Fire optimal escape route planning method and system based on multi-modal fusion algorithm

Abstract

The invention relates to the technical field of route optimization, in particular to a fire optimal escape route planning method and system based on a multi-modal fusion algorithm, and the method comprises the following steps: obtaining temperature, smoke and oxygen sensor readings and building information data, calculating the multi-source fusion criticality of each position and each moment of a building, obtaining a position time sequence fusion hazard value, and calculating the position time sequence fusion hazard value; and based on the position time sequence fusion hazard value, integrating numerical values of different time points, and establishing a time-varying hazard probability distribution diagram. According to the invention, through real-time acquisition and fusion of temperature, smoke and oxygen sensor readings and building information data, hazard degree dynamic evaluation of each position in a building is realized, a position time sequence fusion hazard value is constructed, and space hazard identification accuracy is effectively improved; thermal damage analysis and long-term fragility factor calculation of beams and columns in the building structure are further combined, and the bearing performance change of the building structure under the thermal action is comprehensively evaluated, so that the risk evaluation is more comprehensive.

Description

Technical Field

[0001] The present invention relates to the technical field of route optimization, and in particular to a method and system for planning an optimal fire escape route based on a multimodal fusion algorithm. Background Art

[0002] The optimal fire escape route planning method is a method that uses multi-source data fusion and spatiotemporal risk analysis technology to optimize the path in a fire disaster scenario. Its purpose is to quickly analyze the dangerous conditions in the building and the safety of the spatial passage when a fire occurs.

[0003] In actual application, existing technologies overly rely on a rough assessment of the overall dangerous condition of the building, usually using single sensor monitoring or static building structure data as the basis for risk judgment, while failing to effectively analyze the thermal damage and bearing capacity changes of the building structure itself, ignoring the impact of structural damage on the safety of escape routes, resulting in the possibility of ignoring key risk locations in actual fire situations. Therefore, improvements are needed. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings existing in the prior art and to propose a method and system for planning an optimal fire escape route based on a multimodal fusion algorithm.

[0005] In order to achieve the above object, the present invention adopts the following technical scheme: a method for planning an optimal fire escape route based on a multimodal fusion algorithm, comprising the following steps: Obtain temperature, smoke, oxygen sensor readings and building information data, calculate the multi-source fusion hazard degree at each location and time of the building, obtain the location time series fusion hazard value, integrate the values ​​at different time points based on the location time series fusion hazard value, and establish a time-varying hazard probability distribution map; Based on the time-varying hazard probability distribution map and the information of the middle beam and column in the building information data, the damage of the structural components caused by the thermal effect is analyzed, and the long-term fragility factor of the components is calculated; based on the long-term fragility factor of the components, the long-term fragility factor of the components is superimposed on the time-varying hazard probability distribution map to generate a time-varying structural hazard probability distribution map; According to the time-varying structural hazard probability distribution diagram, the risk of each channel section within the travel time is extracted, and the channel time-integrated risk integral of each channel section is calculated. Based on the channel time-integrated risk integral, it is converted into the survival probability of passing the channel section to establish the path segment cost value; The path segment cost value is applied to analyze the escape routes from the starting point to the exit in the path network, the survival index of each route is cumulatively calculated, the path is selected according to the survival index, and the survival time escape route is obtained.

[0006] Preferably, the step of obtaining the position time series fusion hazard value is: deploying positioning temperature sensors, smoke sensors and oxygen sensors in the building structure, and synchronously collecting data at the current time under each position number every 10 seconds, generating original temperature data, original smoke data and original oxygen data with timestamp, position number and sensor type; Based on the original temperature data, original smoke data and original oxygen data, the fluctuating values, missing values ​​and mutation values ​​that appear in two consecutive time periods are detected by differential sliding windows and replaced with upper and lower neighbor values ​​to generate continuous and complete temperature data, smoke data and oxygen data sequences; According to the continuous and complete temperature data, smoke data and oxygen data sequences, the position time series fusion hazard value of each sampling node is calculated.

[0007] Preferably, the step of acquiring the time-varying hazard probability distribution map is: dividing the time window of fixed period according to the position time series fusion hazard value, and performing interval classification and numerical distribution statistics on the position time series fusion hazard value of each spatial node in each time window to generate a hazard level distribution sequence; Based on the hazard level distribution sequence, a spatial topological network is established and grid interpolation is performed according to the three-dimensional coordinate information of the building structure space, and the position time series fusion hazard value corresponding to each spatial node is mapped into spatial coordinates to generate a time-varying hazard probability distribution map that matches the building space coordinate system.

[0008] Preferably, the step of obtaining the long-term vulnerability factor of the component is: according to the time-varying hazard probability distribution diagram and the beam and column information in the building information data, extracting the node index corresponding to each component number, and calling the spatial coordinates, component material thermal conductivity, component cross-sectional area, component surface temperature, component length and component surface reflectivity from the corresponding node to generate a component thermal environment input parameter set; Based on the component thermal environment input parameter set, the thermal energy transfer per unit area of ​​the component in a single cycle is calculated using the thermal conductivity of the material, and the thermal deflection effect caused by the temperature difference in the length direction of the component and the reflectivity of the component surface is combined to obtain the component temperature difference stress tensor value sequence; The long-term fragility factor of the component is calculated according to the component temperature difference stress tensor value sequence.

[0009] Preferably, the step of acquiring the time-varying structural hazard probability distribution map is: according to the long-term fragility factor of the component, calling the time-varying hazard probability distribution map, extracting the spatial three-dimensional coordinate information of each spatial node, matching the spatial coordinates with the component number, and combining the long-term fragility factor of the component, mapping the long-term fragility factor of the component to the node spatial coordinates, and generating a long-term fragility factor space mapping table associated with the node and the component; Based on the long-term vulnerability factor spatial mapping table associated with the nodes and components, a spatial topological network is constructed to generate a time-varying structural hazard probability distribution map.

[0010] Preferably, the step of obtaining the time-integrated risk integral of the channel is: according to the time-varying structural hazard probability distribution diagram, extracting the three-dimensional coordinates of the starting point and the end point of each channel segment in the building space coordinate system, and according to the channel segment structure number, obtaining the hazard probability value of the spatial position covered by the channel segment in the time-varying structural hazard probability distribution diagram time step by time, forming a time series set of structural hazard probabilities of each channel segment at each time step; Based on the set of structural danger probability time series, combined with the length of each channel segment, the average walking speed under the current evacuation state and the spatial span from the start point to the end point of the channel segment, the travel time interval required to cross the channel segment is calculated, and the structural danger probability value within the time interval is screened to obtain the danger probability sequence within the travel time interval of the channel segment; According to the danger probability sequence within the passage time interval of the passage section, the passage time integral risk integral is calculated, and the calculation formula is: in, For the The channel time integral risk integral of each channel segment, For the The passage time of each channel section, For the Channel segments at time The structural hazard probability value extracted from the time-varying structural hazard probability distribution diagram at any moment, For the The length of the channel segment, For the The average walking speed of each channel segment, is the structural danger probability value corresponding to the passage section at the beginning of passage, For the The spatial distance between the nodes at both ends of a channel segment.

[0011] Preferably, the step of obtaining the path segment cost value is: extracting the channel time product risk integral of each channel segment, and obtaining the total travel time, starting point and end point spatial index numbers corresponding to the channel segment, and arranging them in combination with the node topological order of the channel segment to generate a channel segment risk time accumulation parameter set; Based on the channel segment risk time accumulation parameter set, the structural attribute information of the channel segment is obtained, including the channel segment wall material, the ground anti-slip level and the number of adjacent channel segments, to form a joint matrix of structural code and channel time product risk; According to the structural coding and the channel time product risk joint matrix, the path segment cost value is calculated, and the calculation formula is: in, For the The path segment cost of a channel segment, For the The channel time integral risk integral of each channel segment, For the The wall material of each channel segment, For the The anti-slip level of the ground in each channel section, For the The number of adjacent channel segments per channel segment, For the The radian value of the angle between the starting point and the end point of a channel segment.

[0012] Preferably, the step of acquiring the survival time escape route is: according to the path segment cost value, acquiring the topological structure data of all path segments in the path network, mapping the network node connection relationship and the corresponding marking of the path segment cost value, and performing a path search from the path node starting point to the exit, generating a set of all feasible escape paths from the starting point to the exit; Based on the set of all feasible escape paths from the starting point to the exit, traverse and accumulate the cost values ​​of each path segment by segment to generate a set of cumulative survival indicators of the path; According to the cumulative survival index set of the path, the path segment cost values ​​of all feasible escape paths are compared one by one, and the escape path with the maximum path segment cost value is selected to establish a survival time escape route.

[0013] The present invention provides an escape route planning system, comprising: Environmental data acquisition module, which collects data from temperature, smoke, oxygen sensors and building information to obtain environmental and building data sets; A multi-source data fusion module integrates data at each location and time based on the environment and building data sets to generate a time-varying hazard probability distribution map; The structural damage analysis module uses the time-varying hazard probability distribution diagram to perform thermal damage analysis on the central beams and columns of the building to obtain component fragility factors; A risk assessment module superimposes the component vulnerability factor with the time-varying hazard probability distribution map, calculates the risk value of each channel section according to the risk of each channel section during the travel time, converts the risk value into the survival probability of the channel, and establishes the path segment cost value; The path planning module analyzes possible escape routes from the starting point to the exit in the path network according to the path segment cost value, selects the escape route with the best survival time, and obtains the optimal escape path.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: The present invention realizes dynamic assessment of the degree of hazard at various locations inside the building by acquiring and integrating temperature, smoke, oxygen sensor readings and building information data in real time, constructs a position time-series fusion hazard value, and effectively improves the accuracy of spatial hazard identification; further combines the thermal damage analysis of beams and columns in the building structure with the calculation of long-term vulnerability factors, comprehensively evaluates the changes in the bearing capacity of the building structure under thermal effects, and makes the risk assessment more comprehensive; and based on the time-varying structural hazard probability distribution data, refines the risk integral of each channel section, accurately derives the survival probability of personnel passing through each channel section, and improves the accuracy and reliability of the calculation of the escape path cost; through the accurate accumulation and optimization comparison of the path segment cost values, the escape route selection is more in line with the safe evacuation needs of personnel in the actual building environment, and the scientificity and effectiveness of personnel escape path planning in fire scenarios are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] See also Figure 1 The present invention provides a technical solution, a method for planning an optimal fire escape route based on a multimodal fusion algorithm, comprising the following steps: Obtain temperature, smoke, oxygen sensor readings and building information data, calculate the multi-source fusion hazard degree at each location and time of the building, obtain the location time series fusion hazard value, integrate the values ​​at different time points based on the location time series fusion hazard value, and establish a time-varying hazard probability distribution map; Based on the time-varying hazard probability distribution map and the information of the middle beam and column in the building information data, the damage to the structural components caused by thermal effects is analyzed and the long-term fragility factors of the components are calculated; based on the long-term fragility factors of the components, the long-term fragility factors of the components are superimposed on the time-varying hazard probability distribution map to generate a time-varying structural hazard probability distribution map; According to the time-varying structural hazard probability distribution diagram, the risk of each channel section within the travel time is extracted, and the channel time-integrated risk integral of each channel section is calculated. Based on the channel time-integrated risk integral, it is converted into the survival probability of passing the channel section to establish the path segment cost value; The path segment cost is applied to analyze the escape routes from the starting point to the exit in the path network, the survival index of each route is accumulated and calculated, the path is selected according to the survival index, and the escape route with survival time is obtained.

[0018] The steps for obtaining the position-time fusion hazard value are as follows: deploy positioning temperature sensors, smoke sensors, and oxygen sensors in the building structure, and synchronously collect data at the current time under each position number every 10 seconds to generate raw temperature data, raw smoke data, and raw oxygen data with timestamps, position numbers, and sensor types; Based on the original temperature data, original smoke data and original oxygen data, the fluctuating values, missing values ​​and mutation values ​​that appear in two consecutive time periods are detected by differential sliding windows and replaced with upper and lower neighbor values ​​to generate continuous and complete temperature data, smoke data and oxygen data sequences; According to the continuous and complete temperature data, smoke data and oxygen data sequence, the position time series fusion hazard value of each sampling node is calculated. The calculation formula is: in, For the The position time series fusion hazard value of each node, For the The average temperature of each node in the last 30 seconds. For the The average smoke particle density of each node in the last 30 seconds, For the The average oxygen concentration of each node in the last 30 seconds, For the The local area of ​​the space where the node is located, For the The node corresponds to the window ventilation area of ​​the space. For the The relative height of a node from the ground.

[0019] Specifically, based on the sensor deployment requirements inside the building, the corresponding coordinate position is determined in each area and the specific position number is marked. The temperature sensor, smoke sensor and oxygen sensor are respectively installed at these marked building space positions. The temperature sensor adopts a thermal resistor type with higher stability, the smoke sensor adopts a photoelectric detection method, and the oxygen sensor adopts an electrochemical method. To ensure the accuracy of the measurement data, a collection cycle of ten seconds is required so that each position number can output a set of accurate sensor detection values ​​at the current moment. At the same time, the timestamp is bound to the position number and the sensor type. Since the layout inside the building often has complex scenes such as floors, corridors and rooms, it is necessary to record the floor identification and plane coordinates corresponding to each position number under a unified spatial index system, determine the coordinate center point through the pre-drawn building structure plan, and establish a reference value corresponding to the actual physical size for each installation position. Subsequently, a periodic synchronous collection command is enabled for each sensor, where the temperature data is stored in degrees Celsius, the smoke particle concentration data is stored in milligrams per cubic meter, and the oxygen concentration is stored in percentage. The timestamp is recorded in the precise format of year-month-day-hour-minute-second, and the sensor type is represented by its own digital number. For example, the temperature sensor is marked as T001, the smoke sensor is marked as S001, and the oxygen sensor is marked as O001. In this way, the data source and data collection time of the sensor can be directly read in one record. After the collection is completed every ten seconds, the system will classify all newly acquired sensor readings into the corresponding spatial location directory, and associate them with the location number and timestamp in the real-time database in an indexed manner. When the amount of data accumulates, it can be archived in batches according to days or hours. When archiving, the index information is retained together with the sensor data so that the specific information of a certain time period, a certain location or a certain type of sensor can be quickly retrieved through the index number later. At the same time, after each collection is completed, the system will perform a brief statistics on the recorded temperature, smoke, and oxygen values ​​according to the defined monitoring requirements, and record the maximum, minimum and average values. When a device is found to be offline, the data missing mark of the corresponding location number will be automatically marked. Finally, the original temperature data, original smoke data, and original oxygen data with timestamps, location numbers, and sensor types can be formed on this basis.

[0020] Based on the original temperature data, original smoke data and original oxygen data obtained previously, combined with the numerical items recorded in adjacent time periods, differential sliding window detection and upper and lower adjacent value replacement repair are performed on the fluctuation values, missing values ​​and mutation values. In the specific implementation process, the three types of temperature, smoke and oxygen data are first sorted and classified in chronological order, and an independent data sequence is established under each position number. Then, the data is segmented according to the sliding window length m and step length n set by the system. In the sliding window, the numerical difference between two adjacent records is first calculated and accumulated. If it is detected that k consecutive differences reach the set threshold, the interval can be regarded as a fluctuation interval, and a secondary comparison is performed on all data in the fluctuation interval. Data that obviously exceeds the normal range of the position is marked as a mutation value. For the judgment of missing values, the corresponding time is compared when reading each record. If the gap lasts for more than 20 seconds, it is directly regarded as a missing value. After locating the mutation value or missing value, it is necessary to introduce the upper and lower neighbor value replacement strategy, that is, call the valid values ​​of the two adjacent records for linear interpolation to generate a smooth value after replacement. In order to ensure the credibility of the repaired data, a differential comparison will be performed again after replacement. If the difference between the replaced value and the previous and next data still exceeds the pre-established threshold range, it will be further marked as high-risk data, and the original value will be retained for subsequent investigation. After completing the data sliding window detection and upper and lower neighbor value replacement repair of all position numbers, the repaired temperature, smoke, and oxygen data will be reorganized and stored in the form of timestamps and position numbers corresponding to the original data, and summarized according to the position number and time order to obtain the final continuous and complete temperature data, smoke data, and oxygen data sequences.

[0021] formula: The benefit of the formula is that by integrating environmental factors such as the average temperature, the average smoke particle density, and the oxygen concentration, different physical quantities are calculated in a unified expression, so that when judging the local fire hazard, the influence of temperature, smoke, and oxygen content can be taken into account at the same time. At the same time, differentiated parameters of local area and window ventilation area are introduced, and combined with relative height, the ventilation and smoke accumulation conditions at different spatial levels in the building can be effectively distinguished, thereby obtaining a hazard quantification result that is closer to the real fire scene, and realizing a more accurate quantification and early warning of the spatial hazard level in the overall system or method.

[0022] The steps to obtain are as follows: This parameter indicates the location The average temperature in the last 30 seconds needs to be in the repaired temperature data at position The continuous 30-second records are accumulated and then divided by the number of record entries to obtain the average temperature level during the period. The specific value can be obtained by placing a temperature sensor at the location in the building fire hazard scenario. The temperature sensors at the node are continuously collected. For example, at the node with the position number A12, the temperatures measured in 30 seconds (a total of 3 data, each with an interval of 10 seconds) are 78.1, 79.5, and 82.0 degrees Celsius, respectively. After adding them up one by one, the total is 239.6 degrees Celsius. The number of record entries is 3, so the final AMT The value of .

[0023] The steps to obtain are as follows: This parameter indicates the location The average value of smoke particle density in the last 30 seconds. The specific value is also extracted based on the smoke data that has been repaired in the previous step. The acquisition method can directly imitate the calculation process of the temperature average value. In the continuous 30 seconds, multiple readings of smoke concentration are counted and averaged. For example, at the node with position number A12, the smoke particle density is measured as 0.025, 0.030, and 0.027 mg / m3 respectively within 30 seconds. The sum is 0.082 mg / m3, which is divided by 3 to get 0.0273 mg / m3.

[0024] The steps to obtain are as follows: This parameter indicates the location The average oxygen concentration in the last 30 seconds is extracted from the oxygen sensor data. The rolling window average method consistent with temperature and smoke is used, that is, all oxygen concentration readings in 30 seconds are added up and divided by the number of reading items. For example, at the node with position number A12, the oxygen concentration in 30 seconds is 19.2%, 19.0%, and 18.7%, respectively, which is 56.9% after accumulation. The total number of reading items is 3, and the oxygen concentration is calculated as O for .

[0025] The steps to obtain are as follows: This parameter indicates the location The local area of ​​the space where it is located needs to be confirmed in advance based on the structural design drawings or actual measurement data inside the building and If the building plane structure is divided into several rooms or partitions, the position The area of ​​the sub-region is determined as For example, the room corresponding to position A12 is 15 square meters, so A The value is 15. To obtain this value, it is necessary to mark the plane range of each sub-area according to the known positions of walls and partitions when mapping the overall building information and measure the side lengths. Then, the area calculation formula is used. , or irregular areas, the plane polygon segmentation method can be used to accurately obtain For example, if the length of room A12 is 5 meters and the width is 3 meters, then its area is square meters.

[0026] The steps to obtain are as follows: This parameter indicates the location The window ventilation area of ​​the corresponding space needs to be determined through on-site measurement or the identification of window positions and sizes in architectural drawings. If there are multiple windows in the space, the area of ​​all ventilated windows is added together to obtain the area of ​​the corresponding space. For example, in the room at position A12, the window size is 1.2 meters wide and 1.0 meters high. The ventilation area of ​​a single window can be regarded as 1.2 square meters. If only this window can be opened in the room, then That is 1.2.

[0027] The steps to obtain are as follows: This parameter indicates the location The relative height to the ground can be obtained by actual measurement when installing the sensor, recording the vertical distance between the center of the sensor and the ground. For example, the sensor at position A12 is installed at a height of 2.8 meters, then H =2.8.

[0028] Calculation process: The following example selects the monitoring result of position A12 for calculation, and its parameter values ​​are as follows: Calculate the numerator first : Calculate the denominator again : So the first part is: Then calculate : Then calculate : At the same time, the denominator for: So the second part is: Add the two parts: Therefore, the A12 position The result shows that under the combined effect of the above parameters such as temperature, smoke, oxygen concentration and ventilation, location A12 has a certain degree of fire hazard risk. If this result is compared with the other locations By making a horizontal comparison, we can determine the relative danger level of A12 in the entire space. If the value is greater than 25, it can be considered to be highly dangerous, and if it is less than 10, the danger level is relatively low.

[0029] The steps for obtaining the time-varying hazard probability distribution map are as follows: according to the position time series fusion hazard value, the fixed period time window is divided, and the position time series fusion hazard value of each spatial node in each time window is interval graded and numerical distribution statistics are performed to generate a hazard level distribution sequence; Based on the hazard level distribution sequence, the spatial topology network is established and grid interpolation is performed according to the three-dimensional coordinate information of the building structure space. The position time series corresponding to each spatial node is fused with the hazard value for spatial coordinate mapping to generate a time-varying hazard probability distribution map that matches the building space coordinate system.

[0030] Specifically, according to the previously obtained position time series fusion hazard value, a fixed period is first selected as the time window for the monitoring records at multiple times inside the building, and the continuous fusion hazard values ​​are grouped according to the time period corresponding to each window. All node data in each time window are read one by one and a hazard value list is established. Then, all hazard values ​​in this list are sorted in ascending or descending order, and the hazard value range is segmented and determined in combination with the pre-established interval classification threshold. If some hazard values ​​are found to be lower than the minimum threshold, they are classified as a lower hazard level. If any hazard value exceeds the highest threshold, it is marked as the highest hazard level. The setting of these thresholds refers to the statistical results of the monitoring data of the same type of buildings in the early stage of the trial operation, and then obtained after analyzing. For example, the distribution of hazard values ​​of all locations within one month is statistically analyzed. The median H1 and tertile H2 of the distribution are calculated, and H1 and H2 are used as classification thresholds respectively. In this way, all hazard values ​​can be divided into lower hazard intervals, medium hazard intervals and higher hazard intervals in the implementation process. When judging each record, its corresponding fused hazard value is first obtained, and then compared with H1 and H2 one by one. When the hazard value in the record falls in different intervals, it is assigned to the corresponding level and the correspondence between the node and the level is registered. After the registration is completed, the number and frequency of nodes in each interval are counted, and a hazard level distribution table is constructed based on this. Finally, according to the hazard level sequence of the nodes listed in the table and the order and quantity ratio of their occurrence, the distribution statistics are recorded in time windows, and the interval classification and numerical distribution statistics of the hazard values ​​in each window are completed to generate a hazard level distribution sequence.

[0031] Based on the hazard level distribution sequence obtained in the previous article, the coordinate positions of all nodes are listed with reference to the three-dimensional coordinate information of the building structure space, and a spatial topological network is constructed according to the adjacent relationship of the nodes in the three-dimensional space. The nodes and their connecting edges are marked one by one in the topological network, and the distance or spatial index difference between adjacent nodes is recorded. When selecting a specific grid interpolation method, the building area can be divided into a uniform three-dimensional grid first, and then the position time series of each node in the distribution sequence is fused with the hazard value, and mapped to the corresponding grid unit according to its coordinates. When there are multiple nodes in some grid units, the hazard values ​​of all nodes are comprehensively calculated in the unit and the interpolation is calculated in a weighted average manner. The weight coefficient is determined by the distance between nodes or the importance of nodes. This coefficient can be obtained by recording all nodes. The weight coefficient is calculated based on the coordinate difference within the grid. If the node is closer to the grid center, the weight coefficient value is relatively large. For nodes whose distance exceeds the specified value, the weight coefficient can be appropriately reduced and weighted average can be performed. The setting of the specified value is based on the statistical results of the farthest monitoring radius inside the building. For example, in actual measurement, the spatial distance distribution between different nodes is observed. A distance threshold D is calculated by the average distance and standard deviation, and nodes with a distance greater than D are assigned a lower weight. After all grid cells have completed the interpolation of the node time series fusion hazard value, a continuously distributed hazard probability data field can be formed in the entire three-dimensional coordinate system. Finally, the coordinates and hazard values ​​corresponding to the data field are stored as a set of time-varying records to generate a time-varying hazard probability distribution map matching the building space coordinate system.

[0032] The steps for obtaining the long-term vulnerability factor of a component are as follows: according to the time-varying hazard probability distribution diagram and the information of the middle beam and column in the building information data, the node index corresponding to each component number is extracted, and the spatial coordinates, component material thermal conductivity, component cross-sectional area, component surface temperature, component length and component surface reflectivity are called from the corresponding node to generate the component thermal environment input parameter set; Based on the component thermal environment input parameter set, the thermal energy transfer per unit area of ​​the component in a single cycle is calculated using the material thermal conductivity, and the thermal deflection effect caused by the temperature difference in the component length direction and the component surface reflectivity is combined to obtain the component temperature difference stress tensor value sequence; According to the component temperature difference stress tensor value sequence, the long-term vulnerability factor of the component is calculated using the following formula: in, For the The long-term vulnerability factor of each component, For the The heat transfer per unit area of ​​a component, For the The thermal conductivity of the material of each component is For the The total length of the components, For the The surface temperature of each component, For the The cross-sectional area of ​​a component, For the The surface reflectivity of a component.

[0033] Specifically, according to the previously obtained time-varying hazard probability distribution map and the list of central beams and columns listed in the building information data, when comparing each component number, first read the node index corresponding to the component and find its spatial coordinates in the three-dimensional coordinate system in the component information, then combine the building plan and facade surveying records, distinguish the component types of central beams and columns in the coordinate set, then retrieve the component material thermal conductivity listed in the building material database and match the required numerical entries, check the cross-sectional area value of the component with the cross-sectional shape and size marked on the construction drawing one by one, and confirm its final cross-sectional area in the form of "length × width" or "π × radius²", etc. For the component surface temperature, match the component node from the previously recorded temperature distribution data, read the detected temperature value, and The value is compared again with the temperature data below the lower limit of the effective range. After confirming that the temperature data is in the range of 0℃ to 1000℃, it is recorded in the component surface temperature item. At the same time, the component length identification in the building structure drawing is retrieved, and the component length is finally confirmed in combination with on-site measurement. For the value of surface reflectivity, reference can be made to the reflectivity measurement results in the existing material inspection report. For example, the surface reflectivity of a metal component is in the range of 0.2 to 0.6. The obtained value is registered under the component number. In this process, it is also necessary to record the correspondence between each element and the component number one by one, and store the spatial coordinates, component material thermal conductivity, component cross-sectional area, component surface temperature, component length and component surface reflectivity in the same data table according to the node index order. This data table is the component thermal environment input parameter set.

[0034] Based on the component thermal environment input parameter set obtained earlier, when calculating the heat energy transfer per unit area of ​​each component in a single cycle according to the material thermal conductivity list, a fixed time length is first selected as the basis for a single cycle, such as 30 seconds or 60 seconds as the basic cycle time, the component material thermal conductivity is read from the parameter set and its heat transfer mode is recorded, and the vertical and horizontal transfer paths are comprehensively considered according to the heat diffusion process from the surface to the inside of the component in the unit area dimension, and the temperature difference in the length direction of the component is obtained by reading the temperature measurement point, and then the surface reflectivity value is compared. When the surface reflectivity exceeds the preset threshold value of 0. 5, the thermal conductance adjustment item of the component is corrected downward and deducted numerically. When the surface reflectivity is less than 0.2, the thermal conductance adjustment item is adjusted upward. These thresholds are obtained from the previous reflectivity detection data statistics of the same material. By adding or subtracting the heat energy transfer amount and the thermal conductance adjustment item, the influence of the temperature difference distribution of each component in the internal stress evolution process can be obtained. Finally, a stress data sequence related to the component temperature difference is formed within a single cycle. When there are multiple temperature measurement points in the length direction of the same component, a list of local stress values ​​is established one by one and merged into a complete component temperature difference stress tensor value sequence.

[0035] formula: The benefit of the formula is that it integrates multiple physical quantities such as heat transfer per unit area, material thermal conductivity, component surface temperature, cross-sectional area and surface reflectivity into the calculation expression, so that the multi-factor comprehensive effects of component thermal stress characteristics and fire environment can be quantified. In addition, the introduction of absolute value differences and inverse trigonometric functions in the denominator can take into account the reflection characteristics of high-temperature radiation and temperature field gradient differences on the surfaces of different types of components, thereby helping to assess the risk of structural damage to components after long-term exposure.

[0036] The steps to obtain are: This parameter represents the The heat transfer per unit area of ​​each component requires calculating the total heat transfer experienced by the component during the monitoring period, and then converting it into unit area. The specific method includes first collecting the surface and internal temperature sensor records of the component, combining the specific heat capacity of the building material and other data to quantify the heat flow, and dividing the total heat value by the surface area of ​​the component to obtain the unit area transfer value. For example, in the recording period, The total heat transfer value monitored by each component is 10,000 joules, and the surface area of ​​the component is 5 square meters. According to the formula Joules per square meter.

[0037] The steps to obtain are: This parameter represents the The thermal conductivity of each component can be obtained by testing the materials used or consulting the corresponding material manual. For example, the thermal conductivity of steel structures can be between 20 and 50, and that of concrete can be between 1.4 and 3.3. For steel-concrete composite components, the thermal conductivity of multiple layers of materials can be combined to calculate the equivalent thermal conductivity. If the component is made of steel and the thermal conductivity is confirmed to be 45W / (m·K) through testing.

[0038] The steps to obtain are: This parameter represents the The total length of each component can be obtained directly from the architectural design drawings or on-site measurements. For example, for a 6-meter-long steel beam, .

[0039] The steps to obtain are: This parameter represents the The surface temperature of a component comes from the temperature sensors deployed on the surface of the component or in the vicinity. It is necessary to select the average or peak value of the component surface temperature during continuous monitoring and keep a consistent recording method. For example, in a monitoring cycle, the surface temperature of a steel column is measured multiple times as 450°C, 458°C and 461°C respectively. By averaging these readings, we can get .

[0040] The steps to obtain are: This parameter represents the To determine the cross-sectional area of ​​a component, you need to first determine the cross-sectional shape of the component and measure the cross-sectional geometric dimensions. If it is a rectangular cross-section, multiply the long side by the short side. If it is a circular cross-section, use For example, if the width of a column section is 0.4 meters and the height is 0.4 meters, then the cross-sectional area is square meters.

[0041] The steps to obtain are: This parameter represents the The surface reflectivity of a component is determined by testing the optical properties of the component surface or the surface treatment process. Generally, an integrating sphere measuring instrument can be used to obtain the reflectivity of a metal surface. For example, a reflectivity of 0.4 is measured on a stainless steel component, and 0.25 may be recorded on a steel column coated with a fire retardant coating. After acquisition, it is registered under the component number. At the same time, multi-point reflectivity measurement can be considered on site and the average value can be calculated. For example, if 0.35, 0.36, and 0.37 are recorded in three measurements, they can be summarized as 0.36.

[0042] Calculation process: The following is an example calculation of a component, and its parameter values ​​are as follows: Calculate first : So the term is: Recalculate : After taking the square: Then calculate : After taking the square: Add the two terms and take the square root: For the denominator, calculate first : Recalculate : So the denominator is: So the overall formula is: The results show that under the combined influence of the current unit area heat transfer, thermal conductivity, component surface temperature, cross-sectional area and surface reflectivity, the long-term fragility factor of the component reaches about 1104.19. The larger the value, the more likely the component is to deteriorate in structural performance in a long-term fire environment. If compared with the long-term fragility factors calculated for other components, the relative risk ranking of the component in the entire building can be determined. When it is greater than 800, it can be considered to be significantly more fragile, and when it is less than 200, it is less fragile. Therefore, the result can be incorporated into the subsequent structural safety determination process and further reinforcement or evaluation can be carried out.

[0043] The steps for obtaining the time-varying structural hazard probability distribution map are as follows: according to the long-term vulnerability factor of the component, the time-varying hazard probability distribution map is called, the spatial three-dimensional coordinate information of each spatial node is extracted, the spatial coordinates are matched with the component number, and the long-term vulnerability factor of the component is combined to map the node spatial coordinates of the long-term vulnerability factor of the component, and a long-term vulnerability factor space mapping table associated with the node and the component is generated; Based on the spatial mapping table of long-term vulnerability factors associated with nodes and components, a spatial topological network is constructed to generate a time-varying structural hazard probability distribution map.

[0044] Specifically, according to the long-term vulnerability factor of the component obtained above, first read the three-dimensional coordinate records of all spatial nodes in the established time-varying hazard probability distribution diagram, and completely match the coordinate information of each node with the corresponding component number. When matching, check the component layout list inside the building to obtain the coordinate range occupied by each component, and then check whether the node coordinates fall within the boundary of the component one by one. If the coordinates are consistent with the geometric range of the component shape, the node is considered to belong to the area where the component is located. Then, when calling the long-term vulnerability factor of the component, first compare it with the corresponding relationship between each component number, find the value of the long-term vulnerability factor of the component and record it. If the long-term vulnerability factor of the component is lower than some preset thresholds, such as lower than 200, it is classified and marked as low vulnerability. If the long-term vulnerability factor of the component is greater than 800, it is marked as high vulnerability. These thresholds are obtained by monitoring multiple component samples in the early stage. The quantiles of the vulnerability value distribution are calculated by statistical methods, and the quantiles are used as the basis for grade division. After completing the vulnerability factor marking, compare the node The point coordinates and component information are summarized, and the coordinates need to be retained in the order of node index, and the fragile factor values ​​corresponding to the component numbers are recorded, and the data are merged into the same mapping list. If the same component covers multiple node areas, its coordinates and fragile factors need to be registered one by one. If the node coordinates overlap with the boundary area of ​​two components, the measured component edge plane intersection data is referred to for more accurate coordinate difference calculation. For the case where the landing point at the intersection is closer to the center of a component, the node is assigned to the component closer to it and registered under the corresponding component long-term fragile factor entry. In the process of recording the correspondence between nodes and components, the fields of "node index", "three-dimensional coordinates", "component number" and "component long-term fragile factor" are retained for each node. After all nodes are registered, a final set of mapping data can be obtained, which can be called again in the subsequent networked calculation or spatial visualization stage. After completing the above matching and registration operations, a long-term fragile factor spatial mapping table associated with nodes and components is generated.

[0045] Based on the long-term vulnerability factor spatial mapping table associated with nodes and components, the adjacency relationship of all nodes in the building space is first read and the distance between them is determined. The nodes within the specified range are connected to form a topological edge. The specified range can be inferred from the building plan and facade mapping data. For example, 5 meters is taken as the boundary of adjacent judgment in a large-span space, and 3 meters is taken as the boundary of adjacent judgment in a narrow corridor. If the horizontal and vertical distances between nodes do not exceed the range, a connection is established. At the same time, for the vulnerability factor values ​​that have been mapped to the corresponding components, they are connected to the node index so that they can be accessed in the topological network. After the connection construction of the adjacency relationship is completed, the danger probability of each node at the current moment is retrieved from the previous time-varying danger probability distribution map record. To, and with its corresponding component long-term fragility factor for parallel comparison, during the comparison, if the long-term fragility factor exceeds the high fragility threshold and the corresponding danger probability of the node is also higher than 0.7, then in the topological network for this node connection relationship set a higher warning attribute classification, if the fragility factor is low or the danger probability is lower than 0.3, then the node and its connection relationship is marked as a lower warning level, through the network through a traversal of all nodes and edges and classification according to the above thresholds to complete the construction process of the spatial topological network, the danger probability data and the fragility factor data can be combined to obtain an updated distribution map form, for each grid point or node position can be retrieved based on the time of the danger probability value of the time-varying distribution map to obtain the time-varying structure danger probability distribution map.

[0046] The steps for obtaining the time-integrated risk integral of the channel are as follows: according to the time-varying structural hazard probability distribution map, the three-dimensional coordinates of the starting point and the end point of each channel segment in the building space coordinate system are extracted, and according to the channel segment structure number, the hazard probability value of the spatial position covered by the channel segment is obtained in the time-varying structural hazard probability distribution map at each time step, forming a time series set of the structural hazard probability of each channel segment at each time step; Based on the set of structural hazard probability time series, combined with the length of each channel section, the average walking speed under the current evacuation state and the spatial span from the start to the end of the channel section, the travel time interval required to cross the channel section is calculated, and the structural hazard probability value within the time interval is screened to obtain the hazard probability sequence within the channel section travel time interval; According to the danger probability sequence within the passage time interval of the channel section, the channel time-integrated risk integral is calculated, and the calculation formula is: in, For the The channel time integral risk integral of each channel segment, For the The passage time of each channel section, For the Channel segments at time The structural hazard probability value extracted from the time-varying structural hazard probability distribution diagram at any moment, For the The length of the channel segment, For the The average walking speed of each channel segment, is the structural danger probability value corresponding to the passage section at the beginning of passage, For the The spatial distance between the nodes at both ends of a channel segment.

[0047] Specifically, according to the previously obtained time-varying structural hazard probability distribution map, when extracting the three-dimensional coordinates of the starting point and the end point of each channel segment, the building space coordinate system is first searched, the structure number of each channel segment registered in the building information data is read, and the two end nodes corresponding to the number are compared one by one. Then, the specific X, Y, and Z values ​​of these nodes in the three-dimensional coordinate system are recorded. If there is an error in the coordinate information of the node, the data with a deviation within the range of 0 meters to 0.5 meters is first fine-tuned, and the data exceeding this range is removed and marked and reviewed during subsequent positioning. Then, the time steps are traversed in sequence according to the channel segment number, and all spatial positions covered by the channel segment on the time-varying structural hazard probability distribution map are queried at each time step, and the hazard probability values ​​of all relevant positions are compared. Collect and form a danger probability set for the channel segment, and exclude the probability values ​​of positions outside the channel segment or that do not meet the channel coverage boundary requirements. For the boundary requirements, a reference value of 2 meters can be set by measuring the radial distance between the channel cross section and the node coordinates. When the distance from the node to the center line of the channel segment is greater than 2 meters, the node is excluded. Then the remaining probability values ​​are arranged in sequence according to the time sequence number and stored in the same set. These sequences are divided according to each time step, and the adjacent time steps are separated by 10 seconds. When the danger probability of multiple consecutive time steps is higher than 0.6, it can be regarded as a large danger tendency. The time step number is also noted when recording. After comparing the above sequence with the channel segment number, the time series set of structural danger probability of each channel segment at each time step is obtained.

[0048] Based on the structural hazard probability time series set obtained above, first extract the length information of each channel segment from the building data, read the three-dimensional coordinate difference between the starting point and the end point of the channel segment, and calculate the spatial span of the segment. For example, in the case where the difference in the X direction is 2 meters, the difference in the Y direction is 1 meter, and the difference in the Z direction is 0 meters, the three-dimensional distance can be calculated by the Pythagorean theorem to be about 2.24 meters. When obtaining the average walking speed under the current evacuation state, it is necessary to select the registered speed value from the previous crowd monitoring results. If the average walking speed of the passers-by is between 0.8 meters per second and 1.2 meters per second, an interval value can be selected according to the width of the internal channel of the building and the density of the flow of people. When the walking speed is confirmed, Finally, divide the channel length by the speed to get a travel time. At the same time, compare the calculated time with the three-dimensional span of the channel segment. If the difference between the two is not large, they are considered to be consistent. Next, combine the travel time and the time step sequence for screening, and retrieve the generated structural hazard probability time series in the corresponding time interval. For example, all hazard probability data matching the duration from 0 seconds to the end of the travel are screened. As long as the hazard probability falls between 0 and 1, it will be registered. Values ​​higher than 0.8 will be marked. After obtaining the complete hazard probability sequence in the travel time interval of the channel segment, these probability values ​​are arranged in order and correspond one by one to the time step sequence, so as to obtain the hazard probability sequence in the travel time interval of the channel segment.

[0049] formula: The benefit of the formula is that it combines the changes in the channel length, walking speed, and structural hazard probability values ​​at different times, accumulates the effects of these factors in the same expression by integration, and uses the hazard probability at the starting point of the channel as a reference. The addition of the absolute value difference term can more quickly reflect the risk increment when a fire suddenly changes.

[0050] The steps to obtain are: This parameter represents the The travel time of each channel segment needs to be calculated based on the length of the channel segment and the current walking speed. If the length of the channel segment is known, it is recorded as , the current average walking speed is , then you can For example, in a 20-meter-long passage section, when the crowd moves at a speed of 0.8 meters per second, the calculated travel time is Second.

[0051] The steps to obtain are: This parameter represents the Channel segments at time At each moment, the structural hazard probability value is extracted from the previously obtained time-varying structural hazard probability distribution map, and the value range is between 0 and 1. During implementation, it is necessary to retrieve the spatial position covered by the channel segment at each time step, average the probability values ​​of the relevant positions or assign different weight coefficients according to the distance from the channel center line, and then record them in a sequence indexed by time. For example, if the hazard probability values ​​corresponding to the channel segment are 0.3, 0.5, and 0.52 at t=0 seconds, t=1 second, and t=2 seconds, respectively, then Considered as part of the time node sequence.

[0052] The steps to obtain are: This parameter represents the The length of each passage section needs to be obtained by referring to the internal design drawings of the building or through actual on-site measurement. In some curved or irregular-shaped passages, the length can be accumulated after segmented measurement. For example, for a corridor + turning combined passage, the straight length of the front section is 6 meters, the turning section is 1.5 meters, and the straight length of the rear section is 4 meters, which can be combined to obtain .

[0053] The steps to obtain are: This parameter represents the The average walking speed of each channel segment needs to be obtained from actual crowd monitoring or simulation experiments. For example, when counting multiple batches of people, the ratio of the actual walking time of each batch of people in the channel segment to the channel length is calculated, and then the average of all batches is calculated. If several speed values ​​are obtained, the middle interval can be further extracted as the final For example, in the measurement of 5 batches of people, the speeds are 0.7, 0.8, 0.9, 0.8, and 1.0 meters per second respectively. The mean value of 0.84 meters per second can be calculated first.

[0054] The acquisition steps are as follows: This parameter represents the structural hazard probability value corresponding to the channel section at the beginning of the passage, which is taken from the time-varying structural hazard probability distribution diagram mentioned in the same paragraph, but is extracted at the numerical position at time t=0. If there are multiple node probabilities in the spatial area covered by the channel section, a representative probability value is obtained by calculating the weighted or averaged method described above. For example, before passing, the distribution layer corresponding to time t=0 seconds is selected, and all probability values ​​0.2, 0.28, and 0.25 within the channel section are retrieved. The sum of these values ​​and the average probability divided by 3 is 0.243, which can be recorded. .

[0055] The steps to obtain are: This parameter represents the The spatial distance between the nodes at both ends of the channel segment is The measurement meaning is different. It focuses on the coordinate difference of the nodes at both ends of the channel segment and calculates the straight-line distance. For example, if the starting point coordinates are (2, 4, 0) and the end point coordinates are (6, 6, 0), then the three-dimensional space distance formula is Conclusion Right now rice.

[0056] Calculation process: The following is an example scenario where , Between 0 and 25 seconds, the average fluctuates around 0.4, and can reach a maximum of 0.65 at certain moments, taking the channel segment length as , average walking speed , the probability of danger at the initial moment , the distance between nodes Now we analyze the integral step by step. First, calculate the first term in the integral. ,make At time t, it is 0.5, then , the other part , the first term at this moment is about , let’s look at the second item , at t = 5 seconds =0.5, =0.3, then the absolute value difference is 0.2, , 1+4.24=5.24, so the second term is approximately Finally, the sum of these two items is integrated, that is, the continuous or discrete sum of t from 0 to 25 seconds is taken. For example, a discrete step length of 1 second is used for approximate summation. If the sum is calculated every second, , and Add and square root, then divide by , and put Add them together and finally get the cumulative integral value of the entire channel. In this example, if the probability of danger at multiple moments fluctuates between 0.3 and 0.65, the calculated Approximately between 220 and 250.

[0057] The results show that under the conditions of a channel section with a length of 20 meters, a speed of 0.8 meters per second, an initial danger probability of 0.3 and a node distance of 18 meters, by combining the structural danger probability at each moment with the channel space factors, the calculated channel time-integrated risk integral fluctuates between 220 and 250. The larger the value, the higher the degree of accumulation of risk during the crossing process. When it exceeds 250, it can be regarded as a more obvious danger, and when it is less than 150, it can be regarded as a relatively weak risk.

[0058] The steps for obtaining the path segment cost value are as follows: extract the channel time product risk integral of each channel segment, obtain the total travel time, starting point and end point spatial index numbers corresponding to the channel segment, and sort them out in combination with the node topological order of the channel segment to generate a channel segment risk time accumulation parameter set; Based on the channel segment risk time accumulation parameter set, the structural attribute information of the channel segment is obtained, including the channel segment wall material, ground anti-slip level and the number of adjacent channel segments, to form a joint matrix of structural code and channel time accumulation risk; According to the structural coding and channel time product risk joint matrix, the path segment cost is calculated. The calculation formula is: in, For the The path segment cost of a channel segment, For the The channel time integral risk integral of each channel segment, For the The wall material of each channel segment, For the The anti-slip level of the ground in each channel section, For the The number of adjacent channel segments per channel segment, For the The radian value of the angle between the starting point and the end point of a channel segment.

[0059] Specifically, according to the channel time-product risk score and the corresponding total travel time, starting point and end point spatial index numbers of each channel segment obtained previously, it is necessary to read these data one by one and check the field integrity. When it is found that the time field of a channel segment is not in the range of 0 seconds to 7200 seconds, the record can be marked and the data collector can be notified for review. For the spatial index numbers of the starting point and the end point, the overall building layout diagram must be compared first to confirm that the number does belong to the same floor or the same area. When the starting point and the end point belong to different floors, a height difference identification must be made in the floor index matching table to ensure that the vertical channel data is distinguished in subsequent analysis. Then, the node topological order of the channel segment is summarized, and the channel segment number and the node index are matched one by one. At the same time, the total time required for the channel segment to pass and the accumulated risk score are recorded. If the risk score exceeds the empirical threshold of 300, a conspicuous note is made for the channel segment. The way to obtain this empirical threshold is Referring to the channel risk score distribution recorded during multiple fire drills of the same type of buildings in the past 12 months, after calculating the mean plus 2 times the standard deviation of these distribution data, a higher interval value was obtained and set as the evaluation threshold. When sorting according to the node topological order, it is necessary to ensure that the mutual connection relationship between the starting point and the end point can be normally indexed in the topological map, and the node sequence of each channel segment is deduplicated and the content that conforms to the spatial geometric order is retained. If there is a breakpoint in the channel segment connection relationship, a breakpoint mark is added to the data table, and the channel segment number and its start and end node index are recorded and arranged in chronological order. A channel segment risk time accumulation parameter set is formed, which includes fields such as channel time-product risk score, total travel time, start and end point index, and connection relationship order. When the data is recorded, the time-risk parameter information corresponding to each channel segment can be displayed at the end of the table, which is convenient for the subsequent steps to combine this information with the structural properties of the channel segment.

[0060] On the premise of having the cumulative parameter set of the channel section risk time, it is necessary to obtain the structural attribute data of the channel section based on the building information. First, check the wall material of the channel section from the building drawings or on-site inspection reports. When it is found that refractory bricks or steel plates are used on the channel wall, record the corresponding material number. If the wall material is classified as a concrete structure, register the corresponding code 2 or 3 in the material field. These codes are given by the channel material comparison table compiled in the early stage. Then, the anti-slip level of the ground should also be distinguished according to the nature of the building use. For example, the friction coefficient is queried in the inspection report of the channel paving material. If the friction coefficient is between 0.4 and 0.6, it corresponds to anti-slip level 2. If the friction coefficient exceeds 0.6, it is registered as level 3. These levels are obtained by averaging and summing multiple sets of friction force measurement values ​​on site. The maximum value is obtained by comparison. When the number of adjacent channel segments needs to be confirmed, the spatial topological relationship diagram of the channel segment is checked, and the numbers of other channels connected at the starting point and end point of the channel segment are found and counted. When the number links exceed 3, it is multi-adjacent. The number is recorded and stored in the adjacent channel segment field. Then, according to the wall material number, the ground anti-slip grade number, the number of adjacent channel segments and the time-product risk integral value in the aforementioned channel segment risk time accumulation parameter set, they are combined and arranged into a structural code and channel time-product risk joint matrix. In this matrix, each row corresponds to a channel segment, and each column corresponds to its material, ground grade, number of adjacent channel segments and time-product risk integral, ensuring that all fields can be quickly indexed during query. This matrix can be used in the subsequent process of calculating the path segment cost value.

[0061] formula: The benefit of the formula is that it incorporates the time-product risk integral of the channel through exponential decay, and makes comprehensive adjustments in the denominator or multiplication factor based on dimensions such as wall material, ground anti-slip level and the number of adjacent channel sections. At the same time, an absolute value of the radian value of the angle between the starting point and the end point is added at the end, which can make the turning or straight channel sections show differentiation in the cost evaluation, providing a quantitative means of multi-factor balance for the overall evacuation path optimization.

[0062] The steps to obtain are: This parameter represents the The channel time-integrated risk integral of each channel segment is obtained through the previously executed channel risk integral calculation.

[0063] The steps to obtain are: This parameter represents the The wall material of each channel section. Since the wall material itself is not numerical, it is necessary to establish a quantitative standard first. After integrating the data of wall material and fire state tolerance in the previous section, each material can be converted into a quantitative value in the form of a score or number. For example, ordinary concrete wall is set to 3 points, refractory brick wall is set to 2 points, metal steel plate wall is set to 5 points, and wooden wall is set to 6 points. These values ​​can be obtained through comprehensive measurement after combining multi-dimensional indicators such as heat resistance temperature and deformation characteristics. For example, the fire resistance time of various wall materials at a certain temperature is tested and the time when cracks appear is recorded. The average value is calculated after logarithmic transformation of all time lengths, and the result is linearly mapped to the range of 1 to 10. When it is found that a certain material has a longer average fire resistance time, a smaller value is assigned. If the material has a shorter fire resistance time, a larger value is assigned. When a certain channel section is given, if the detection confirms that its wall is made of refractory brick material, it can be recorded in the corresponding channel section. , when other materials appear in the same building and are subsequently classified into this field, they are recorded as different integer scores in a similar way.

[0064] The steps to obtain are: This parameter represents the The anti-slip level of the ground in each channel section is also assigned by quantitative method. For example, when testing the friction coefficient of the ground paving materials, the friction coefficient of each material in a wet and slippery state is counted, and then the measurement results are normalized and calculated. Then the level is set according to a certain classification method, and the higher friction coefficient is assigned to 1 and the lower friction coefficient is assigned to 5 or 6. When calculation is required, these level values ​​are directly substituted into the formula. For example, the test channel is paved with granite surface and the wet friction coefficient is 0.3. After comparing with the data of ceramic tiles and composite materials commonly used in this scenario, the anti-slip level is 3.5 in the normalized formula. For example, In this way, when the maximum friction coefficient is 0.7, if the measured value is 0.3, then , after rounding up .

[0065] The steps to obtain are: This parameter represents the The number of adjacent channel segments of a channel segment. The connection between different channel nodes can be confirmed through the channel topology relationship table obtained above. When forming the topology, count how many other channels the starting point and the end point are connected to. For example, if a channel segment has two adjacent lines at the starting point and three adjacent lines at the end point, then .

[0066] The steps to obtain are: This parameter represents the The radian value of the angle between the starting point and the end point of a channel segment can be calculated in a three-dimensional coordinate system by vector method. The end point coordinates are , then the vector from the starting point to the end point can be seen as If you need to calculate the angle with a reference axis such as the X-axis, you can use , then through Calculate the arc. For example, if a section of the channel is 30° relative to the X-axis on the plane, the corresponding arc is about 0.5236.

[0067] Calculation process: The following example is given to illustrate that, for example, a channel segment number k=12 has been obtained in the previous link , , , , , read the first part first ,in Very small, about Magnitude, , the denominator is approximately 1+4.2426=5.2426, and the numerator is approximately , the value after dividing the two is approximately , then watch the second part , first calculate , denominator 1+1.7918=2.7918, overall reciprocal About 0.4802, after taking the absolute value, it is still 0.4802. The result of the second part is about 0.358*0.4802=0.172. Add the two parts together, and the first part is about Extremely small, approximately 0.172 overall, .

[0068] The results show that when the channel segment risk score reaches 320, the exponential decay term Very small. When the wall material and the anti-slip level of the ground are both in a medium or deviation state, and the number of adjacent channel segments and the angle data are as above, the final calculated path segment cost is approximately 0.172. When the value is greater than 0.3, it indicates that the relative comfort or safety is higher. When the value is lower than 0.1, it indicates that the passage effect in this section of the channel is affected by more unfavorable factors.

[0069] The steps of obtaining the survival time escape route are as follows: according to the path segment cost value, the topological structure data of all path segments in the path network are obtained, the mapping of the network node connection relationship and the corresponding marking of the path segment cost value are performed, and the path search from the starting point of the path node to the exit is performed to generate a set of all feasible escape paths from the starting point to the exit; Based on the set of all feasible escape paths from the starting point to the exit, the cost values ​​of each path are traversed and accumulated segment by segment to generate a set of cumulative survival indicators of the path; According to the cumulative survival index set of the path, the path segment cost values ​​of all feasible escape paths are compared one by one, and the escape path with the largest path segment cost value is selected to establish the survival time escape route.

[0070] Specifically, according to the cost value of the path segment obtained above, firstly, the corresponding data of the nodes and channel segments recorded inside the building are queried centrally, and the cost value of each path segment is identified one by one in the corresponding topological structure. When the cost value of the path segment exceeds a certain threshold, the channel segment is marked for identification in the subsequent screening step. This threshold can refer to the statistics of the historical test results of each channel segment in the early stage. For example, the 75% quantile point is calculated for the cost value distribution data of 50 channel segments to obtain the threshold value of 0.2, and then the node connection relationship of all channel segments is indexed according to the sequence of spatial coordinates, and the starting point number and the end point number of the channel segment are read one by one for bidirectional association. When there are multiple parallel channels between the starting point and the end point, different channel segment numbers need to be marked respectively. After the mapping is completed, the path is started. Path search, starting from the specified starting point step by step, traverses the connecting channels in a breadth-first or depth-first manner, and registers the corresponding path segment cost value each time it advances to the next node. If the node has been visited, skip the branch and return to the previous node to continue searching. When a certain exit number is reached, the recorded complete node sequence is included in the result and the channel segment cost value distribution of the current entire path is saved. The cost value of each segment in the path is retained in order. If a fork in the road appears at a node and needs to be further expanded, all channels under this node are included in the query, and recursion or looping is used to ensure that the search covers every traversable node. After traversing all traversable channel segments, multiple optional node sequences from the starting point to the exit are listed, and the corresponding path segment cost values ​​are stored in the same data set.

[0071] After generating the above feasible escape path set from the starting point to the exit, it is necessary to traverse the path segment cost values ​​of each path and accumulate them segment by segment. First, read the node sequence of each path and obtain the cost value of the channel segment according to the channel number between adjacent nodes. If the channel segment cost value is greater than 0.5, it is marked with a conspicuous mark. The value of 0.5 comes from the statistics and preset of the regional distribution of the cost value. The specific process is to collect the cost values ​​of all channel segments, take the mean plus the standard deviation to calculate 0.5 as a more moderate dividing line, and then calculate the cost values ​​of all channel segments under the path according to The paths are added sequentially. If the cost value cannot be found due to complex terrain or abnormal nodes in some channel sections, they are skipped and marked at the corresponding positions. After the accumulation is completed, a total value is obtained, which is recorded as the stage-by-stage cumulative value of the path. If the path is long and has many nodes, it can be calculated in sections and a summation operation is performed at the end of each section. Finally, the cumulative survival index of the path is obtained after the statistics of all sections are completed. For the convenience of subsequent query, the cumulative results are registered together with the corresponding path number or path node sequence and summarized in the same table, where each path is equipped with a node list from the starting point to the end point and the cost value of each section.

[0072] After obtaining the cumulative survival index set of all feasible escape paths, these path indicators are compared with the cost values ​​of each path segment one by one. First, they are sorted according to the path number and the cumulative survival indicators are arranged from large to small or from high to low. At the same time, the path with the higher cost value is identified. When comparing, if the difference in the cumulative survival indicators of multiple paths is found to be no more than 0.05, the specific cost value comparison between the segments in each path can be continued. If multiple paths have a significant drop in cost value at key nodes, they will be marked in the table. After confirming the cost value of the optimal path segment, it will be placed at the top of the result, and the node sequence of this path will be extracted to form the start-to-exit sequence of the path. Here, it is necessary to check again in the node sequence for duplication or cross-connection to ensure that the path is complete and correct. If there is no abnormality, this path will be registered as an available survival time escape route.

[0073] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A fire optimal escape route planning method based on a multimodal fusion algorithm, characterized in that: The following steps are involved: Obtain temperature, smoke, oxygen sensor readings and building information data, calculate the multi-source fusion hazard degree at each location and time of the building, obtain the location time series fusion hazard value, integrate the values ​​at different time points based on the location time series fusion hazard value, and establish a time-varying hazard probability distribution map; Based on the time-varying hazard probability distribution diagram and the information of the middle beam and column in the building information data, the damage to the structural components caused by thermal effects is analyzed, and the long-term fragility factor of the components is calculated; Based on the long-term vulnerability factor of the component, the long-term vulnerability factor of the component is superimposed on the time-varying hazard probability distribution map to generate a time-varying structural hazard probability distribution map; According to the time-varying structural hazard probability distribution diagram, the risk of each channel section within the travel time is extracted, and the channel time-integrated risk integral of each channel section is calculated. Based on the channel time-integrated risk integral, it is converted into the survival probability of passing the channel section to establish the path segment cost value; The path segment cost value is applied to analyze the escape routes from the starting point to the exit in the path network, the survival index of each route is cumulatively calculated, the path is selected according to the survival index, and the survival time escape route is obtained.

2. The method for planning the optimal fire escape route based on a multimodal fusion algorithm according to claim 1 is characterized in that: The steps for obtaining the position time series fusion hazard value are as follows: deploying positioning temperature sensors, smoke sensors and oxygen sensors in the building structure, and synchronously collecting data at the current time under each position number every 10 seconds to generate original temperature data, original smoke data and original oxygen data with timestamp, position number and sensor type; Based on the original temperature data, original smoke data and original oxygen data, the fluctuating values, missing values ​​and mutation values ​​that appear in two consecutive time periods are detected by differential sliding windows and replaced with upper and lower neighbor values ​​to generate continuous and complete temperature data, smoke data and oxygen data sequences; According to the continuous and complete temperature data, smoke data and oxygen data sequences, the position time series fusion hazard value of each sampling node is calculated.

3. The method for planning the optimal fire escape route based on a multimodal fusion algorithm according to claim 1, characterized in that: The step of acquiring the time-varying hazard probability distribution map is as follows: according to the position time series fusion hazard value, a fixed period time window is divided, and the position time series fusion hazard value of each spatial node in each time window is interval graded and numerical distribution statistics are performed to generate a hazard level distribution sequence; Based on the hazard level distribution sequence, a spatial topological network is established and grid interpolation is performed according to the three-dimensional coordinate information of the building structure space, and the position time series fusion hazard value corresponding to each spatial node is mapped into spatial coordinates to generate a time-varying hazard probability distribution map that matches the building space coordinate system.

4. The method for planning the optimal fire escape route based on a multimodal fusion algorithm according to claim 1, characterized in that: The steps of obtaining the long-term vulnerability factor of the component are as follows: according to the time-varying hazard probability distribution diagram and the beam and column information in the building information data, the node index corresponding to each component number is extracted, and the spatial coordinates, component material thermal conductivity, component cross-sectional area, component surface temperature, component length and component surface reflectivity are called from the corresponding node to generate a component thermal environment input parameter set; Based on the component thermal environment input parameter set, the thermal energy transfer per unit area of ​​the component in a single cycle is calculated using the thermal conductivity of the material, and the thermal deflection effect caused by the temperature difference in the length direction of the component and the reflectivity of the component surface is combined to obtain the component temperature difference stress tensor value sequence; The long-term fragility factor of the component is calculated according to the component temperature difference stress tensor value sequence.

5. The method for planning the optimal fire escape route based on a multimodal fusion algorithm according to claim 1, characterized in that: The step of obtaining the time-varying structural hazard probability distribution map is as follows: according to the long-term vulnerability factor of the component, calling the time-varying hazard probability distribution map, extracting the spatial three-dimensional coordinate information of each spatial node, matching the spatial coordinate with the component number, and combining the long-term vulnerability factor of the component, mapping the long-term vulnerability factor of the component to the node spatial coordinate, and generating a long-term vulnerability factor space mapping table associated with the node and the component; Based on the long-term vulnerability factor spatial mapping table associated with the nodes and components, a spatial topological network is constructed to generate a time-varying structural hazard probability distribution map.

6. The method for planning the optimal fire escape route based on a multimodal fusion algorithm according to claim 1, characterized in that: The step of obtaining the time-integrated risk integral of the channel is as follows: according to the time-varying structural hazard probability distribution diagram, extracting the three-dimensional coordinates of the starting point and the end point of each channel segment in the building space coordinate system, and according to the channel segment structure number, obtaining the hazard probability value of the spatial position covered by the channel segment in the time-varying structural hazard probability distribution diagram time step by time step, forming a time series set of the structural hazard probability of each channel segment at each time step; Based on the set of structural danger probability time series, combined with the length of each channel segment, the average walking speed under the current evacuation state and the spatial span from the start point to the end point of the channel segment, the travel time interval required to cross the channel segment is calculated, and the structural danger probability value within the time interval is screened to obtain the danger probability sequence within the travel time interval of the channel segment; According to the danger probability sequence within the passage time interval of the passage section, the passage time integral risk integral is calculated, and the calculation formula is: in, For the The channel time integral risk integral of each channel segment, For the The passage time of each channel section, For the Channel segments at time The structural hazard probability value extracted from the time-varying structural hazard probability distribution diagram at any moment, For the The length of the channel segment, For the The average walking speed of each channel segment, is the structural danger probability value corresponding to the passage section at the beginning of passage, For the The spatial distance between the nodes at both ends of a channel segment.

7. The method for planning the optimal fire escape route based on a multimodal fusion algorithm according to claim 1, characterized in that: The step of obtaining the path segment cost value is as follows: extracting the channel time product risk integral of each channel segment, and obtaining the total travel time, the starting point and the end point space index numbers corresponding to the channel segment, and arranging them in combination with the node topological order of the channel segment to generate a channel segment risk time accumulation parameter set; Based on the channel segment risk time accumulation parameter set, the structural attribute information of the channel segment is obtained, including the channel segment wall material, the ground anti-slip level and the number of adjacent channel segments, to form a joint matrix of structural code and channel time product risk; According to the structural coding and the channel time product risk joint matrix, the path segment cost value is calculated, and the calculation formula is: in, For the The path segment cost of a channel segment, For the The channel time integral risk integral of each channel segment, For the The wall material of each channel segment, For the The anti-slip level of the ground in each channel section, For the The number of adjacent channel segments per channel segment, For the The radian value of the angle between the starting point and the end point of a channel segment.

8. The method for planning the optimal fire escape route based on a multimodal fusion algorithm according to claim 1, characterized in that: The step of acquiring the survival time escape route is: according to the path segment cost value, acquiring the topological structure data of all path segments in the path network, mapping the network node connection relationship and the corresponding marking of the path segment cost value, and performing a path search from the path node starting point to the exit, generating a set of all feasible escape paths from the starting point to the exit; Based on the set of all feasible escape paths from the starting point to the exit, traverse and accumulate the cost values ​​of each path segment by segment to generate a set of cumulative survival indicators of the path; According to the cumulative survival index set of the path, the path segment cost values ​​of all feasible escape paths are compared one by one, and the escape path with the maximum path segment cost value is selected to establish a survival time escape route.

9. The escape route planning system of the fire optimal escape route planning method based on multimodal fusion algorithm according to any one of claims 1 to 8, characterized in that: include: Environmental data acquisition module, which collects data from temperature, smoke, oxygen sensors and building information to obtain environmental and building data sets; A multi-source data fusion module integrates data at each location and time based on the environment and building data sets to generate a time-varying hazard probability distribution map; The structural damage analysis module uses the time-varying hazard probability distribution diagram to perform thermal damage analysis on the central beams and columns of the building to obtain component fragility factors; A risk assessment module superimposes the component vulnerability factor with the time-varying hazard probability distribution map, calculates the risk value of each channel section according to the risk of each channel section during the travel time, converts the risk value into the survival probability of the channel, and establishes the path segment cost value; The path planning module analyzes possible escape routes from the starting point to the exit in the path network according to the path segment cost value, selects the escape route with the best survival time, and obtains the optimal escape path.

Images