. Please evacuate via the right corridor instead." Simultaneously monitor the status of evacuation facilities, such as elevators and safety doors. Assuming that an elevator cannot be used due to power outages due to a fire, the system will add it to the disabled facility list and push alarms to escapees through display devices and sound devices, such as "The elevator is out of service. Please use the stairs to evacuate." Ensure that escapees avoid unavailable facilities and choose safe evacuation paths. Integrate disabled path and disabled facility information to generate comprehensive guidance information. Dynamically adjust the guidance information content based on the escapee's current location and fire dynamic information. For example, if an escapee is near the fire source, the system will prioritize the evacuation path information away from the fire source to ensure that he evacuates quickly. Dynamically push comprehensive guidance information and evacuation alarms to escapees through electronic display devices or sound prompt devices. If the escapee approaches a disabled path or facility, the system will increase the alarm intensity, such as increasing the sound frequency or flashing lights, to prompt the escapee to adjust the evacuation direction. If an escapee mistakenly enters a prohibited path, the system will immediately sound a high-frequency alarm and display "Danger! Please return immediately and take a safe path". The system continuously monitors the changes in the escapee's position and the dynamic information of the fire, and updates the guidance information and alarm content in real time. If the escapee's position or the fire situation changes, such as the escapee moves to a new floor or the fire spreads to a new area, the system will re-judge the safety of the path and facilities, and update the guidance information and alarm. Ensure that the escapee always receives the latest evacuation guidance and improve evacuation efficiency. In this way, the system can adjust the evacuation strategy in real time and dynamically to ensure that the escapee evacuates quickly and safely in an emergency. It not only improves the evacuation efficiency, but also minimizes casualties and property losses.Each step is closely linked to each other, forming a strict logical relationship and chain of thinking to ensure the efficient operation of the system and the safety of the escapees.

[0041] S107. Based on the real-time location of the escapees and the development of the fire, the guidance direction and evacuation path are continuously optimized to ensure the accuracy and real-time nature of user safety guidance and evacuation alarms.

[0042] Obtain the real-time location information of the escapee, including the coordinates, movement speed and direction of the escapee, etc., through indoor positioning technology. Obtain real-time information on the development of the fire, including the location, spread speed and direction of the fire, etc., through fire sensors and monitoring systems. According to the location of the escapee and the development of the fire, machine learning algorithms such as reinforcement learning and graph neural networks are used to calculate the optimal evacuation path in real time. If the calculated evacuation path is different from the current induction direction, the induction direction is updated to guide the escapee to move to the optimal evacuation path. According to the real-time location of the escapee and the optimal evacuation path, determine whether the escapee deviates from the path. If deviated, warn and guide through voice or visual prompts. Obtain the real-time location of the escapee and the development of the fire, continuously optimize the evacuation path, and ensure the real-time and accuracy of the path. According to the location of the escapee and the development of the fire, determine whether an evacuation alarm needs to be issued. If necessary, an alarm is issued through broadcasting or mobile phone push, and the escapee is guided to evacuate quickly.

[0043] Specifically, indoor positioning technology is the key to obtaining real-time location information of escapees. For example, Bluetooth beacons and Wi-Fi signal strength can be used to determine the location of escapees in a building. By strategically arranging multiple Bluetooth beacons in a building and combining them with the receiving capabilities of smartphones, indoor positioning with an accuracy of 1-3 meters can be achieved. This method can not only obtain the coordinates of the escapee, but also calculate the moving speed and direction through continuous positioning. Fire sensors and monitoring systems are important means to obtain real-time information on the development of fires. Devices such as temperature sensors, smoke detectors, and infrared cameras can form a comprehensive fire monitoring network. For example, a high-rise building is equipped with 20 temperature sensors and 10 smoke detectors on each floor, and these devices report data to the central control system once a second. When the temperature in a certain area suddenly rises or the smoke concentration increases, the system can quickly locate the fire. By analyzing the data changes of sensors in adjacent areas, the speed and direction of the fire spread can also be inferred. Machine learning algorithms play an important role in real-time calculation of the optimal evacuation path. Reinforcement learning algorithms can learn the best decisions in different situations by simulating a large number of evacuation scenarios. For example, the system can pre-train a decision model based on factors such as the building's structure, fire location, and personnel distribution. When an actual fire occurs, this model can quickly give the optimal evacuation path based on real-time data. At the same time, graph neural networks can effectively process the complex structural information of the building, converting the floor plan into a network representation of nodes and edges, so as to more accurately calculate the shortest and safest evacuation path. Real-time updating of the induction direction is a key step to ensure the safety of escapees. The system will continuously compare the current induction direction with the newly calculated optimal path, and will immediately update the guidance once a difference is found. For example, if the original evacuation path is through the central staircase, but the system detects that the stairwell is already filled with smoke, the new optimal path may be changed to use the fire elevator. At this time, the system will immediately update the signs and voice prompts in all relevant areas to guide the escapees to change direction. Warning and induction of escapees who deviate from the path is an important measure to ensure evacuation efficiency. The system can set a "safe corridor", and when the escapee's position deviates from this range, an alarm will be triggered. For example, if an escapee is supposed to move north but the system detects that he is moving east, a specific warning tone will be immediately played through the speakers in the area, and an arrow will flash on a nearby display to indicate the correct direction. If the escapee is carrying a smartphone, they can also be reminded to correct their movement direction through push notifications or vibrations. Continuous optimization of evacuation routes is a necessary means to deal with dynamic fire situations. The system will adjust the evacuation strategy in real time based on the continuously updated fire development data and the distribution of escapees. For example, if it is detected that a major evacuation route has been blocked by fire, the system will immediately recalculate the evacuation route for all affected areas and update the guidance information accordingly.This dynamic optimization ensures that every escapee always has the safest and quickest evacuation route. Issuing evacuation alarms based on the location of the escapee and the development of the fire is the final safety guarantee. The system sets multiple alarm trigger conditions, such as fire spread speed, smoke concentration, temperature, etc. When these indicators reach the danger threshold, the system automatically starts the evacuation procedure. For example, when the smoke concentration on a certain floor exceeds 50 mg per cubic meter, the floor and the floors above will receive a mandatory evacuation alarm. The alarm can be issued simultaneously through the building's broadcast system, emergency lighting system, and push notifications to registered users' mobile phones to ensure comprehensive coverage of the information.

[0044] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A sensor-based fire escape guidance method, characterized in that: The method comprises: Based on the ambient temperature information, smoke concentration information and combustible gas content information collected by the sensor, combined with the three-dimensional space information inside the building, the heat transfer simulation and smoke flow simulation are constructed through the computational fluid dynamics model to obtain the fire spread trend, fire spread speed and fire expansion direction; According to the fire spread trend, fire spread speed and fire expansion direction, the area most seriously affected by the fire is determined, and the location of the fire point, the location of the heat source and the places that may induce risks in the future are dynamically marked in the system; According to the building structure model and the location information of the escapees, combined with the fire spread trend, fire spread speed and fire expansion direction, the safe evacuation path and evacuation facilities are calculated through graph theory algorithms; Dynamically adjust the update frequency of fire induction and evacuation alarms according to the fire development and preset trigger thresholds to ensure that the real-time alarms delivered to escapees are forward-looking and accurate; If the fire spread trend, fire spread speed or fire expansion direction changes, the fire-affected area, fire starting point location, heat source location and subsequent risk locations will be recalculated, and the evacuation routes and evacuation facilities will be updated; Dynamically push fire guidance information and evacuation alarms to escapees at different locations through electronic display devices or sound prompts, informing them of the evacuation routes and facilities that should not be taken or used at present; According to the real-time location of the escapees and the development of the fire, the guidance direction and evacuation path are continuously optimized to ensure the accuracy and real-time nature of user safety guidance and evacuation alarms.

2. The method according to claim 1, characterized in that The method is based on the ambient temperature information, smoke concentration information and combustible gas content information collected by the sensor, combined with the three-dimensional space information inside the building, and constructing a heat transfer simulation and a smoke flow simulation through a computational fluid dynamics model to obtain the fire spread trend, fire spread speed and fire expansion direction, including: Sensors continuously monitor ambient temperature information data within the building; If the temperature of any point in the ambient temperature information data is higher than the preset temperature threshold, the location information of the point and the smoke concentration data information and combustible gas content data information of the surrounding environment of the point are collected; According to the location information of the point where the temperature is higher than the preset temperature threshold and the smoke concentration and combustible gas content data information collected around the location, the parameters of the fire spread calculation model in the computational fluid dynamics model are constructed; The computational fluid dynamics model generates a three-dimensional spatial computational grid of the building based on the geometric structure of the building, and performs differential calculations on the three-dimensional spatial computational grid to obtain a simulated flow field; Construct a three-dimensional space based on the structural data information of the building, use the thermal radiation calculation model and the thermal convection calculation model to combine the material structure distribution information and thermal conductivity calculation data information inside the building to construct a heat transfer calculation model, calculate the temperature field in the simulated flow field, and obtain the heat data of each grid in the three-dimensional space; Obtain the heat data output by the heat transfer model, fuse the smoke concentration and combustible gas data information to perform differential calculation to construct a smoke flow model, and obtain the spatial distribution of smoke particles in each grid. The smoke flow model includes a physical model of air entrainment and diffusion behavior, simulating the state of smoke particles at different times in each grid in three-dimensional space; According to the physical model in the smoke flow model and the state information of smoke particles at different times in each grid, a smoke concentration calculation model is established to predict the development trend of fire, and the burning rate information and fire stage information of the fire in each grid are obtained, so as to make real-time judgment on the state of the fire area in the building; If the fire in any grid is in the smoldering stage and the ambient temperature of the grid rises to the first threshold, the state of the smoke particles is discretized by constructing an adaptive hybrid difference format, and the combustion rate of the surrounding grids is obtained as the combustion reaction rate. The extreme point is searched in the combustion rate using the gradient descent method to obtain the fire spread speed at different extreme points. The burning rate is determined by the gradient descent method, and combined with the fire spread speed information of the extreme point, a three-dimensional geometric fire simulation model is constructed by integrating the three-dimensional spatial geometric dimensions using the differential model and the finite element model. The temperature field and smoke field space are simulated using the computational fluid dynamics model, so as to perform numerical simulation to obtain the expansion direction of the fire in the three-dimensional space. It also includes: based on the ambient temperature, smoke concentration and combustible gas content information, combined with the three-dimensional space information inside the building, the heat transfer and smoke flow simulation is constructed through the computational fluid dynamics model to obtain the fire spread trend, spread speed and expansion direction.

3. The method according to claim 2, characterized in that The method is based on the ambient temperature smoke concentration and combustible gas content information, combined with the three-dimensional space information inside the building, and constructs a heat transfer and smoke flow simulation through a computational fluid dynamics model to obtain the fire spread trend, spread speed and expansion direction, including: Obtain the three-dimensional spatial information inside the building, including room layout, door and window locations, vent locations, etc., and construct a three-dimensional model of the building; Temperature sensors, smoke concentration sensors and combustible gas sensors are arranged inside the building to collect data on ambient temperature, smoke concentration and combustible gas content in real time; Based on the 3D model of the building and the data collected by the sensors, a computational fluid dynamics model is constructed to simulate the heat transfer and smoke flow inside the building; The temperature distribution, smoke concentration distribution and combustible gas concentration distribution at different time points and locations are calculated through computational fluid dynamics models; According to the changes in temperature distribution, smoke concentration distribution and combustible gas concentration distribution, the location and size of the fire can be judged, and the spread trend and expansion direction of the fire can be determined; Use machine learning algorithms, such as support vector machines or neural networks, to predict the speed and extent of a fire spread over a period of time based on historical fire data and current fire parameters; Combine the predicted information such as fire spread trend, spread speed and expansion direction with the three-dimensional model of the building to generate an intuitive fire spread visualization diagram, providing auxiliary decision-making basis for fire rescue.

4. The method according to claim 1, characterized in that: According to the fire spread trend, fire spread speed and fire expansion direction, the area most seriously affected by the fire is determined, and the location of the fire point, the location of the heat source and the subsequent locations that may induce risks are dynamically marked in the system, including: The heat source distribution image of the fire scene is obtained by infrared thermal imaging camera, and the fire starting point and main heat source location are determined according to the location, area and temperature change of the heat source in the image; Computer vision algorithms are used to analyze continuously acquired heat source distribution images, and the direction and speed of fire spread are determined by the shape, size and position changes of the heat source area; According to the speed and direction of fire spread, combined with the internal structure diagram of the building and the distribution of combustibles, the fire spread trend in the future and the most likely affected areas are predicted; Correlate and analyze the fire spread trend with the distribution of people in the building, determine the areas most seriously affected by the fire and requiring priority evacuation, and dynamically mark them with warning colors in the system; Based on the internal structure and material characteristics of the building, the support vector machine algorithm is used to identify risk locations that may cause secondary disasters and mark them in the system; Taking into account the fire spread trend, speed, direction and distribution of risk locations, the ant colony optimization algorithm is used to plan the evacuation route, and the route information is dynamically pushed to the on-site commanders and evacuation guides; Continuously track the development of fires, dynamically update fire spread prediction results, affected area divisions and evacuation route planning based on real-time on-site information, and provide accurate decision-making basis for fire fighting and personnel evacuation.

5. The method according to claim 1, characterized in that The method calculates safe evacuation paths and evacuation facilities based on the building structure model and the location information of the escapees, combined with the fire spread trend, fire spread speed and fire expansion direction, through a graph theory algorithm, including: Obtain building structural model information, including the building's floor plan, number of floors, and room distribution, and construct a topological structure diagram of the building; Obtain personnel location information, determine the room or area where each person is located, and map the personnel location to the building topology diagram; Obtain fire spread information, including the location, spread trend, spread speed and spread direction of the fire, and map the fire spread information into the building topology diagram; According to the building topology diagram, personnel location and fire spread information, a spatiotemporal weighted graph is constructed, in which nodes represent rooms or areas, edges represent the connectivity between adjacent rooms or areas, and the weight of the edges represents the impact of fire spread; Use the shortest path algorithm, such as the Dijkstra algorithm or the A algorithm, to calculate the shortest evacuation path from each person's location to the safe exit on the spatiotemporal weighted graph to obtain a personalized evacuation path; According to the building topology diagram and evacuation route, determine the evacuation facilities that need to be opened, such as emergency lighting, signs and fire doors, and control the status of these facilities; Monitor the spread of fire and the progress of personnel evacuation in real time, and dynamically adjust evacuation routes and facilities according to the latest situation to ensure the safety and effectiveness of evacuation; It also includes: based on the building structure model and the escapee's location information, combined with the fire spread trend, spread speed and spread direction, the safety factor, accessibility and congestion of the evacuation path are calculated through graph theory algorithms to obtain the optimal safe evacuation path and evacuation facilities.

6. The method according to claim 5, characterized in that The method calculates the safety factor, accessibility and congestion of the evacuation path through a graph theory algorithm based on the building structure model and the location information of the escapees, combined with the fire spread trend, spread speed and spread direction, so as to obtain the optimal safe evacuation path and evacuation facilities, including: According to the building structure model, the layout information of the building is obtained, including the location coordinates and connectivity relationships of rooms, corridors, stairs, etc., and the building topology map is constructed; Obtain information about the location of the fire and the location of the personnel, and mark the fire source nodes and personnel nodes in the building topology map; According to the fire spreading trend, spreading speed and spreading direction, the cellular automaton model is used to simulate the dynamic spreading process of fire in the building, and the fire impact range at different times is obtained; In the building topology graph, remove the nodes and edges affected by the fire and update the building topology structure; At the same time, based on the fire spreading trend, the nodes and edges that may be affected in the future are predicted to reduce their safety factor; The Dijkstra shortest path algorithm is used to calculate the evacuation path with the personnel node as the starting point and the evacuation exit as the end point; During the calculation process, the safety factor, length and congestion of the path are comprehensively considered to obtain the optimal evacuation path with the minimum comprehensive cost; If the congestion degree of a path exceeds a preset threshold, the path is judged to be a congested path. According to the density of people and the evacuation flow, a cellular automaton model is used to simulate the evacuation process, predict the evacuation time of the congested path, and use it as a weighted item of the path cost. According to the distribution of evacuation paths in the building topology map, determine the safe exits and evacuation passages that need to be opened, and determine the configuration of evacuation facilities based on the flow of people on the path, such as safety signs, emergency lighting and evacuation guidance devices, so as to achieve efficient and orderly evacuation.

7. The method according to claim 1, characterized in that The updating frequency of fire induction and evacuation alarms is dynamically adjusted according to the fire development and the preset trigger threshold to ensure that the real-time alarms delivered to the escapees are forward-looking and accurate, including: Obtain the current fire situation data at each location sent back by the sensing devices in the building structure, and collect the locations of the evacuees sent back by the devices on the personnel; The constructed deep learning network analyzes the current fire data at each location to obtain the current fire severity at all locations in the building; Different fire induction trigger thresholds and personnel evacuation trigger thresholds are set according to differences in population density and fire severity change trends; Set the calculation formula for the trigger threshold of each position. The formula needs to include variables such as position and space size; The induction threshold is used to induce personnel to evacuate to a certain area, and is usually a smaller value; The evacuation threshold is used to trigger personnel to escape from danger, and usually takes a larger value; If the initial density of people is below the average density of buildings, the induction threshold is increased and the evacuation threshold is decreased; Determine thresholds based on personnel and fire information; Get a set of dynamic thresholds; According to the current position information of the personnel and the dynamic threshold, if it is determined that the current position of the personnel is greater than the induction threshold, the induction information is sent to the personnel; If it is greater than the evacuation threshold, an evacuation alarm is sent; Each alarm carries the target evacuation direction, and each alarm is sent at a preset time interval, which is represented by T; T is affected by the number of people around; The calculation formula of T is as follows: T = k1 / (sum(D_{i})); D represents the distance, and D_{i} represents the distance between the person and other person i; i refers to each person that can be found, and the integers 1, 2, and 3 are used to mark the persons in ascending order; The summation symbol ∑ sums each item D_{i} from i=1 to i=m, where m represents the total number of personnel, and m changes in real time according to the currently collected personnel positions; k is the proportionality coefficient, which adjusts the influence of the crowd density around the person on the sending time interval T, and is set to 100; Calculate the real-time update frequency of alerts sent at each location; The moving average method is used to fit the fire data at all locations and to construct a time series prediction model for fire changes at each location. Build a model for each location; The model outputs the fire prediction for the next two time intervals T. The model also outputs the residual of this time series data; The distribution range of the residuals is determined by the random forest method, and a confidence interval of the fire prediction data is determined; Through this model, the possible fire level at each location in the future is determined, and combined with the judgment logic in step 3, an estimate of the frequency of future alarm updates is obtained; If the estimated frequency is lower than the frequency threshold, the frequency is increased, otherwise the frequency is decreased; According to the locations of all evacuees, the preset maximum number of evacuees at each location, and the trend of fire severity at each location, the optimal evacuation route is obtained. The optimal route is output as a real-time alarm, and each route is numbered. Personnel choose the path according to the instructions of this number. Each person gets different route information, and the route number changes in real time, constantly changing according to the fire situation and personnel location; According to the path numbers connected to each location, the numbers are converted into real-time guidance information and sent down to determine the optimal route for the next time; By combining the fire information with the long short-term memory network, the route changes in the future period are determined; Obtain the historical speed distribution information of evacuees on different evacuation routes under the model, obtain the age and physical fitness indicators of the current evacuees, and obtain the instruction information for accelerating or decelerating the evacuees based on the speed ratio between the evacuees and other people on the same route; If the age is lower than the average age, the evacuation speed indicator value is increased, otherwise the speed indicator value is decreased; If the density of people around a certain location is higher than a certain threshold, the evacuation speed of people in this area will be reduced; Collect the alarm history information sent back by all devices with alarm functions, and use the gradient boosting decision tree method to obtain the rule base based on the actual route information selected by the personnel and the corresponding fire information. Select the optimal evacuation route according to different fire levels and different building structures, and determine the optimal route for sending real-time alarms; The rule base is automatically updated every other day, and the matching degree between the results obtained by this method and the results obtained by the fifth step method is calculated; The solution with the highest matching degree is used for output to form a route alert that is ultimately sent to all personnel.

8. The method according to claim 1, characterized in that If the fire spread trend, fire spread speed or fire expansion direction changes, the affected area, fire starting point location, heat source location and subsequent risk locations will be recalculated, and the evacuation routes and evacuation facilities will be updated, including: Obtain dynamic information such as fire spread trend, spread speed and spread direction in real time through the fire monitoring system; Based on the fire dynamic information obtained, the fire spread model is used to calculate the scope of the affected area; In the calculated affected area, the fire point and heat source location are determined by heat source location algorithm; According to the fire spreading trend and speed, combined with the building layout, predict the subsequent fire risk location; Comprehensive information on affected areas, fire points, heat source locations and risk locations, and use the shortest path algorithm to plan evacuation routes; Dynamically adjust and update the guidance direction of evacuation indication facilities according to the planned evacuation route; If the spread of fire changes, return to step 1 to reacquire the fire dynamic information, iterate the subsequent steps, and realize the real-time update of evacuation routes and facilities.

9. The method according to claim 1, characterized in that: The electronic display device or sound prompts are used to dynamically push fire induction information and evacuation alarms to escapees at different locations, informing them of the evacuation routes and facilities that should not be taken or used at present, including: Obtain the current location information of the escapee and determine the escapee's distribution through the preset sensor network; If the location information of the escapee is obtained successfully, proceed to the next step, otherwise the location information is collected again; Determine the location and spread of fires based on real-time data from the fire monitoring system, and generate fire dynamic information; If the fire information changes, the fire dynamic information database will be updated to ensure the real-time nature of the information; Combine fire dynamic information with preset evacuation route planning to determine the safety of each evacuation route and generate a list of prohibited routes; If a path is determined to be unsafe, it will be added to the list of disabled paths, and a disabled path alert will be pushed to the escapee through a display device or sound device; Simultaneously monitor the status of evacuation facilities, such as elevators and safety doors, to determine the list of available and disabled facilities; If a facility is determined to be unavailable, it will be added to the list of disabled facilities, and a disabled facility alert will be pushed to the survivors through a display device or sound device; Integrate prohibited paths and prohibited facilities information to generate comprehensive guidance information; Dynamically adjust the content of the guidance information according to the current location of the escapee and the dynamic information of the fire, so as to ensure that the information received by the escapee is targeted and real-time; Dynamically push comprehensive guidance information and evacuation alarms to escapees through electronic display devices or sound prompt devices; If the escapee approaches a prohibited path or facility, the alarm intensity will be increased to prompt the escapee to adjust the evacuation direction; Continuously monitor the location changes of escapees and fire dynamics information, and update guidance information and alarm content in real time; If the location of the escapee or the fire situation changes, the safety of the route and facilities will be re-evaluated, and the guidance information and alarms will be updated to ensure that the escapee always receives the latest evacuation instructions.

10. The method according to claim 1, characterized in that The described method continuously optimizes the guidance direction and evacuation path according to the real-time location of the escapees and the development of the fire, ensuring the accuracy and real-time nature of the user's safety guidance and evacuation alarm, including: Obtain the real-time location information of the escapee, including the escapee's coordinates, movement speed and direction, etc., through indoor positioning technology; Obtain real-time information on fire development, including the location, spread speed and direction of the fire, through fire sensors and monitoring systems; Based on the location of the escapees and the development of the fire, machine learning algorithms such as reinforcement learning and graph neural networks are used to calculate the optimal evacuation path in real time; If the calculated evacuation path is different from the current guidance direction, the guidance direction is updated to guide the escapees to move to the optimal evacuation path; Based on the escapee's real-time location and optimal evacuation path, determine whether the escapee has deviated from the path. If so, warn and guide through voice or visual prompts; Obtain the real-time location of escapees and fire development status, continuously optimize the evacuation route, and ensure the real-time and accuracy of the route; Based on the location of the escapees and the development of the fire, determine whether it is necessary to issue an evacuation alarm. If necessary, an alarm will be issued via radio or mobile phone push, and the escapees will be guided to evacuate quickly.

Images