Fire early warning method and system based on three-dimensional modeling

Abstract

The present application provides a fire warning method and system based on three-dimensional modeling. The method can determine the abnormal indicators of fire risk in different spatial units by combining the spatial structural characteristics of the three-dimensional digital twin model with the state evolution pattern of the physical field, and then determine the risk cloud map through all the abnormal indicators; extract the risk subdomain of the fire based on the risk cloud map, and then generate the heat flux density field matrix between different spatial units; determine the risk fit of different spatial units by combining the heat flux diffusion information with the monitoring information of the fire situation in the target building, and perform time series aggregation on the fire risk in the target building based on all the risk fits combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential; monitor the fire spread situation of the target building based on the fire risk propagation potential, and output a fire spread warning signal. Using the solution of the present application, the diffusion trajectory of the fire heat flow can be accurately identified in the three-dimensional model.

Description

Technical Field

[0001] The present application relates to the field of fire warning technology, and more specifically, to a fire warning method and system based on three-dimensional modeling. Background Art

[0002] Fire early warning is a critical safety system that uses advanced technology to monitor, analyze, and assess various characteristics before a fire breaks out in real time, issuing advance warnings and alerting relevant personnel to take countermeasures. Its core goal is to accurately identify danger signals in the early stages of a fire, buying valuable time for evacuation and firefighting, and minimizing loss of life and property. Fire early warning plays a vital role in all types of settings. In densely populated areas like commercial complexes, office buildings, and residential communities, it can promptly alert people to evacuate dangerous areas. In industrial sites like factories and warehouses, it can notify staff in advance to take measures such as power outages and valve closures to prevent the fire from spreading. With the continuous advancement of technology, fire early warning systems are developing towards intelligent, integrated, and unmanned features. They are deeply integrated with systems such as video surveillance and fire alarm systems to form a comprehensive, multi-layered fire prevention and control system, providing a more solid safety guarantee for people's production and daily lives.

[0003] In fire warning, three-dimensional modeling is a core technical means to accurately locate, dynamically deduce and visually warn fire hazards by constructing spatial three-dimensional digital models; this technology breaks through the limitations of traditional two-dimensional plane monitoring and integrates building structures, environmental parameters and real-time monitoring data into a three-dimensional digital twin scene, providing multi-dimensional spatial analysis capabilities for fire prevention. Currently, 3D modeling technology has expanded from single-building applications to city-wide monitoring. By integrating multi-source 3D data, it is building a three-dimensional fire prevention and control network covering communities and parks. Its core value lies in upgrading fire warning from "two-dimensional alarms" to "three-dimensional rehearsals," making prevention and control measures more spatially targeted and significantly improving fire response efficiency in complex scenarios. However, existing fire warning methods based on 3D modeling often remain at the level of fire source location and rely on simplified algorithms for dynamic simulation of fire spread. This method lacks adaptability to complex building structures (i.e., it fails to couple building material properties such as wall thermal conductivity and ventilation system resistance with spatial structural parameters). As a result, traditional methods cannot accurately calculate the diffusion trajectory of heat flow in different spatial units (for example, in an underground shopping mall fire, it is difficult to predict the vertical spread speed of smoke through escalator shafts and the concentration distribution of each store). As a result, they cannot provide accurate fire warning paths for personnel evacuation and firefighting deployment in complex scenarios. Therefore, how to accurately identify the diffusion trajectory of fire heat flow in 3D models has become a difficult problem facing the industry. Summary of the Invention

[0004] The present application provides a fire warning method and system based on three-dimensional modeling, which can accurately identify the diffusion trajectory of fire heat flow in the three-dimensional model.

[0005] In a first aspect, the present application provides a fire early warning method based on three-dimensional modeling, comprising the following steps:

[0006] Obtaining a three-dimensional digital twin model of a target building, wherein the three-dimensional digital twin model includes a plurality of spatial units in the target building;

[0007] Based on the spatial structural characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field, the fire risk fluctuation indicators at different spatial units are determined. Then, risk fitting is performed on all spatial units using all fluctuation indicators to obtain a risk cloud map of the fire distribution in the target building.

[0008] Extracting a fire risk subdomain from the target building according to the risk cloud map, and then generating a heat flux density field matrix between different spatial units in the risk subdomain based on a fire risk evolution model;

[0009] The heat flux density field matrix is ​​combined with the monitoring information of the fire situation in the target building to determine the risk fitting degree of different spatial units, and then the fire risk in the target building is aggregated in time series based on all the risk fitting degrees combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential energy of different spatial units;

[0010] The fire spread situation of the target building is monitored based on all fire risk propagation potentials, and a fire spread warning signal is output.

[0011] In some embodiments, determining the fire risk anomaly indicators at different spatial units based on the spatial structural characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field specifically includes:

[0012] Obtaining spatial structural characteristics of each spatial unit in the three-dimensional digital twin model;

[0013] Perform dynamic simulation of the target building based on the state evolution model of the physical field to obtain the evolution process data of the physical field in each spatial unit;

[0014] The spatial structure characteristics are correlated with the evolution process data to determine the abnormal fire risk indicators at different spatial units.

[0015] In some embodiments, risk fitting is performed on all spatial units using all abnormal indicators to obtain a risk cloud map of fire distribution in the target building, specifically including:

[0016] Fit and calculate the risk level of each space unit in the target building based on all abnormal indicators;

[0017] The fitting calculation results are spatially mapped to generate a risk cloud map of fire distribution in the target building.

[0018] In some embodiments, extracting a fire risk subdomain from the target building according to the risk cloud map specifically includes:

[0019] Performing threshold determination on the risk level of each spatial unit in the risk cloud map;

[0020] The set of spatial units that meet the threshold discrimination conditions is extracted as the fire risk subdomain.

[0021] In some embodiments, generating a heat flux density field matrix between different spatial units in the risk subdomain based on the fire risk evolution model specifically includes:

[0022] Obtaining status information of each spatial unit in the risk subdomain;

[0023] Setting boundary conditions during fire evolution using the state information;

[0024] A heat flux density field matrix between different spatial units in the risk subdomain is generated according to the boundary conditions and combined with a fire risk evolution model.

[0025] In some embodiments, determining the risk fit of different spatial units by combining the heat flux density field matrix with the monitoring information of the fire situation in the target building specifically includes:

[0026] Obtain monitoring information on fire conditions in target buildings;

[0027] determining a plurality of information differences between the monitoring information and the heat flux density field matrix;

[0028] The risk fit of different spatial units is determined by all information differences.

[0029] In some embodiments, the fire risk in the target building is aggregated in time series based on all risk fitting degrees combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential of different spatial units, specifically including:

[0030] Obtaining the local temperature field of each spatial unit in the three-dimensional digital twin model;

[0031] By combining all risk fitting degrees with the local temperature field of each spatial unit for time series analysis, the time series risk aggregation model of the target building is obtained;

[0032] Determining the fire risk value of each spatial unit at different time points according to the temporal risk aggregation model;

[0033] The fire risk spread potential of different spatial units is determined through all fire risk values.

[0034] In some embodiments, monitoring the fire spread situation of a target building based on all fire risk propagation potentials and outputting a fire spread warning signal specifically includes:

[0035] Analyze the changing trend of fire situation in target buildings based on all fire risk propagation potential;

[0036] Evaluate the changing trend of the fire situation according to preset fire warning rules;

[0037] When the evaluation result reaches the warning condition, a fire spread warning signal is generated and output.

[0038] In some embodiments, the spatial unit is a finite element unit with uniform spatial properties.

[0039] In a second aspect, the present application provides a fire warning system based on three-dimensional modeling, comprising:

[0040] An acquisition module, configured to acquire a three-dimensional digital twin model of a target building, wherein the three-dimensional digital twin model includes a plurality of spatial units in the target building;

[0041] A processing module is used to determine the anomaly index of fire risk at different spatial units based on the spatial structural characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field, and then perform risk fitting on all spatial units through all the anomaly indexes to obtain a risk cloud map of fire distribution in the target building;

[0042] The processing module is further configured to extract a fire risk subdomain from the target building according to the risk cloud map, and then generate a heat flux density field matrix between different spatial units in the risk subdomain based on the fire risk evolution model;

[0043] The processing module is further configured to determine the risk fit of different spatial units by combining the heat flux density field matrix with the monitoring information of the fire situation in the target building, and then perform time series aggregation of the fire risk in the target building based on all the risk fits combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential energy of the different spatial units;

[0044] The execution module is used to monitor the fire spread situation of the target building based on all fire risk propagation potentials and output a fire spread warning signal.

[0045] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0046] In the fire warning method and system based on three-dimensional modeling provided by the present application, a three-dimensional digital twin model of a target building is obtained, and the three-dimensional digital twin model includes multiple spatial units in the target building; the abnormal indicators of fire risks at different spatial units are determined based on the spatial structure characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field, and then all spatial units are subjected to risk fitting through all abnormal indicators to obtain a risk cloud map of the fire distribution in the target building; the risk subdomain of the fire is extracted from the target building according to the risk cloud map, and then a heat flux density field matrix between different spatial units in the risk subdomain is generated based on the fire risk evolution model; the risk fitting degree of different spatial units is determined through the heat flux density field matrix combined with the monitoring information of the fire situation in the target building, and then the fire risk in the target building is time-series aggregated according to all risk fitting degrees combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential energy of different spatial units; the fire spread situation of the target building is monitored based on all fire risk propagation potential energy, and a fire spread warning signal is output.

[0047] It can be seen that in this application, first, after obtaining the three-dimensional digital twin model of the target building, the fire risk subdomain is extracted based on the abnormal indicators of fire risks in different spatial units. Then, by extracting the state information of the spatial units in the risk subdomain (such as material combustion performance, ventilation parameters) and setting it as the boundary conditions of the fire risk evolution model, the fire dynamics equations are combined to simulate the heat flow transfer process between units, and key parameters such as heat conduction rate, direction and delay time can be quantified (such as updating the temperature field and heat flux density field by solving the energy conservation equation in the specification); the parameters directly reflect the actual propagation path of the heat flow in the building structure, for example, the heat conduction rate is faster when passing through a unit with lower wall thermal resistance, or the diffusion direction is changed by the influence of the ventilation system; in addition, by introducing the difference analysis between real-time monitoring data and model prediction values ​​(such as Euclidean distance to calculate the heat conduction rate deviation), the simulation results of the heat flow trajectory can be dynamically corrected to avoid path misjudgment caused by simplified algorithms; then, the heat flux density field matrix is ​​combined with the monitoring information of the fire in the target building to determine the heat flow paths of different spatial units. The risk fit of each element is then combined with the local temperature field of each spatial unit to perform a temporal aggregation of the fire risk in the target building based on all risk fits, thereby obtaining the fire risk propagation potential of different spatial units. Specifically, by inputting the fire risk value and the physical properties of the spatial unit (such as wall thermal resistance and combustible calorific value) into a physical field coupling model or a machine learning model, indicators such as the heat flow conduction rate and the heat release rate gradient are calculated. Combined with the dynamic weighting assigned during the fire stage (for example, focusing on heat conduction rate in the early stage), a comprehensive assessment of the potential ability of each unit to spread fire to the surrounding area can be achieved. This quantitative assessment can identify key paths for heat flow diffusion. For example, units with high fire risk propagation potential often constitute the main channels for heat flow diffusion. The continuous distribution of their risk value field formed through spatiotemporal interpolation can intuitively represent the spatial diffusion range of the heat flow (such as the risk value field generated through Kriging interpolation in the manual). Finally, the fire spread situation of the target building is monitored based on all fire risk propagation potentials, and a fire spread warning signal is output. In summary, this solution can accurately identify the diffusion trajectory of fire heat flow in a three-dimensional model. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a flow chart of a fire warning method based on three-dimensional modeling according to some embodiments of the present application;

[0049] Figure 2 is a flowchart illustrating implementation of threshold discrimination according to some embodiments of the present application;

[0050] Figure 3 is a schematic diagram of a process for determining risk fitness according to some embodiments of the present application;

[0051] Figure 4is a structural diagram of a fire warning system based on three-dimensional modeling according to some embodiments of the present application;

[0052] Figure 5 This is a diagram of the internal structure of a computer device for implementing a fire warning method based on three-dimensional modeling according to some embodiments of the present application. DETAILED DESCRIPTION

[0053] In order to better understand the technical solution in this embodiment, the technical solution in this embodiment will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0054] refer to Figure 1 , which is a flow chart of a fire warning method based on 3D modeling according to some embodiments of the present application. The fire warning method 100 based on 3D modeling mainly includes the following steps:

[0055] In step 101, a three-dimensional digital twin model of a target building is obtained, where the three-dimensional digital twin model includes a plurality of spatial units in the target building.

[0056] In specific implementation, a three-dimensional sensing device (such as a lidar device, a structured light scanning module or a depth camera) can be deployed in the target building to perform a full-coverage spatial scan of the target building, and obtain real-time spatial structure images including walls, doors and windows, passages, ceilings, etc.; at the same time, by configuring a multi-source environmental acquisition module (such as a temperature sensor, a smoke sensor, an infrared imaging device) on the same monitoring platform or in adjacent locations, the environmental status images inside the target building at different time points can be obtained as the three-dimensional digital twin model in this application.

[0057] It should be noted that the three-dimensional digital twin model is a highly simulated representation of the real target building in the digital space, with mapping characteristics in multiple dimensions such as geometry, physics, and behavior; the model not only includes the static structural information of the target building, such as wall structure, channel distribution, door and window positions, etc., but also includes its dynamic evolution information, such as temperature changes, risk diffusion, personnel flow and other multi-dimensional state data; in addition, the spatial unit in the three-dimensional digital twin model is the smallest modeling unit after the target building is discretely divided, that is: the spatial unit is a finite element unit with unified spatial properties, where each spatial unit carries its corresponding geometric coordinates, structural properties and physical state parameters (such as temperature value, risk value, ventilation rate, etc.).

[0058] In step 102, the abnormal indicators of fire risks in different spatial units are determined based on the spatial structure characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field, and then all spatial units are subjected to risk fitting through all abnormal indicators to obtain a risk cloud map of fire distribution in the target building.

[0059] In some embodiments, determining the fire risk anomaly indicators at different spatial units based on the spatial structural characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field can be achieved by the following steps:

[0060] Obtaining spatial structural characteristics of each spatial unit in the three-dimensional digital twin model;

[0061] Perform dynamic simulation of the target building based on the state evolution model of the physical field to obtain the evolution process data of the physical field in each spatial unit;

[0062] The spatial structure characteristics are correlated with the evolution process data to determine the abnormal fire risk indicators at different spatial units.

[0063] Preferably, the unit volume, surface area and aspect ratio can be extracted from the geometric data of the three-dimensional digital twin model; then, the spatial unit adjacency matrix is ​​constructed and the connectivity of each unit is calculated; then, the fire resistance limit parameters of the fire separation components at the unit boundary are detected, the separation defect ratio is calculated, and the Dijkstra algorithm is used to calculate the shortest distance from the unit to the nearest safe exit and the path tortuosity; finally, the ventilation volume and ventilation efficiency of the unit are calculated based on fluid mechanics, the thermal physical parameters of the enclosing structure are extracted and the equivalent thermal resistance is calculated, and then the unit volume, surface area, aspect ratio, connectivity, fire separation parameters, evacuation path parameters, ventilation parameters and heat conduction parameters are used as the spatial structural characteristics of each spatial unit; in other embodiments, the three-dimensional spatial unit can be projected onto a two-dimensional plane by calculating the edge nodes with limited resources, and simplified geometric features such as perimeter and area can be extracted, and high-dimensional structural features can be compressed by principal component analysis. This application does not limit this.

[0064] It should be noted that the spatial structure characteristics described in this application refer to a multi-dimensional parameter set formed by extracting and calculating the geometric data, topological relationships, structural safety and environmental parameters of each spatial unit in the three-dimensional digital twin model, and calculating indicators such as unit volume, surface area, connectivity, fire separation parameters, and ventilation parameters; the spatial structure characteristics are represented as a set of structured parameters, each parameter corresponds to a quantitative description of the spatial unit in terms of geometric form, connection relationship, safety performance and environmental characteristics, and the parameter values ​​directly reflect the differences in the structural properties of the spatial units, which can be used for subsequent analysis of the characteristics of different spatial units in terms of fire risk bearing capacity.

[0065] In specific implementation, the target building is dynamically simulated according to the state evolution pattern of the physical field, and the evolution process data of the physical field in each spatial unit can be obtained in the following way, namely: first, the physical field simulation model is determined, among which the Fourier heat conduction equation is selected for the heat conduction process, and the large eddy simulation or regional model is used for the smoke diffusion, and the model parameters are set according to the characteristics of the target building, such as material thermal conductivity, ambient wind speed, etc.; then, the spatial units in the three-dimensional digital twin model are used as simulation calculation grids, and initial conditions are assigned to each spatial unit, including initial temperature, oxygen concentration, combustible material distribution, etc.; then, the simulation time step is set, 10 seconds / step is used in the initial warning stage, and it is switched to 1 second / step in the fire development stage. step, start the simulation calculation; finally, at the end of each time step, extract the physical field parameters in each spatial unit, including temperature, heat flux density, flue gas concentration, CO concentration, etc., to form the evolution process data of the physical field in each spatial unit; wherein, as a preferred embodiment, parallel computing technology can be used to distribute the simulation tasks of different spatial units to multiple computing cores to improve the simulation efficiency; in other implementations, a reduced-order modeling method can be introduced to simplify the physical field model through techniques such as orthogonal decomposition (POD) to reduce the computational complexity while ensuring the computational accuracy, and this application does not impose any limitation on this.

[0066] It should be noted that the evolution process data described in this application refers to the dynamic simulation of the target building according to the state evolution pattern of the physical field. By setting the simulation model parameters, dividing the calculation grid, assigning initial conditions and performing simulation calculations, the temperature, heat flux density, smoke concentration, CO concentration and other physical field parameters of each spatial unit are extracted at each time step, and then the time series data set is formed by arranging them in chronological order; the evolution process data is represented as a set of multi-dimensional time series values, each value corresponds to the instantaneous state of the physical field parameters in the spatial unit at a specific time point, and its numerical changes directly reflect the dynamic evolution of the physical field in the spatial unit over time, which can be used for the subsequent construction of fire risk evolution models and prediction of fire development trends.

[0067] In specific implementation, the spatial structure characteristics are correlated with the evolution process data to determine the anomaly index of fire risk at different spatial units. This can be achieved in the following manner: first, a feature correlation matrix is ​​constructed with the spatial structure feature parameters as row vectors and the evolution process data parameters as column vectors; then, feature selection is performed based on the feature correlation matrix, and parameter pairs with absolute values ​​of correlation coefficients ≥ 0.6 are retained to construct a high-correlation feature subset; finally, the spatial structure feature parameters and the evolution process data parameters in the high-correlation feature subset are input into a multi-layer perceptron model, and the abnormal risk probability value of each spatial unit is output through a soft maximum function, and the abnormal risk probability value is then used as the anomaly index of fire risk at the corresponding spatial unit; among them, as a preferred embodiment, a random forest algorithm can be used to sort the importance and reduce the dimension of the high-correlation feature subset; in other implementations, evidence theory can be used to fuse multi-source features to improve the robustness of anomaly indicators, and this application does not limit this.

[0068] It should be noted that the anomaly index described in this application refers to a quantitative evaluation parameter constructed by performing correlation analysis on spatial structure characteristic parameters and evolution process data parameters; the anomaly index is represented by a probability value output by a multi-layer perceptron model through a soft maximum function, and this value corresponds to the risk abnormality degree of each spatial unit in a specific time dimension, and its numerical value directly reflects the abnormal coupling state of the physical field parameters and structural characteristic parameters in the spatial unit.

[0069] In some embodiments, risk fitting is performed on all spatial units using all abnormal indicators to obtain a risk cloud map of fire distribution in the target building. This can be achieved by using the following steps:

[0070] Fit and calculate the risk level of each space unit in the target building based on all abnormal indicators;

[0071] The fitting calculation results are spatially mapped to generate a risk cloud map of fire distribution in the target building.

[0072] In specific implementation, the risk level of each spatial unit in the target building is fitted and calculated based on all the anomaly indicators. The following method is used: first, the number of components of the Gaussian mixture model is determined based on the Bayesian information criterion; then, the anomaly indicators are standardized and preprocessed, and each indicator is mapped to the interval [0, 1] to obtain a normalized value; then, all normalized values ​​are input into the Gaussian mixture model, and the mean, covariance and weight parameters of each Gaussian component are iteratively calculated to construct a risk fitting model; then, the mean vector, covariance matrix and mixing coefficient of the risk fitting model are iteratively updated through the expectation maximization algorithm to maximize the log-likelihood function; finally, the anomaly indicator of each spatial unit is input into the updated risk fitting model, and the posterior probability that the spatial unit belongs to the high-risk category is output, and this probability value is used as the fitting calculation result of the risk level of the corresponding spatial unit. Among them, as a preferred embodiment, the Z-score normalization method can be used in the standardization preprocessing; in other implementations, the process of determining the number of components of the Gaussian mixture model can be optimized by cross-validation, and this application does not limit this.

[0073] In specific implementation, the fitting calculation results are spatially mapped to generate a risk cloud map of the fire distribution in the target building. This can be achieved in the following way: first, the target building is gridded and divided into square or hexagonal grid units of uniform size, and a coordinate mapping relationship between each grid unit and the risk level value obtained by the fitting calculation is established; then, the discrete risk level values ​​are processed using the Kriging interpolation algorithm, and the grid gap areas are estimated and filled with risk values ​​through spatial autocorrelation analysis to construct a continuous risk value distribution surface; then, a risk level color mapping table is set, and low, medium, and high risk levels are respectively corresponding to color gradients such as blue, yellow, and red, and the surface risk values ​​are converted into corresponding color values; finally, the risk curve with color information is mapped to the target building. The three-dimensional digital twin model of the surface and the target building is layered and rendered to generate a visual fire distribution risk cloud map; as a preferred embodiment, the grid density can be dynamically adjusted according to the complexity of the spatial unit during grid division; in other implementations, the inverse distance weighted interpolation method can be used instead of the Kriging interpolation algorithm to estimate the risk value, which is not limited in this application; preferably, establishing a coordinate mapping relationship between each grid unit and the risk level value specifically includes: constructing a Cartesian coordinate system to divide the target building into square grids, and determining the coordinates (x, y) of the lower left corner vertex of each grid; then, each grid coordinate is matched one-to-one with the risk level value obtained by fitting calculation, and stored in a hash table in the form of key-value pairs to form a mapping relationship between grid coordinates and risk values.

[0074] In step 103, a fire risk subdomain is extracted from the target building according to the risk cloud map, and then a heat flux density field matrix between different spatial units in the risk subdomain is generated based on the fire risk evolution model.

[0075] In some embodiments, extracting the fire risk subdomain from the target building according to the risk cloud map can be achieved by using the following steps:

[0076] Performing threshold determination on the risk level of each spatial unit in the risk cloud map;

[0077] The set of spatial units that meet the threshold discrimination conditions is extracted as the fire risk subdomain.

[0078] Preferably, the maximum inter-class variance method can be used to automatically calculate the threshold value of the risk level values ​​of all spatial units in the risk cloud map to obtain the risk threshold value, wherein the maximum inter-class variance method determines the optimal segmentation threshold value by maximizing the inter-class variance; if different scene characteristics need to be considered, a manual intervention mechanism can be set to manually adjust the threshold value based on historical fire data or expert experience; therefore, as a preferred embodiment, reference Figure 2 As shown, this figure is a flowchart of threshold discrimination shown in some embodiments of the present application. The threshold discrimination of the risk level of each spatial unit in the risk cloud map can be achieved in the following manner, namely: comparing the risk level value of each spatial unit with the risk threshold, marking the spatial units with risk level values ​​greater than or equal to the risk threshold as high-risk units, and marking the spatial units with risk level values ​​less than the risk threshold as low-risk units, to complete the threshold discrimination process; wherein, as a preferred embodiment, a dynamic sliding window method can be used to calculate the regional threshold of the risk cloud map to adapt to the differences in risk characteristics under different environments in the target building; in other implementations, a machine learning model (such as a support vector machine) can be used to train the threshold discrimination model, and the discrimination result can be automatically output according to the input spatial unit characteristics, which is not limited in the present application.

[0079] In specific implementation, the following method can be used to extract a set of spatial units that meet the threshold judgment conditions as a fire risk subdomain, namely: first, a connectivity analysis is performed on the spatial units that have passed the threshold judgment, and an eight-neighborhood connected area marking algorithm is used to identify adjacent high-risk spatial units whose risk level values ​​are greater than or equal to the threshold; then, the interconnected high-risk spatial units are clustered, and each cluster set is regarded as a potential risk subdomain; then, the subdomains obtained by clustering are filtered, and a minimum area threshold is set (such as clusters containing less than 3 spatial units are eliminated) to remove isolated small areas caused by noise or misjudgment; finally, the retained high-risk spatial unit cluster set is used as the final fire risk subdomain; among them, as a preferred embodiment, a topological distance constraint can be introduced in the connectivity analysis to limit the merging of spatial units across fire protection zones to ensure that the risk subdomain meets the actual physical isolation conditions; in other implementations, the boundary division of the risk subdomain can be optimized by the minimum spanning tree algorithm in graph theory, which is not limited in this application.

[0080] It should be noted that the fire risk subdomain described in this application refers to a set of spatial units that meet the threshold conditions and are interconnected, which are extracted after threshold judgment is performed on the risk levels of spatial units in the risk cloud map; the fire risk subdomain is structurally a clustered area formed by high-risk spatial units through eight-neighborhood connectivity analysis, which is retained after minimum area filtering. Its range directly reflects the degree of concentration and spatial distribution characteristics of fire risks in the target building, and can be used for subsequent fire resource scheduling and fire prevention and control strategy formulation.

[0081] In some embodiments, generating a heat flux density field matrix between different spatial units in the risk subdomain based on the fire risk evolution model can be achieved by the following steps:

[0082] Obtaining status information of each spatial unit in the risk subdomain;

[0083] Setting boundary conditions during fire evolution using the state information;

[0084] A heat flux density field matrix between different spatial units in the risk subdomain is generated according to the boundary conditions and combined with a fire risk evolution model.

[0085] Preferably, after extracting the risk level value of each spatial unit from the grid coordinate mapping table corresponding to the risk subdomain, the physical property data of the spatial unit can be obtained through a geographic information system or a building information model, including the combustion performance grade of building materials (such as Class A non-combustible, Class B1 flame retardant), ventilation conditions (such as the proportion of door and window area), and fire source load density (such as MJ / m²); then, combined with real-time monitoring data (such as temperature and smoke concentration sensor data) or historical statistical data, a state information vector containing multidimensional data such as risk level, combustion performance, ventilation parameters, and fire source load is generated as the state information of each spatial unit in the risk subdomain.

[0086] In specific implementation, the boundary conditions for fire evolution set by the state information can be achieved in the following manner, namely: first, the building physical property parameters in the state information (such as the thermal conductivity coefficient of the wall material and the heat release rate of the combustible combustion) are mapped to the material boundary conditions of the fire risk evolution model; then, the environmental parameters (such as the space ventilation volume and the air exchange rate) are converted into the fluid dynamic boundary conditions of the model; then, the fire risk level corresponding to the risk level value is mapped to the probability trigger boundary condition of the model, and the fire probability threshold corresponding to different risk levels is set; among them, as a preferred embodiment, a parameter sensitivity analysis method can be used to weight the key parameters in the state information (such as the ventilation coefficient and the fire source load), and the high-weight parameters are preferentially set as strong constraint boundary conditions; in other implementations, a mapping relationship table between state information and boundary conditions can be established through a machine learning algorithm, and this application does not limit this.

[0087] In specific implementation, the heat flux density field matrix between different spatial units in the risk subdomain is generated according to the boundary conditions in combination with the fire risk evolution model, which can be achieved in the following manner: first, based on the computational grid topology of the fire risk evolution model, a one-to-one mapping relationship between the spatial units in the risk subdomain and the model grid nodes is established, and the material physical parameters (including material thermal conductivity, heat storage characteristics, etc.) and fluid dynamics parameters (such as vent airflow velocity, spatial pressure distribution, etc.) in the boundary conditions are configured for each network node to complete the initial parameter configuration of the model; secondly, based on the quantitative mapping relationship between the risk level value and the fire probability threshold (the quantitative mapping relationship can be constructed by fitting historical fire data or theoretical model), a probability weighted random sampling strategy (such as multinomial distribution sampling or roulette wheel selection method) is adopted to determine the initial fire source node in the risk subdomain; according to the fire source load density of the initial fire source node (characterizing the potential heat release capacity of combustibles per unit area), its initial heat release rate (reflecting the initial combustion of combustibles per unit area) is calculated. Heat release intensity); at the same time, based on Fourier's law of heat conduction (the direction of heat flow is opposite to the direction of temperature gradient, and the heat flux density is related to the thermal conductivity of the material and the temperature gradient), the initial heat flux density between the fire source node and the adjacent nodes is calculated; then, by numerically solving the fire dynamics control equations including the energy conservation equation (describing the dynamic balance law of heat input, output and storage) and the momentum conservation equation (describing the fluid dynamics law of the change of momentum of heat flow), the control equations are time discretized using the finite difference method or the finite element method, setting the time step of the initial stage to 0.1 second and the time step of the development stage to 0.01 second. The temperature field, heat flux density field and smoke concentration field distribution of each node are iteratively calculated within each time step; finally, the heat flux density vector (including amplitude and direction information) of each spatial unit node at the current moment is extracted, and the heat flux transfer direction between nodes is used as the row index and the receiving node is used as the column index to construct the heat flux density field matrix (the matrix element represents the heat flux density value from the source unit to the target unit at the corresponding moment, in watts per square meter).

[0088] In other implementations, reduced-order modeling technology (such as orthogonal decomposition) may also be introduced to compress the high-dimensional thermal flow field to reduce computational complexity, which is not limited in this application.

[0089] It should be noted that the heat flux density field matrix described in this application refers to a multidimensional matrix generated based on the fire risk evolution model, which characterizes the intensity and direction of heat flux transfer between spatial units within the risk subdomain; the matrix uses the heat flux transfer path between spatial units as the row index and the receiving unit as the column index, and the matrix elements are heat flux density values ​​(unit: W / m²). The numerical value reflects the strength of the heat flux transfer, and the positive or negative sign or vector direction characterizes the direction of heat flux diffusion (such as from the high temperature unit to the low temperature unit); in addition, the fire risk evolution model refers to a mathematical calculation model constructed based on the fire dynamics theory, which is used to simulate the occurrence, development and spread of fire in spatial units within the risk subdomain.

[0090] In step 104, the risk fitting degrees of different spatial units are determined by combining the heat flux density field matrix with the monitoring information of the fire situation in the target building, and then the fire risk in the target building is temporally aggregated based on all risk fitting degrees combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential energy of different spatial units.

[0091] In some embodiments, reference Figure 3 As shown in FIG, this figure is a schematic diagram of a process for determining risk fitness shown in some embodiments of the present application. The risk fitness of different spatial units can be determined by combining the heat flux density field matrix with the monitoring information of the fire situation in the target building by the following steps:

[0092] First, in 1041, monitoring information of the fire situation in the target building is obtained;

[0093] Then, in 1042 , a plurality of information differences between the monitoring information and the heat flux density field matrix are determined;

[0094] Finally, in 1043, the risk fit of different spatial units is determined through all information differences.

[0095] As a preferred embodiment, a multi-type IoT sensor network can be deployed to cover all spatial units of the target building, including but not limited to temperature sensors (such as thermocouples, infrared temperature detectors), smoke concentration sensors (such as light scattering smoke detectors), gas composition sensors (such as CO, CO2 concentration detectors) and flame detectors (such as ultraviolet / infrared dual-band flame sensors) to collect physical environment data in real time; then, the sensor data is transmitted to the edge computing node or cloud server through wireless communication protocol for data cleaning and preprocessing (such as removing outliers and filling missing values); then, a mapping relationship between monitoring information and spatial units is established, and the collected data is matched to the corresponding spatial unit through positioning technology or sensor deployment address tags; finally, a time-series monitoring data set is generated at a preset frequency (such as seconds or minutes) as monitoring information of the fire in the target building.

[0096] In specific implementation, multiple information difference degrees between the monitoring information and the heat flux density field matrix are determined, that is, the inconsistency between the monitoring information and the heat flux density field matrix is ​​used as the information difference degree; specifically, the following steps are performed: first, the real-time parameters (such as heat conduction rate, conduction direction, and heat transfer delay time) representing the heat transfer between spatial units in the monitoring information are aligned with the corresponding unit pairs in the heat flux density field matrix and the model prediction values ​​at the same time to form a parameter pair set; then, for each parameter pair, a distance measurement algorithm (Euclidean distance algorithm) is used to calculate the information difference degree, for example, the Euclidean distance is used to calculate the absolute deviation between the measured value and the simulated value for the heat conduction rate parameter, and the angle difference formula is used to calculate the degree of direction deviation for the conduction direction parameter; then, the time dimension difference evaluation is introduced, and the dynamic time warping algorithm is used to calculate the information difference degree. The temporal matching degree between the monitoring information sequence and the heat flux density field matrix sequence is quantified to quantify the time axis offset error caused by the difference in the change of the heat conduction rate (such as the temporal misalignment between the measured heat conduction delay and the model prediction value); finally, a multi-dimensional difference vector is constructed, and the difference of each parameter (difference in heat conduction rate, difference in conduction direction, difference in delay time) is normalized with the time difference and then weighted and fused to form a comprehensive difference index as the information difference in this application; among them, as a preferred embodiment, the weight coefficient of each parameter difference can be objectively calculated by the entropy weight method (such as giving a higher weight to the difference in heat conduction rate to reflect the energy transfer efficiency); in other embodiments, a fuzzy logic algorithm can be used to qualitatively evaluate the difference of nonlinear parameters (such as the difference in the modulation effect of flue gas concentration on heat conduction), which is not limited in this application.

[0097] In specific implementation, the risk fit of different spatial units can be determined by all information differences in the following way, namely: first, based on historical fire case data or laboratory simulation samples, a mathematical relationship model is trained using machine learning algorithms (such as linear regression, random forest, neural network), and multidimensional information such as temperature difference, smoke concentration difference, gas composition difference and time series difference of each spatial unit is used as input variables, and the initial value of risk fit (with a value range of [0, 1]) is output through model parameter learning; for example, when using a neural network model, the number of neurons in the input layer can be set to be consistent with the difference dimension, the hidden layer captures the nonlinear relationship through a linear rectification function, and the output layer normalizes the result into a probability value through a sigmoide function; then, according to the characteristics of the fire evolution stage (such as the temperature rise rate is dominant in the initial stage and the smoke diffusion is dominant in the development stage), the fire risk is determined by the neural network model. The core and heyday stages use gas products as key indicators) and the weight vectors of various difference parameters are automatically adjusted through the fuzzy logic controller; for example, in the early stage of a fire (0-10 minutes), the temperature difference weight is set to 0.5, the smoke concentration weight is set to 0.3, and the gas composition weight is set to 0.2; after entering the development stage (10-30 minutes), it is automatically adjusted to 0.3 for temperature, 0.5 for smoke, and 0.2 for gas, so as to adapt to the changes in the risk-dominant factors in different fire development stages; finally, the risk fit is divided into four levels: low risk (<0.3), medium risk (0.3-0.6), high risk (0.6-0.8), and extremely high risk (>0.8), and a risk fit matrix with risk fit level labels is generated based on the grid coordinate system, where each element in the risk fit matrix corresponds to a spatial unit, including coordinates and corresponding risk fit.

[0098] It should be noted that the risk fitting degree mentioned in this application refers to the spatial unit fire risk credibility index obtained by quantifying the multi-dimensional difference of the fusion monitoring information and the heat flow diffusion information; the risk fitting degree is in numerical form (the value range is [0, 1]) and can be used to characterize the degree of consistency between the model prediction results and the actual fire situation. The higher the value, the higher the credibility of the fire risk prediction.

[0099] In some embodiments, the fire risk in the target building is aggregated in time series based on all risk fitting degrees combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential of different spatial units. This can be achieved by the following steps:

[0100] Obtaining the local temperature field of each spatial unit in the three-dimensional digital twin model;

[0101] By combining all risk fitting degrees with the local temperature field of each spatial unit for time series analysis, the time series risk aggregation model of the target building is obtained;

[0102] Determining the fire risk value of each spatial unit at different time points according to the temporal risk aggregation model;

[0103] The fire risk spread potential of different spatial units is determined through all fire risk values.

[0104] As a preferred embodiment, temperature data of each spatial unit can be collected in real time by temperature sensors (such as thermocouples and infrared temperature detectors) deployed in the target building, and the data can be matched with the corresponding grid units in the three-dimensional digital twin model using a positioning system; then, for areas where sensors are not deployed, a spatial interpolation algorithm (such as inverse distance weighted interpolation and Kriging interpolation) is used to estimate the missing data based on the measured temperature values ​​of neighboring sensors; then, the collected and estimated discrete temperature data are mapped to the grid nodes of the three-dimensional digital twin model to form the temperature distribution of each spatial unit at a certain moment; finally, the temperature distribution data at each moment is stored in a time series, and a continuous local temperature field change sequence is constructed as the local temperature field of each spatial unit in the three-dimensional digital twin model. In other embodiments, other methods can also be used to obtain it, which is not limited here.

[0105] In specific implementation, all risk fitting degrees are combined with the local temperature field of each spatial unit for time series analysis to obtain the time series risk aggregation model of the target building. This can be achieved in the following way: first, the risk fitting degree of each spatial unit at different time points is spliced ​​with the parameters (such as temperature value, temperature rise rate) in the local temperature field at the corresponding time point to obtain a multi-dimensional time series feature vector containing the time dimension; then, the time series modeling algorithm is used to analyze the feature vector, for example, the memory unit of the long short-term memory network is used to capture the long-term dependence relationship between the risk fitting degree and the local temperature field, or the feature vectors of different time scales are extracted through the dilated causal convolution of the time series convolutional network. feature; then, an attention mechanism is introduced to dynamically assign weights of different time steps and feature dimensions, and higher weights are given to the characteristic parameters of the key stages of the fire (such as the initial temperature rise and the smoke diffusion in the development stage); finally, the network parameters (such as the gating threshold of the long short-term memory network and the convolution kernel size of the temporal convolution network) are optimized through model training, so that the model output can characterize the coupled evolution characteristics of the risk fit and the temperature field, and obtain the temporal risk aggregation model of the target building; wherein, as a preferred embodiment, a sliding time window can be used to frame the time series data; in other implementations, the model output can be corrected a posteriori in combination with expert rules, which is not limited in this application.

[0106] In specific implementation, determining the fire risk value of each spatial unit at different time points according to the temporal risk aggregation model can be achieved in the following manner: first, the multi-dimensional temporal feature vector of each spatial unit (including a risk fit sequence, a temperature value sequence, a temperature rise rate sequence, etc.) is input into the trained temporal risk aggregation model, and the risk aggregation feature value of the unit at each time point is calculated through the temporal feature extraction layer of the model (such as the hidden layer of the long short-term memory network, the convolutional layer of the temporal convolutional network); then, the risk aggregation feature value is mapped to a specific fire risk value using a fully connected layer or a regression layer, thereby obtaining the fire risk value of each spatial unit at different time points; preferably, mapping the risk aggregation feature value to a specific fire risk value using a fully connected layer or a regression layer specifically includes: performing a matrix multiplication operation on the neuron weight matrix of the fully connected layer and the risk aggregation feature value, superimposing a bias, and outputting a one-dimensional scalar as the corresponding fire risk value through an activation function (such as a linear function). In other embodiments, other methods can also be used for implementation, which are not limited here.

[0107] In specific implementation, the fire risk propagation potential of different spatial units can be determined by all fire risk values ​​in the following way, namely: first, establish a spatiotemporal mapping relationship between the fire risk value of each spatial unit at different time points and the grid nodes of the three-dimensional digital twin model to form a weighted spatiotemporal data point set, then use the spatiotemporal interpolation algorithm in the existing technology to smooth the discrete data points to generate a continuous risk value field, and then extract the structural parameters (such as wall thermal resistance, ventilation volume, space volume), material parameters (such as combustible calorific value, etc.) of each spatial unit from the three-dimensional digital twin model. The risk value field data and the unit physical properties are input into the physical field coupling model (such as the multi-field coupling algorithm based on the heat conduction equation and the energy conservation equation) or the machine learning regression model (such as a trained neural network) to calculate the heat flow conduction rate, heat release rate gradient, and smoke diffusion potential difference of each spatial unit. Finally, according to the fire development stage (such as the initial stage and the development stage), the weight is dynamically allocated (such as the initial stage focuses on the heat conduction rate weight, and the development stage focuses on the smoke diffusion potential difference weight), and the weighted fusion comprehensive index value is used as the fire risk propagation potential energy of the corresponding spatial unit.

[0108] It should be noted that the fire risk value described in this application refers to a quantitative indicator that characterizes the degree of fire risk of a spatial unit at a specific time point, calculated by a time-series risk aggregation model; the fire risk value maps the risk aggregation characteristic value to a scalar value in the interval [0, 100] through a fully connected layer or a regression layer, and the larger the fire risk value, the higher the fire risk; in addition, the fire risk propagation potential refers to a fire risk propagation capability indicator quantified by fusing the fire risk value of the spatial unit with the physical property parameters; the physical field coupling model refers to a multi-field coupling calculation model constructed based on physical laws such as the heat conduction equation and the energy conservation equation; the model takes the risk value field data of the spatial unit (such as a continuous fire risk value distribution) and structural parameters (wall thermal resistance, ventilation volume), and material parameters (combustible calorific value, thermal conductivity) as input, and solves the coupling relationship of multiple physical fields such as heat conduction, heat release, and smoke diffusion through numerical calculation methods, and outputs physical quantities such as heat flow conduction rate and heat release rate gradient that characterize the fire propagation capability.

[0109] In step 105, the fire spread situation of the target building is monitored based on all fire risk propagation potentials, and a fire spread warning signal is output.

[0110] In some embodiments, monitoring the fire spread situation of a target building based on all fire risk propagation potentials and outputting a fire spread warning signal can be achieved by the following steps:

[0111] Analyze the changing trend of fire situation in target buildings based on all fire risk propagation potential;

[0112] Evaluate the changing trend of the fire situation according to preset fire warning rules;

[0113] When the evaluation result reaches the diffusion warning condition, a fire diffusion warning signal is generated and output.

[0114] In specific implementation, the following method can be used to analyze the changing trend of the fire situation in the target building based on all fire risk propagation potential energies, namely: first, extract the characteristics of all fire risk propagation potential energies to obtain parameters such as risk diffusion speed (displacement distance of the center of mass of the spatial unit per unit time), risk propagation direction (azimuth of the line connecting the center of mass of the spatial unit at adjacent moments), and risk intensity change rate (first-order derivative of the fire risk propagation potential energy of the spatial unit per unit time); then, construct a multidimensional feature vector of the above parameters according to the time series, and use time series prediction algorithms such as linear regression and exponential smoothing to perform trend fitting on the feature vector to obtain the future change curve of each parameter; this is used as the output of the changing trend of the fire situation in the target building.

[0115] In specific implementation, the following method is used to evaluate the changing trend of the fire situation according to the preset fire warning rules, namely: inputting the risk diffusion speed, risk propagation direction, risk intensity change rate and other parameters in the fire situation change trend into the rule evaluation module, and comparing them with the preset multi-level thresholds (such as the risk diffusion speed threshold ≥5m / min, the risk intensity change rate threshold ≥10 units / min); calculating the comprehensive evaluation score of each parameter by weighted sum algorithm, mapping the comprehensive evaluation score to the preset warning level interval (blue warning [0, 30), yellow warning [30, 60), orange warning [60, 80), red warning [80, 100]), and generating a structured evaluation report including warning level, triggering parameters, and confidence as the evaluation result; as a preferred embodiment, when the evaluation result meets the diffusion warning condition, generating and outputting a fire diffusion warning signal, namely: comparing the comprehensive evaluation score in the evaluation result with the preset warning threshold, triggering the warning signal generation process when the comprehensive evaluation score is greater than or equal to the warning threshold; generating a warning signal including warning level, timestamp, risk area location, etc. according to the warning level. The system generates JSON format warning information with indicators and development trend forecasts; and asynchronously pushes the warning information to the data interface of the fire command center, the three-dimensional digital twin system visualization module and the on-site sound and light alarm controller through the message queue, and records the audit log containing the warning trigger time, release channel, and recipient confirmation status; when the comprehensive evaluation score is greater than or equal to the warning threshold, no processing is performed; wherein, the warning threshold can be set according to the warning scenario, for example: when the warning scenario is a densely populated but low-risk area (such as schools, hospitals, shopping malls (no flammable and explosive materials)), a larger warning threshold is set to avoid false alarms causing panic or evacuation chaos; when the warning scenario is a flammable and explosive area (such as a chemical plant, gas station, hazardous materials warehouse), a smaller warning threshold is set to avoid chain reactions such as explosions.

[0116] In addition, in another aspect of the present application, in some embodiments, the present application provides a fire warning system based on three-dimensional modeling, referring to Figure 4 , which is a schematic diagram of the structure of a fire warning system based on three-dimensional modeling according to some embodiments of the present application. The fire warning system 200 based on three-dimensional modeling includes: an acquisition module 201, a processing module 202, and an execution module 203, which are described as follows:

[0117] Acquisition module 201, in this application, acquisition module 201 is mainly used to obtain a three-dimensional digital twin model of a target building, wherein the three-dimensional digital twin model includes multiple spatial units in the target building;

[0118] Processing module 202, in this application, is mainly used to determine the anomaly index of fire risk at different spatial units based on the spatial structural characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field, and then perform risk fitting on all spatial units through all anomaly indicators to obtain a risk cloud map of fire distribution in the target building;

[0119] In addition, the processing module 202 in the present application is further configured to extract a fire risk subdomain from the target building according to the risk cloud map, and then generate a heat flux density field matrix between different spatial units in the risk subdomain based on the fire risk evolution model;

[0120] In addition, the processing module 202 in the present application is further configured to determine the risk fit of different spatial units by combining the heat flux density field matrix with the monitoring information of the fire situation in the target building, and then perform time series aggregation on the fire risk in the target building based on all the risk fits combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential energy of different spatial units;

[0121] Execution module 203, in this application, execution module 203 is mainly used to monitor the fire spread situation of the target building based on all fire risk propagation potentials, and output a fire spread warning signal.

[0122] In addition, the present application also provides a computer device, which includes a memory and a processor, wherein the memory stores code, and the processor is configured to obtain the code and execute the above-mentioned fire warning method based on three-dimensional modeling.

[0123] In some embodiments, reference Figure 5 , which is an internal structure diagram of a computer device for implementing a fire warning method based on three-dimensional modeling according to some embodiments of the present application. The fire warning method based on three-dimensional modeling in the above embodiment can be Figure 5 The computer device 300 shown in FIG. 1 is implemented as shown in FIG. 1 , and the computer device 300 includes at least one processor 301 , a communication bus 302 , a memory 303 , and at least one communication interface 304 .

[0124] The processor 301 may be a general-purpose central processing unit (CPU), or an application-specific integrated circuit (ASIC) or one or more processors for controlling the execution of the fire warning method based on three-dimensional modeling in the present application.

[0125] The communication bus 302 is used to transmit information between the above components.

[0126] Memory 303 may be, but is not limited to, a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disk storage, an optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer. Memory 303 may exist independently and be connected to processor 301 via communication bus 302. Memory 303 may also be integrated with processor 301.

[0127] Memory 303 is used to store program code for implementing the present invention, and is controlled by processor 301 for execution. Processor 301 is used to execute the program code stored in memory 303. The program code may include one or more software modules. The fire warning method based on 3D modeling in the above embodiment can be implemented by processor 301 and one or more software modules in the program code stored in memory 303.

[0128] The communication interface 304 uses any device such as a transceiver to communicate with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area network (WLAN), etc.

[0129] In a specific implementation, as an example, a computer device may include multiple processors, each of which may be a single-CPU processor or a multi-CPU processor. A processor herein may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0130] The aforementioned computer device may be a general-purpose computer device or a dedicated computer device. In a specific implementation, the computer device may be a desktop computer, a portable computer, a network server, a personal digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. The embodiments of this application do not limit the type of computer device.

[0131] In addition, the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned fire warning method based on three-dimensional modeling.

[0132] In summary, in the fire warning method and system based on three-dimensional modeling disclosed in the embodiment of the present application, a three-dimensional digital twin model of a target building is obtained, and the three-dimensional digital twin model includes multiple spatial units in the target building; the abnormal indicators of fire risks in different spatial units are determined according to the spatial structure characteristics of the three-dimensional digital twin model combined with the state evolution pattern of the physical field, and then all spatial units are subjected to risk fitting through all abnormal indicators to obtain a risk cloud map of the fire distribution in the target building; the risk subdomain of the fire is extracted from the target building according to the risk cloud map, and the fire risk subdomain is obtained. Based on the fire risk evolution model, a heat flux density field matrix is ​​generated between different spatial units in the risk subdomain; the risk fitting degree of different spatial units is determined by combining the heat flux density field matrix with the monitoring information of the fire situation in the target building, and then the fire risk in the target building is temporally aggregated according to all risk fitting degrees combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential energy of different spatial units; the fire diffusion situation of the target building is monitored based on all fire risk propagation potential energy, and a fire diffusion warning signal is output; the diffusion trajectory of the fire heat flow can be accurately identified in the three-dimensional model.

[0133] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0134] Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if such changes and modifications fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such changes and modifications.

Claims

1. A fire warning method based on three-dimensional modeling, characterized in that: The steps include: Obtaining a three-dimensional digital twin model of a target building, wherein the three-dimensional digital twin model includes a plurality of spatial units in the target building; Based on the spatial structural characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field, the fire risk fluctuation indicators at different spatial units are determined. Then, risk fitting is performed on all spatial units using all fluctuation indicators to obtain a risk cloud map of the fire distribution in the target building. Extracting a fire risk subdomain from the target building according to the risk cloud map, and then generating a heat flux density field matrix between different spatial units in the risk subdomain based on a fire risk evolution model; The heat flux density field matrix is ​​combined with the monitoring information of the fire situation in the target building to determine the risk fit of different spatial units. Then, based on all the risk fits and the local temperature field of each spatial unit, the fire risk in the target building is time-series aggregated to obtain the fire risk propagation potential of different spatial units. The fire risk propagation potential refers to the fire risk propagation capacity index quantified by fusing the fire risk value of the spatial unit with the physical property parameters. Monitor the fire spread of target buildings based on all fire risk propagation potentials and output fire spread warning signals; The heat flux density field matrix between different spatial units in the risk subdomain is generated based on the fire risk evolution model and specifically includes: Obtaining status information of each spatial unit in the risk subdomain; Setting boundary conditions during fire evolution using the state information; generating a heat flux density field matrix between different spatial units in the risk subdomain according to the boundary conditions and a fire risk evolution model; Among them, the fire risk in the target building is aggregated in time series based on all risk fitting degrees combined with the local temperature field of each spatial unit, and the fire risk propagation potential of different spatial units is obtained, including: Obtaining the local temperature field of each spatial unit in the three-dimensional digital twin model; By combining all risk fitting degrees with the local temperature field of each spatial unit for time series analysis, the time series risk aggregation model of the target building is obtained; Determining the fire risk value of each spatial unit at different time points according to the temporal risk aggregation model; The fire risk spread potential of different spatial units is determined through all fire risk values.

2. The method according to claim 1, wherein The fire risk anomaly indicators at different spatial units are determined based on the spatial structural characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field, including: Obtaining spatial structural characteristics of each spatial unit in the three-dimensional digital twin model; Perform dynamic simulation of the target building based on the state evolution model of the physical field to obtain the evolution process data of the physical field in each spatial unit; The spatial structure characteristics are correlated with the evolution process data to determine the abnormal fire risk indicators at different spatial units.

3. The method according to claim 1, wherein By fitting all spatial units with all abnormal indicators, the risk cloud map of fire distribution in the target building is obtained, which specifically includes: Fit and calculate the risk level of each space unit in the target building based on all abnormal indicators; The fitting calculation results are spatially mapped to generate a risk cloud map of fire distribution in the target building.

4. The method according to claim 1, wherein The fire risk subdomains extracted from the target building according to the risk cloud map specifically include: Performing threshold determination on the risk level of each spatial unit in the risk cloud map; The set of spatial units that meet the threshold discrimination conditions is extracted as the fire risk subdomain.

5. The method according to claim 1, wherein The risk fitting degree of different spatial units is determined by combining the heat flux density field matrix with the monitoring information of the fire situation in the target building, specifically including: Obtain monitoring information on fire conditions in target buildings; determining a plurality of information differences between the monitoring information and the heat flux density field matrix; The risk fit of different spatial units is determined by all information differences.

6. The method according to claim 1, wherein Monitor the fire spread of target buildings based on all fire risk propagation potentials and output fire spread warning signals. Specifically, Analyze the changing trend of fire situation in target buildings based on all fire risk propagation potential; Evaluate the changing trend of the fire situation according to preset fire warning rules; When the evaluation result reaches the warning condition, a fire spread warning signal is generated and output.

7. The method according to claim 1, wherein The spatial unit is a finite element unit with uniform spatial properties.

8. A fire warning system based on three-dimensional modeling, which uses the method according to any one of claims 1 to 7 to perform fire warning, characterized in that: The system includes: An acquisition module, configured to acquire a three-dimensional digital twin model of a target building, wherein the three-dimensional digital twin model includes a plurality of spatial units in the target building; A processing module is used to determine the anomaly index of fire risk at different spatial units based on the spatial structural characteristics of the three-dimensional digital twin model and the state evolution pattern of the physical field, and then perform risk fitting on all spatial units through all the anomaly indexes to obtain a risk cloud map of fire distribution in the target building; The processing module is further configured to extract a fire risk subdomain from the target building according to the risk cloud map, and then generate a heat flux density field matrix between different spatial units in the risk subdomain based on the fire risk evolution model; The processing module is further configured to determine the risk fit of different spatial units by combining the heat flux density field matrix with the monitoring information of the fire situation in the target building, and then perform time series aggregation of the fire risk in the target building based on all the risk fits combined with the local temperature field of each spatial unit to obtain the fire risk propagation potential energy of the different spatial units; The execution module is used to monitor the fire spread situation of the target building based on all fire risk propagation potentials and output a fire spread warning signal.

Images