Fire spread simulation system for forest and town interface based on space-time semantic driving

Abstract

The application discloses a forest and town junction domain fire spread simulation system based on space-time semantic driving, comprising the following steps: integrating vegetation, terrain, meteorological, building and other data through a multi-source space-time semantic data acquisition module, extracting semantic features and carrying out correlation analysis through a forest fire town spread evaluation calculation module, and then generating fire environment parameters through a fire environment semantic coupling response module; a self-adaptive grid division processing module dynamically adjusts the grid according to the parameters, a dual-domain fire state semantic monitoring module realizes dual-domain fire state cross-domain monitoring, and finally a junction domain fire spread simulation output module completes dynamic simulation. The system improves the comprehensive utilization efficiency of multi-source data, the overallness of evaluation results, the adaptability of the grid to the fire environment and the correlation of fire state monitoring, and improves the simulation timeliness and accuracy as a whole, thereby providing precise technical support for fire prevention and control decision-making.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation management technology, and in particular to a forest-township boundary fire spread simulation system based on spatiotemporal semantics. Background Technology

[0002] The boundary between forests and towns, as a transitional zone between ecosystems and human settlements, features complex vegetation types, dense building distribution, and diverse terrain. When fires occur, they are prone to cross-regional spread, posing a serious threat to the lives and property of residents and the ecological environment. Currently, fire spread simulation in boundary areas needs to integrate multi-source spatiotemporal data on vegetation, topography, meteorology, and buildings to accurately capture dynamic changes in the fire environment and fire propagation patterns, supporting fire prevention and control decision-making. However, traditional simulation methods struggle to achieve deep correlation and semantic analysis of multi-source data and lack targeted monitoring of the differences in fire conditions between forest and urban areas. This results in simulation results that fail to meet the timeliness and accuracy requirements of actual fire prevention and control. Therefore, there is an urgent need to construct a fire spread simulation system that can integrate multiple model algorithms and achieve collaborative monitoring of both forest and urban areas.

[0003] Existing technologies for simulating fire spread in forest-urban boundary areas have two significant drawbacks: First, the data processing and evaluation stages lack the ability to semantically couple multi-source spatiotemporal data, often only performing independent analysis of single data points. This fails to establish deep correlations between vegetation distribution, topographic features, meteorological conditions, and urban building attributes, resulting in evaluation results that cannot comprehensively reflect the combined impact of various factors on fire spread, thus affecting the basic accuracy of subsequent simulations. Second, there is a disconnect between fire propagation simulation and dynamic monitoring. Grid division often adopts a fixed pattern, failing to adapt to real-time changes in the fire environment. Furthermore, the monitoring of fire state parameters in forest and urban areas lacks a cross-domain correlation mechanism, making it difficult to accurately capture changes in fire rate and intensity attenuation during cross-domain fire propagation. This leads to significant deviations between simulation results and actual fire spread processes, failing to provide precise technical support for the formulation of prevention and control measures. Summary of the Invention

[0004] To overcome the shortcomings and deficiencies of existing technologies, this invention provides a forest-town boundary fire spread simulation system based on spatiotemporal semantics.

[0005] A forest-urban boundary fire spread simulation system based on spatiotemporal semantics includes: a multi-source spatiotemporal semantic data acquisition module, a forest fire-urban spread assessment and calculation module, a fire environment semantic coupling response module, an adaptive grid partitioning and processing module, a dual-domain fire state semantic monitoring module, and a boundary fire spread simulation output module. The multi-source spatiotemporal semantic data acquisition module collects vegetation type distribution data, topographic elevation data, real-time meteorological monitoring data, urban building density data, road network distribution data, and historical fire case data within the forest-urban boundary area, and transmits the collected data to the forest fire-urban spread assessment and calculation module. The forest fire-urban spread assessment and calculation module performs semantic feature extraction and correlation analysis on the received data based on a multi-source spatiotemporal semantic forest fire-urban spread assessment model, and outputs the assessment results to the fire environment semantic coupling response module. The fire environment semantic coupling response module uses a semantic coupling fire environment response algorithm to couple the evaluation results with real-time fire environment parameters, generating fire environment response parameters that are transmitted to the adaptive grid partitioning processing module. The adaptive grid partitioning processing module uses an adaptive grid fire propagation model to dynamically partition the boundary domain space according to the fire environment response parameters, and outputs the grid partitioning results to the dual-domain fire state semantic monitoring module. The dual-domain fire state semantic monitoring module performs real-time semantic monitoring and data feedback on the fire state parameters of the forest domain and urban domain within the grid partitioning area, and transmits the feedback data and grid partitioning results to the boundary domain fire spread simulation output module. The boundary domain fire spread simulation output module combines the feedback data and grid partitioning results to dynamically simulate the fire spread process in the forest-urban boundary domain and output the results.

[0006] Furthermore, the expression for the multi-source spatiotemporal semantic forest fire town spread assessment model in the forest fire town spread assessment calculation module is as follows: ;in, For the assessment value of forest fire spread to towns, For the first Weighting coefficients for multi-source data For the first Semantic feature extraction values ​​of class data, For the first Multi-source spatiotemporal data, For the number of associated parameter types, For the first Class data and the first The correlation coefficients of parameters affecting fire spread For the first Parameters affecting fire spread The time decay coefficient, For the first Data collection time for class data The current simulation time is used; the expression for the semantic coupling response module of the fire environment to couple the evaluation results with real-time fire environment parameters using the semantic coupling fire environment response algorithm is as follows: ;in, The coupled response value of the fire environment, Couple ( ) is a coupled operation function. For real-time wind speed parameters, For air density parameters, For ambient temperature parameters, This is a relative humidity parameter. The number of fire environmental factors. For semantic mapping functions, For the first Individual fire behavior parameters For the first Environmental parameters.

[0007] Furthermore, the semantically coupled fire environment response algorithm expression in the fire environment semantic coupling response module is as follows: ;in, For the semantic coupling coefficient of the fire environment, To evaluate the semantic transformation value of the result, For gradient operators, For time variables, This is the coupling adjustment coefficient. The number of environmental monitoring points, For the first Temperature parameters at each monitoring point For the first The vegetation moisture content parameter at each monitoring point; the expression for the adaptive grid partitioning processing module to dynamically partition the boundary space based on the fire environment response parameters using the adaptive grid fire propagation model is as follows: ;in, To adjust the coefficients for adaptive meshing, For the initial grid coordinate function, They are x and y in a spatial rectangular coordinate system, respectively. Axis coordinates This represents the maximum value of the fire environment coupled response. For grid adjustment coefficients, Let be the partial derivative of the fire environment coupled response value in spatial coordinates. For the number of grid cells, For the first The area and distance weighting function of each grid cell. For the first The area of ​​each grid cell, For the first The distance from each grid cell to the fire point.

[0008] Furthermore, the adaptive grid fire propagation model expression in the adaptive grid partitioning processing module is as follows: ;in, This represents the probability value of fire propagation. For divergence operators, This is the semantic quantization value of the fire environment coupled response value. For the fire propagation velocity vector, For the propagation regulation coefficient, Divide the grid area into several parts. For the first Fuel load function for each grid region For the first Fuel type parameters for each grid region For the first Wind speed influence function for each grid region For the first The wind speed parameters for each grid region; the expression for semantic feature extraction and correlation analysis of the received data by the forest fire town spread assessment calculation module based on the multi-source spatiotemporal semantic forest fire town spread assessment model is as follows:

[0009] ,

[0010] in, Values ​​for analyzing the spread of forest fires to towns. For semantic feature extraction function, This is data on the distribution of vegetation types. For terrain elevation data, For real-time meteorological monitoring data, For correlation analysis functions, For urban building density data, For traffic network distribution data, This is data on historical fire cases. The historical data influence coefficient. This represents the maximum value of historical fire case data.

[0011] Furthermore, in the process of real-time semantic monitoring and data feedback of fire state parameters in the forest domain and urban domain within the grid-divided area, the dual-domain fire state semantic monitoring module uses the following fire state semantic monitoring model expression: ;in, This is the semantic monitoring value for fire state. For semantic monitoring functions, For flame temperature parameters, For smoke concentration parameters, For thermal radiation intensity parameters, The area represents the grid area after adaptive grid adjustment, and TotalArea represents the total monitored area of ​​the boundary region. To monitor the number of sub-regions, For the first Domain type weighting function for each sub-region For the first The forest area ratio of each sub-region For the first The proportion of urban areas in each sub-region; the expression for the dynamic simulation of the fire spread process in the forest-urban boundary area by the boundary area fire spread simulation output module, which combines feedback data and grid division results, is as follows: ,in, This represents the result of a fire spread simulation. For dynamic simulation functions, The time step function, To simulate the time step, To simulate the number of parameter types, For parameter weighting function, For the first Each forest domain simulation parameter, For the first Each town area simulation parameter.

[0012] Furthermore, when the forest fire town spread assessment calculation module performs semantic feature extraction and association analysis on the received data based on the multi-source spatiotemporal semantic forest fire town spread assessment model, the semantic feature association model expression used is as follows: ,in, For semantic feature relevance, For the first The semantic value of the fire spread effect parameter. For the first Class data and the first The semantic distance of each parameter. The distance attenuation coefficient is used; when the adaptive mesh generation processing module dynamically meshes the boundary domain space according to the fire environment response parameters using the adaptive mesh fire propagation model, the mesh density adjustment model expression used is: ;in, This represents the grid density value. This represents the minimum value of the fire environment coupled response. Vegetation is the terrain influence function. This is the vegetation influence function.

[0013] Further, the adaptive grid partitioning processing module includes a grid initial generation unit, a fire environment parameter mapping unit, a dynamic grid adjustment unit, and a grid quality verification unit. The grid initial generation unit, based on the spatial coordinate range of the forest-town boundary, topographic elevation data, and urban building distribution boundaries, performs initial grid partitioning of the boundary space according to a preset grid precision, generating an initial grid dataset including grid number, grid coordinates, vegetation type proportion within the grid, and building density within the grid. The fire environment parameter mapping unit receives fire environment response parameters transmitted by the fire environment semantic coupling response module, associates and maps the fire environment response parameters with the initial grid dataset according to the grid number, calculates the mean, maximum, and rate of change of the fire environment response parameters corresponding to each initial grid unit, and generates a grid-fire environment parameter mapping table. The dynamic grid adjustment unit is based on an adaptive grid fire propagation model. According to the rate of change of fire environment response parameters in the grid-fire environment parameter mapping table, it densifies grid cells with a rate of change of fire environment response parameters greater than a preset threshold and merges grid cells with a rate of change of fire environment response parameters less than the preset threshold. At the same time, it adjusts the grid division angle in combination with terrain slope parameters to ensure the fit between the grid boundary and the terrain contour lines. The grid quality verification unit performs quality verification on the dynamically adjusted grid cells, calculates the aspect ratio, interior angle, and area deviation of each grid cell, and removes grid cells with an aspect ratio exceeding the preset range, interior angle exceeding the reasonable range, or area deviation exceeding the allowable threshold. The removed areas are then re-grid until all grid cells meet the quality requirements, and the final grid division result is output.

[0014] Furthermore, the dual-domain fire status semantic monitoring module includes a forest domain fire status monitoring unit, an urban domain fire status monitoring unit, a cross-domain fire status association unit, and a fire status data feedback unit. The forest domain fire status monitoring unit selects grid cells with a forest domain proportion greater than a preset ratio as monitoring points within the grid division area output by the adaptive grid division processing module. It collects real-time parameters such as flame temperature, smoke concentration, thermal radiation intensity, and vegetation burning rate at each monitoring point, and converts the collected parameters into standardized semantic data through semantic encoding. The urban domain fire status monitoring unit selects grid cells with an urban domain proportion greater than a preset ratio as monitoring points within the grid division area, and collects real-time parameters such as building surface temperature and building combustion rate at each monitoring point. The parameters of fire status, quantity of flammable materials, and road network traffic status are converted into standardized semantic data using the same semantic coding rules as the forest fire monitoring unit. The cross-domain fire correlation unit performs correlation analysis on the standardized semantic data of monitoring points in the forest and urban areas, identifies the transmission relationship of fire parameters between adjacent forest and urban monitoring points, calculates the time delay and intensity attenuation coefficient of cross-domain fire parameter propagation, and generates a cross-domain fire correlation matrix. The fire data feedback unit integrates the forest fire data, urban fire data, and cross-domain fire correlation matrix according to a preset data format, adds data collection timestamps and grid number information, and transmits them in real time to the boundary fire spread simulation output module.

[0015] Furthermore, the boundary domain fire spread simulation output module includes a simulation parameter integration unit, a dynamic spread calculation unit, a simulation result rendering unit, and a simulation data output unit. The simulation parameter integration unit receives feedback data transmitted from the dual-domain fire state semantic monitoring module and the grid partitioning results output by the adaptive grid partitioning processing module. It matches and correlates the fire state parameters in the feedback data with the grid attributes in the grid partitioning results, extracting the real-time fire state parameters, initial environmental parameters, and spatial attribute parameters of each grid cell to construct a simulation calculation parameter library. The dynamic spread calculation unit, based on the multi-source spatiotemporal semantic forest fire town spread assessment model and the adaptive grid fire propagation model, uses data from the simulation calculation parameter library as input. The fire spread direction, spread speed, and combustion state changes of each grid cell at each time node are calculated according to a preset time step to generate a dynamic spread calculation dataset. The simulation result rendering unit uses 3D rendering technology to visualize the fire spread state at each time node based on the dynamic spread calculation dataset, presenting the dynamic effects of flame shape, smoke diffusion, vegetation burning, and building damage, while also annotating the fire state parameter values ​​of each grid cell. The simulation data output unit encapsulates the dynamic spread calculation dataset and the visualization rendering results according to a preset format to generate a simulation result package including time series data, spatial distribution data, and visualization files, supporting external devices to read and display the simulation result package.

[0016] Beneficial Effects: This invention proposes a forest-town boundary fire spread simulation system based on spatiotemporal semantics. It integrates multiple types of data through a multi-source spatiotemporal semantic data acquisition module, and combines this with a forest fire-town spread assessment and calculation module to achieve data semantic feature extraction and correlation analysis. This effectively overcomes the shortcomings of traditional technologies that lack multi-source data semantic coupling capabilities, significantly improving the comprehensive utilization efficiency of multi-source spatiotemporal data and the comprehensiveness of assessment results. It ensures that the assessment results fully reflect the comprehensive impact of vegetation, topography, meteorology, buildings, and other factors on fire spread, laying an accurate foundation for subsequent simulations. Simultaneously, the fire environment semantic coupling response module and the adaptive mesh generation processing module work collaboratively, based on the fire... The grid division is dynamically adjusted in real time to reflect environmental changes. The dual-domain fire state semantic monitoring module establishes a cross-domain correlation monitoring mechanism for fire state parameters in forest and urban domains. This solves the problems of fixed grids and disconnected dual-domain monitoring in traditional simulations, significantly improving the dynamic adaptability of fire propagation simulation and the correlation of fire state parameter monitoring. It accurately captures the rate changes and intensity attenuation of fire propagation across domains. Finally, combined with the boundary domain fire spread simulation output module, dynamic simulation and result output are realized, which improves the timeliness and accuracy of fire spread simulation in forest and urban boundary domains. The simulation results are more in line with the actual fire spread process, providing more accurate technical support for fire prevention and control decision-making and effectively meeting actual prevention and control needs. Attached Figure Description

[0017] Figure 1 This is a diagram showing the system module composition of the present invention;

[0018] Figure 2 This is a flowchart of the system operation steps of the present invention. Detailed Implementation

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1 As shown, the forest-town boundary fire spread simulation system based on spatiotemporal semantics includes: a multi-source spatiotemporal semantic data acquisition module, a forest fire-town spread assessment and calculation module, a fire environment semantic coupling response module, an adaptive grid partitioning and processing module, a dual-domain fire state semantic monitoring module, and a boundary fire spread simulation output module.

[0021] The multi-source spatiotemporal semantic data acquisition module collects vegetation type distribution data, topographic elevation data, real-time meteorological monitoring data, urban building density data, transportation network distribution data, and historical fire case data within the forest-urban boundary area, and transmits the collected data to the forest fire urban spread assessment and calculation module.

[0022] Specifically, the multi-source spatiotemporal semantic data acquisition module is the core component of the system's data input. During implementation, multiple types of sensors are deployed at 500-meter intervals within the forest-town boundary area, while simultaneously connecting to the meteorological department's real-time monitoring data interface and the urban planning geographic information database. Specific acquisition parameters include vegetation type distribution data (including the coverage area percentage of trees, shrubs, and herbs, with an accuracy of 10m x 10m grid units), topographic elevation data (obtained through a combination of UAV aerial photography and GIS topographic analysis, with elevation measurement errors controlled within ±0.5 meters), real-time meteorological monitoring data (including wind speed, wind direction, air temperature, relative humidity, and precipitation, with a sampling frequency of once per minute), urban building density data (statistics of building footprint per hectare, distinguishing the proportion of building materials such as concrete and wood), road network distribution data (recording road network width, material, and distance from vegetated areas, with a data update cycle of once per month), and historical fire case data (collecting information on the location, ignition time, duration of combustion, burned area, and direction of fire spread in the boundary area over the past 30 years). During implementation, the collected raw data needs to be uniformly converted into a standardized format, and spatiotemporal labels are stored in XML format. Data validation algorithms are used to remove outliers that exceed the normal range (such as abnormal data with wind speeds exceeding 30 m / s) and datasets with missing values ​​exceeding 5%, ensuring that the output standardized multi-source dataset has a completeness of over 98%. This provides an accurate and comprehensive data foundation for subsequent evaluation calculations. The stable operation of this module ensures that the system obtains data that truly reflects the environmental characteristics of the boundary region, avoiding the impact of data deviations on the accuracy of subsequent simulation results.

[0023] The forest fire town spread assessment and calculation module performs semantic feature extraction and correlation analysis on the received data based on the multi-source spatiotemporal semantic forest fire town spread assessment model, and outputs the assessment results to the fire environment semantic coupling response module.

[0024] Specifically, the forest fire urban spread assessment and calculation module receives the standardized dataset output by the multi-source spatiotemporal semantic data acquisition module. During implementation, semantic features are first extracted from various data types. Through semantic word segmentation and feature mapping algorithms, concrete data such as vegetation type, terrain slope, and wind speed are converted into quantifiable semantic feature vectors. The vegetation type feature vector is set to 8 dimensions (corresponding to 8 main vegetation combustion characteristics), and the terrain feature vector is set to 5 dimensions (including key indicators such as slope, aspect, and elevation difference). Subsequently, an association analysis algorithm is used to establish semantic relationships between different data types, calculating the weight of each data type's impact on fire spread. Meteorological data accounts for 35% of the weight (wind speed 15%, temperature 10%, humidity 10%), vegetation data 30%, terrain data 20%, and urban building and transportation data 15%. The weight calculation is determined through the analytic hierarchy process combined with historical fire data verification to ensure that the weight allocation is consistent with the actual degree of fire spread impact. During the assessment process, data needs to be analyzed in 10-minute time windows to generate forest fire urban spread assessment results for each time window. These results include the potential direction of fire spread, the estimated range of spread velocity (in meters per minute, with an accuracy controlled within ±0.2 meters per minute), and the distribution of high-risk areas (risk levels are marked with 10m × 10m grid units, categorized as low, medium, and high). This module transforms multi-source data into quantitative indicators with assessment significance. Through precise weight allocation and dynamic analysis, it improves the ability to predict fire spread trends, providing clear assessment criteria for fire-environment coupled response and avoiding the biased predictions caused by single-data dominance in traditional assessments.

[0025] The fire environment semantic coupling response module uses a semantic coupling fire environment response algorithm to couple the evaluation results with real-time fire environment parameters, and generates fire environment response parameters which are then transmitted to the adaptive mesh generation processing module.

[0026] Specifically, the fire environment semantic coupling response module is based on the assessment results output by the forest fire town spread assessment calculation module, and performs coupling calculations in combination with real-time supplemented fire environment parameters. During implementation, the parameters required for coupling calculations are first determined, including the fire risk level and potential spread speed in the assessment results, as well as the real-time collected wind speed (measurement range 0-25 m / s, accuracy ±0.1 m / s), air density (measurement range 1.1-1.3 kg / m³, accuracy ±0.01 kg / m³), ambient temperature (measurement range -10-40℃, accuracy ±0.5℃), relative humidity (measurement range 20%-100%, accuracy ±2%) and vegetation moisture content (measurement range 10%-30%, accuracy ±1%) around the fire point. The coupling operation employs a step-by-step processing approach. The first step semantically matches the semantic features in the assessment results with real-time fire environment parameters, establishing a correspondence between risk levels and parameters such as wind speed and temperature. For example, a high-risk level corresponds to an environmental condition with a wind speed greater than 5 m / s and a temperature higher than 25℃. The second step calculates the comprehensive impact coefficient of each parameter on the fire environment using a coupling algorithm. A time decay factor is introduced during coefficient calculation, giving higher weight to data from the last 10 minutes compared to earlier data. The decay coefficient is set to 0.9 (decaying by 10% every 10 minutes). The third step generates coupled fire environment response parameters, including a fire environment stability index (range 0-1, where 0 represents extremely unstable and 1 represents extremely stable), a fire spread acceleration probability (in percentage, with an accuracy of ±3%), and the degree of fire environment impact on buildings (divided into three levels: no impact, slight impact, and severe impact). This module enables deep integration of assessment results with the real-time fire environment. By dynamically adjusting parameter weights and accurately calculating the response index, it improves the ability to capture dynamic changes in the fire environment, overcoming the problem of assessment being disconnected from the fire environment in traditional technologies, and providing parameter support that closely matches the actual fire environment for subsequent grid division.

[0027] The adaptive grid partitioning processing module dynamically partitions the boundary domain space according to the fire environment response parameters using the adaptive grid fire propagation model, and outputs the grid partitioning results to the dual-domain fire state semantic monitoring module.

[0028] Specifically, the adaptive grid generation module dynamically divides the forest-urban boundary space into grids based on the fire environment coupling response parameters output by the fire environment semantic coupling response module. During implementation, the initial grid size is first determined, using the overall boundary area as the boundary. The initial grid size is set to 20 meters × 20 meters, with the coverage area extending 1 kilometer outward from the boundary edge (ensuring coverage of potential fire spread areas). Subsequently, the grid density is adjusted based on the fire environment stability index and fire spread acceleration probability in the fire environment coupling response parameters. When the stability index is less than 0.3 and the acceleration probability is greater than 60%, the corresponding area's grid is densified to 5 meters × 5 meters (the densified area boundary is extended outward by 3 initial grid units, using grid cells as units). When the stability index is greater than 0.7 and the acceleration probability is less than 30%, the corresponding area's grid is merged to 40 meters × 40 meters (the merged area must ensure it includes at least 4 initial grid units and does not cross the boundaries of densely built-up urban areas or densely vegetated areas). During grid division, the grid shape needs to be adjusted based on topographic elevation data. In areas with a slope greater than 25°, a rhomboid grid (with an acute angle of 45°) is used, ensuring that the grid boundary is at least 85% parallel to the topographic contour lines. Simultaneously, referring to urban building distribution data, in areas with a building density greater than 30%, the grid boundary must coincide with the edge of the building's exterior wall (deviation controlled within ±1 meter) to prevent the grid from crossing the main building structure. After division, grid units are labeled with attributes, including grid number (using the format "region code-row number-column number"), main vegetation type within the grid, building density, average fire environment response parameters, and grid priority (high-priority grids are used for key monitoring, accounting for approximately 30%). This module enables dynamic adaptation of grid division. By adjusting density and shape according to fire environment parameters, it improves the targeted coverage of high-risk fire areas, avoiding monitoring blind spots or resource waste caused by traditional fixed grids, and providing a precise spatial division basis for dual-domain fire monitoring.

[0029] The dual-domain fire state semantic monitoring module performs real-time semantic monitoring and data feedback on the fire state parameters of the forest domain and urban domain within the grid division area, and transmits the feedback data and grid division results to the boundary domain fire spread simulation output module in a coordinated manner.

[0030] Specifically, the dual-domain fire state semantic monitoring module uses the grid division results output by the adaptive grid division processing module to perform real-time semantic monitoring and data feedback on fire state parameters in the forest and urban domains. During implementation, monitoring priorities are first determined according to grid priority: high-priority grids collect data every 30 seconds, medium-priority grids every minute, and low-priority grids every 2 minutes. Monitoring parameters differentiate between the forest and urban domains. In the forest domain, the focus is on collecting flame temperature (measurement range 0-1200℃, accuracy ±5℃), smoke concentration (unit: mg / m³, measurement range 0-500, accuracy ±10 mg / m³), and vegetation burning rate (unit: cm / min, accuracy ±0.1 cm / min). In the urban domain, the focus is on collecting building surface temperature (measurement range 0-800℃, accuracy ±3℃), building combustion status (divided into unburned, partially burned, and completely burned categories), and remaining flammable material quantity (unit: kg, accuracy ±0.5 kg). Data acquisition employs a combination of sensors and video monitoring. Each high-priority grid is equipped with one temperature sensor and one smoke sensor, and every five high-priority grids are fitted with one high-definition video camera (1080P resolution, 25 frames per second) to supplement combustion status data through video image recognition. The acquired data is semantically encoded into standardized semantic information; for example, "flame temperature 800℃, smoke concentration 300 mg / m³" is encoded as the semantic tag "forest area - high risk - rapid combustion," and associated with the corresponding grid number and acquisition timestamp (time accuracy down to milliseconds). Data feedback uses a real-time transmission mode, with high-priority grid data latency controlled within 1 second and medium-to-low-priority grid data latency controlled within 3 seconds. Feedback data includes the original monitoring values, semantic tags, and data reliability (value range 0-1, determined by sensor error and data consistency verification). The implementation of this module enables accurate differentiation and monitoring of fire status in forest and urban areas. By using differentiated collection frequencies and multiple data verification methods, it improves the real-time performance and reliability of fire data, providing dynamically updated fire status support for fire spread simulation in border areas and overcoming the problem of data confusion between the two domains in traditional monitoring.

[0031] The boundary zone fire spread simulation output module combines feedback data and grid division results to dynamically simulate and output the fire spread process at the forest-town boundary.

[0032] Specifically, the boundary domain fire spread simulation output module integrates feedback data from the dual-domain fire state semantic monitoring module and grid division results from the adaptive grid division processing module to achieve dynamic simulation and result output of the fire spread process. During implementation, a simulation calculation model is first constructed. Based on grid cells, the fire state parameters (temperature, combustion rate, etc.) and spatial attributes (terrain, building distribution, etc.) of each grid are used as inputs. The simulation time step is 5 minutes. The fire spread between grids within each time step is calculated, including the spread direction (determined by the difference in fire state parameters between adjacent grids, with the direction of the largest difference being the main spread direction), the spread distance (calculated based on the combustion rate and time step, while correcting for the influence of terrain slope; for every 10° increase in slope, the spread distance increases by 5%), and the change in grid combustion state (the transition conditions from "unburning" to "partial burning" and then to "complete burning" are determined by the temperature threshold and the duration of combustion; for example, if the temperature exceeds 600℃ and lasts for 2 minutes, it is judged as complete burning). During the simulation, feedback data needs to be integrated and updated in real time to update simulation parameters. When the fire state parameters of a certain grid change exceed a preset threshold (e.g., temperature change greater than 100℃ / minute), the simulation calculation weights of that grid and its adjacent grids are immediately adjusted to ensure that the simulation results are synchronized with the actual fire state. The simulation results output is divided into two categories: data and visualization. The data output includes the fire-affected area at each time step (in square meters, with an accuracy of ±10 square meters), the number of affected grids, and a list of high-risk buildings and vegetation areas. The visualization output uses 3D rendering technology to restore the boundary terrain and buildings at a scale of 1:1000, marking the combustion state with different colors (red for complete combustion, orange for partial combustion, and yellow for unburned but high-risk areas), and dynamically displaying the fire spread process (the playback speed is adjustable, ranging from 1 to 10 times the real-time speed). The output results can be exported to common data formats (such as CSV and GIS vector formats) and video formats (4K resolution, 30 frames per second). It also provides a data query function, which can retrieve specific simulation data by time, grid number, and area location. The implementation of this module can transform multi-module data into intuitive and accurate simulation results. Through dynamic updates and multi-format output, it improves the practicality and readability of simulation results, provides clear and comprehensive technical references for fire prevention and control decisions, and realizes a complete closed loop from data collection to simulation application.

[0033] Preferably, the multi-source spatiotemporal semantic forest fire town spread assessment model expression in the forest fire town spread assessment calculation module is: ;in, For the assessment value of forest fire spread to towns, For the first Weighting coefficients for multi-source data For the first Semantic feature extraction values ​​of class data, For the first Multi-source spatiotemporal data, For the number of associated parameter types, For the first Class data and the first The correlation coefficients of parameters affecting fire spread For the first Parameters affecting fire spread The time decay coefficient, For the first Data collection time for class data The current simulation time is used; the expression for the semantic coupling response module of the fire environment to couple the evaluation results with real-time fire environment parameters using the semantic coupling fire environment response algorithm is as follows: ;in, The coupled response value of the fire environment, Couple ( ) is a coupled operation function. For real-time wind speed parameters, For air density parameters, For ambient temperature parameters, This is a relative humidity parameter. The number of fire environmental factors. For semantic mapping functions, For the first Individual fire behavior parameters For the first Environmental parameters.

[0034] Specifically, the core algorithms of the forest fire town spread assessment calculation module and the fire environment semantic coupling response module are implemented by first clarifying the parameter values ​​and calculation logic of the forest fire town spread assessment model. The weight coefficients of multi-source data are determined based on the contribution of each data point to the spread impact in historical fire cases. The total weight coefficients of different types of data such as vegetation, topography, and meteorology are 1. The time decay coefficient is set to 0.05 to ensure that the data collected closer to the current simulation time has a greater impact on the assessment results. The semantic feature extraction values ​​are obtained by standardizing the original data, and the value range is controlled between 0 and 1. The correlation coefficient is determined by calculating the ratio of the covariance to the standard deviation of the two types of data to ensure that it can accurately reflect the correlation strength between the data. In the implementation of the fire environment coupled response algorithm, the coupling operation function adopts a weighted summation method, integrating the evaluation value with real-time parameters such as wind speed and air density according to different weights. The weights are set as follows: wind speed 0.3, air density 0.2, ambient temperature 0.25, relative humidity 0.15, and vegetation moisture content 0.1. The semantic mapping function converts fire behavior parameters and environmental parameters into values ​​with a unified semantic dimension, also ranging from 0 to 1. The number of fire environment factors is determined based on the actual environmental characteristics of the boundary region, typically including eight key factors such as temperature, humidity, wind speed, and vegetation moisture content. By clarifying the model parameter values ​​and calculation process, the accuracy of the evaluation results and fire environment response values ​​is improved, ensuring a more realistic reflection of the comprehensive impact of multi-source data and fire environment parameters on fire spread, and providing more reliable parameter support for subsequent grid division.

[0035] Preferably, the semantically coupled fire environment response algorithm expression in the fire environment semantic coupling response module is: ;in, For the semantic coupling coefficient of the fire environment, To evaluate the semantic transformation value of the result, For gradient operators, For time variables, This is the coupling adjustment coefficient. The number of environmental monitoring points, For the first Temperature parameters at each monitoring point For the first The vegetation moisture content parameter at each monitoring point; the expression for the adaptive grid partitioning processing module to dynamically partition the boundary space based on the fire environment response parameters using the adaptive grid fire propagation model is as follows: ;in, To adjust the coefficients for adaptive meshing, For the initial grid coordinate function, They are x and y in a spatial rectangular coordinate system, respectively. Axis coordinates This represents the maximum value of the fire environment coupled response. For grid adjustment coefficients, Let be the partial derivative of the fire environment coupled response value in spatial coordinates. For the number of grid cells, For the first The area and distance weighting function of each grid cell. For the first The area of ​​each grid cell, For the first The distance from each grid cell to the fire point.

[0036] Specifically, the semantic coupling fire environment response algorithm of the fire environment semantic coupling response module and the adaptive grid fire propagation model of the adaptive grid partitioning processing module are implemented by first determining the parameter settings of the semantic coupling fire environment response algorithm. The coupling adjustment coefficient is calibrated based on the historical data of the fire environment stability of the boundary area and is set to 0.8. The number of environmental monitoring points is determined by a density of one per 2 square kilometers to ensure that there are no blind spots in the monitoring coverage. The temperature parameter is the average value collected in real time from the monitoring points, and the vegetation moisture content parameter is obtained by on-site sampling and measurement with an accuracy controlled within ±1%. The semantic conversion value converts the risk level of the assessment result into a quantitative value of 0-1. The gradient operator operation is realized by calculating the rate of change of the fire environment parameters in space to ensure that the spatial differences of the fire environment can be captured. In the implementation of the adaptive grid fire propagation model, the initial grid coordinate function is determined based on the latitude and longitude range of the boundary region. The x and y axes represent horizontal spatial coordinates, and the z axis represents elevation. The grid adjustment coefficient is set to 1.2 to ensure a reasonable adjustment range for grid density, avoiding excessive densification or merging. The area of ​​a grid cell is obtained by calculating the difference in grid boundary coordinates, and the distance is the straight-line distance from the center of the grid cell to the fire point. The weight function is set according to the influence of area and distance on fire propagation; the smaller the area and the closer the distance, the greater the weight of the grid cell. By refining the algorithm parameters and grid partitioning logic, the accuracy of the fire environment coupling coefficient calculation and the rationality of grid adjustment are improved, ensuring that the grid partitioning can accurately match the dynamic changes of the fire environment and provide more suitable spatial units for dual-domain fire monitoring.

[0037] Preferably, the adaptive grid fire propagation model expression in the adaptive grid partitioning processing module is: ;in, This represents the probability value of fire propagation. For divergence operators, This is the semantic quantization value of the fire environment coupled response value. For the fire propagation velocity vector, For the propagation regulation coefficient, Divide the grid area into several parts. For the first Fuel load function for each grid region For the first Fuel type parameters for each grid region For the first Wind speed influence function for each grid region For the first The wind speed parameters for each grid region; the expression for semantic feature extraction and correlation analysis of the received data by the forest fire town spread assessment calculation module based on the multi-source spatiotemporal semantic forest fire town spread assessment model is as follows:

[0038] ,

[0039] in, Values ​​for analyzing the spread of forest fires to towns. For semantic feature extraction function, This is data on the distribution of vegetation types. For terrain elevation data, For real-time meteorological monitoring data, For correlation analysis functions, For urban building density data, For traffic network distribution data, This is data on historical fire cases. The historical data influence coefficient. This represents the maximum value of historical fire case data.

[0040] Specifically, the adaptive grid fire propagation model of the adaptive grid partitioning processing module and the evaluation model of the forest fire town spread assessment calculation module are implemented by first clarifying the parameter configuration of the adaptive grid fire propagation model. The propagation adjustment coefficient is determined based on the historical fire propagation speed data of the boundary area and is set to 0.9. The number of grid areas is calculated based on the initial grid size and the total area of ​​the boundary area. Usually, the boundary area is divided into 200-500 grid areas. The fuel load function is determined based on the vegetation type. The fuel load is set to 10 tons / hectare for tree areas, 5 tons / hectare for shrub areas, and 2 tons / hectare for herbaceous areas. The wind speed influence function is obtained by fitting experimental data on the influence of wind speed on flame propagation speed. The higher the wind speed, the larger the function value, and the value range is 0-2. In the implementation of the forest fire urban spread assessment model, semantic feature extraction functions process vegetation type, topographic elevation, and meteorological data separately. Vegetation type is converted into a 0-1 value according to the degree of combustion difficulty, topographic elevation is converted according to slope, and meteorological data is converted according to its favorable effect on fire spread. A correlation analysis function calculates the correlation strength between building density, transportation network, and historical fire data. The historical data influence coefficient is set to 0.3 to ensure that historical data can reasonably influence the assessment results. The maximum value of historical data is the highest recorded value of this type of data in the past 30 years of historical cases. By clarifying the model parameters and function logic, the accuracy of fire propagation probability calculation and spread analysis is improved, ensuring a more realistic reflection of the impact of factors such as fuel and wind speed on propagation and the correlation between multi-source data.

[0041] Preferably, the fire state semantic monitoring model expression used by the dual-domain fire state semantic monitoring module in the process of real-time semantic monitoring and data feedback of fire state parameters in the forest domain and urban domain within the grid division area is as follows: ;in, This is the semantic monitoring value for fire state. For semantic monitoring functions, For flame temperature parameters, For smoke concentration parameters, For thermal radiation intensity parameters, The area represents the grid area after adaptive grid adjustment, and TotalArea represents the total monitored area of ​​the boundary region. To monitor the number of sub-regions, For the first Domain type weighting function for each sub-region For the first The forest area ratio of each sub-region For the first The proportion of urban areas in each sub-region; the expression for the dynamic simulation of the fire spread process in the forest-urban boundary area by the boundary area fire spread simulation output module, which combines feedback data and grid division results, is as follows: ,in, This represents the result of a fire spread simulation. For dynamic simulation functions, The time step function, To simulate the time step, To simulate the number of parameter types, For parameter weighting function, For the first Each forest domain simulation parameter, For the first Each town area simulation parameter.

[0042] Specifically, the fire semantic monitoring model of the dual-domain fire semantic monitoring module and the dynamic simulation model of the boundary domain fire spread simulation output module are implemented by first determining the parameter settings of the fire semantic monitoring model. The semantic monitoring function converts flame temperature, smoke concentration, and thermal radiation intensity into semantic quantification values ​​of 0-1. The flame temperature is 1 when it exceeds 1000℃ and 0 when it is below 300℃; the smoke concentration is 1 when it exceeds 400 mg / m³ and 0 when it is below 50 mg / m³; and the thermal radiation intensity is 1 when it exceeds 5 kW / m² and 0 when it is below 0.5 kW / m². The grid area is calculated through the grid boundary coordinates. The total monitoring area is the actual control area of ​​the boundary domain. The number of monitoring sub-regions is determined by dividing each 1 square kilometer into one sub-region. The domain type weight function is set according to the proportion of forest domain and urban domain area within the sub-region. The higher the proportion, the greater the weight. In the implementation of the dynamic simulation model, the dynamic simulation function integrates fire monitoring values, grid division results, and fire environment response values, calculating simulation results with different weights. The time step is set to 5 minutes to ensure that the simulation can track fire changes in real time. The number of simulation parameters includes 10 key parameters such as temperature, combustion rate, and spread direction. The parameter weight function is set according to the degree of influence of the parameters on the simulation results. The simulation parameter weights for forest and urban areas are adjusted according to the differences in fire spread characteristics within the areas, with temperature parameters typically having a higher weight in forest areas than in urban areas. By refining the monitoring and simulation parameters, the accuracy of fire monitoring values ​​and simulation results is improved, ensuring that the differences in fire conditions between the two areas and the dynamic process of fire spread can be accurately captured.

[0043] Preferably, when the forest fire town spread assessment calculation module performs semantic feature extraction and association analysis on the received data based on the multi-source spatiotemporal semantic forest fire town spread assessment model, the semantic feature association model expression used is as follows: ,in, For semantic feature relevance, For the first The semantic value of the fire spread effect parameter. For the first Class data and the first The semantic distance of each parameter. The distance attenuation coefficient is used; when the adaptive mesh generation processing module dynamically meshes the boundary domain space according to the fire environment response parameters using the adaptive mesh fire propagation model, the mesh density adjustment model expression used is: ;in, This represents the grid density value. This represents the minimum value of the fire environment coupled response. Vegetation is the terrain influence function. This is the vegetation influence function.

[0044] Specifically, the semantic feature association model of the forest fire town spread assessment calculation module and the grid density adjustment model of the adaptive grid partitioning processing module are implemented by first clarifying the parameter configuration of the semantic feature association model. The semantic distance is obtained by calculating the Euclidean distance between the semantic feature vectors of the two types of data, and the distance decay coefficient is set to 0.1 to ensure that the closer the semantic distance, the greater the influence of the data analysis value on the correlation. The semantic value is obtained by standardizing and semantically encoding the data, and the value range is 0-1. The correlation coefficient is calculated using the Pearson correlation coefficient formula to ensure that it can accurately reflect the degree of linear correlation between the data. The number of data and parameter types is determined according to the actual data collection situation in the boundary area. Typically, the data types include 6 types such as vegetation, topography, and meteorology, and the parameter types include 8 types such as spread rate and burning intensity. In the implementation of the grid density adjustment model, the grid adjustment coefficient is set to 1.2. The minimum value of the fire environment coupling response is taken as the minimum recorded value of this parameter within the simulation period. The terrain influence function is set according to the slope: 1.5 for slopes exceeding 30° and 0.8 for slopes below 5°. The vegetation influence function is set according to the ease of vegetation combustion: 1.2 for tree areas, 1.0 for shrub areas, and 0.9 for herbaceous areas. By clarifying the parameters and function logic of the correlation model and the density adjustment model, the accuracy of semantic feature correlation calculation and the rationality of grid density adjustment are improved, ensuring a more accurate reflection of data correlation and the influence of terrain and vegetation on grid division.

[0045] Preferably, the adaptive grid generation processing module includes a grid initial generation unit, a fire environment parameter mapping unit, a dynamic grid adjustment unit, and a grid quality verification unit. The grid initial generation unit, based on the spatial coordinate range of the forest-town boundary, topographic elevation data, and urban building distribution boundaries, performs initial grid generation on the boundary space according to a preset grid precision, generating an initial grid dataset including grid number, grid coordinates, vegetation type proportion within the grid, and building density within the grid. The fire environment parameter mapping unit receives fire environment response parameters transmitted by the fire environment semantic coupling response module, associates and maps the fire environment response parameters with the initial grid dataset according to the grid number, calculates the mean, maximum, and rate of change of the fire environment response parameters corresponding to each initial grid unit, and generates a grid-fire environment parameter mapping table. The dynamic grid adjustment unit... The dynamic grid adjustment unit is based on an adaptive grid fire propagation model. According to the rate of change of fire environment response parameters in the grid-fire environment parameter mapping table, it densifies grid cells with a rate of change of fire environment response parameters greater than a preset threshold and merges grid cells with a rate of change of fire environment response parameters less than a preset threshold. At the same time, it adjusts the grid division angle in combination with terrain slope parameters to ensure the fit between the grid boundary and the terrain contour lines. The grid quality verification unit performs quality verification on the dynamically adjusted grid cells, calculates the aspect ratio, interior angle, and area deviation of each grid cell, and removes grid cells with an aspect ratio exceeding a preset range, an interior angle exceeding a reasonable range, or an area deviation exceeding an allowable threshold. The removed areas are then re-grid until all grid cells meet the quality requirements, and the final grid division result is output.

[0046] Specifically, the adaptive grid generation module includes four units. During implementation, the initial grid generation unit is started first. Based on the latitude and longitude boundary of the forest-town boundary area, the grid accuracy is preset to 10 meters × 10 meters. Combined with the topographic elevation data (elevation data resolution of 1 meter) and the town building distribution vector map (building boundary accuracy ±0.5 meters) imported from the GIS system, the boundary area space is divided into grids. The generated initial grid dataset must include a 12-digit grid number (the first 6 digits are the regional coordinate code, and the last 6 digits are the row and column numbers), the latitude and longitude coordinates of the four corners of the grid, the proportion of trees / shrubs / herbaceous vegetation in the grid (accuracy ±1%), and the building area per square kilometer (i.e., building density, calculation accuracy ±0.01). Subsequently, the fire environment parameter mapping unit receives fire environment response parameters (including fire environment stability index, fire spread acceleration probability, etc.) transmitted from the fire environment semantic coupling response module. It matches these parameters one-to-one according to grid number, and uses the arithmetic mean method to calculate the mean (sampling frequency 1 time / minute), maximum value, and rate of change at 5-minute intervals for each grid cell over 30 minutes (rate of change calculation accuracy ±0.001). This generates a mapping table including grid number and corresponding parameters, with the table update frequency matching the parameter acquisition frequency. Next, the dynamic grid adjustment unit, based on the adaptive grid fire propagation model, sets a threshold of 0.15 for the rate of change of fire environment response parameters. When the rate of change of a grid parameter exceeds the threshold, it is split into four 5m × 5m sub-grids; when it is below the threshold of 0.05, the four adjacent grids are merged into a large 20m × 20m grid. Simultaneously, combined with terrain slope data (slope calculation uses a 3×3 window analysis method), in areas with a slope > 25°, the long side of the grid is adjusted to be parallel to the contour lines to ensure a fit ≥ 90%. Finally, the mesh quality verification unit calculates the aspect ratio (allowable range 1:1 to 1:3), interior angle (allowable range 60° to 120°), and area deviation value (allowable deviation ±5%) for each mesh. Mesh that do not meet the requirements are marked as abnormal and re-divided until the proportion of abnormal meshes is <1%. The final output includes mesh division results including mesh attributes and quality verification results.

[0047] Preferably, the dual-domain fire status semantic monitoring module includes a forest domain fire status monitoring unit, an urban domain fire status monitoring unit, a cross-domain fire status association unit, and a fire status data feedback unit. The forest domain fire status monitoring unit selects grid cells with a forest domain proportion greater than a preset ratio as monitoring points within the grid division area output by the adaptive grid division processing module. It collects real-time parameters such as flame temperature, smoke concentration, thermal radiation intensity, and vegetation burning rate at each monitoring point, and converts the collected parameters into standardized semantic data through semantic encoding. The urban domain fire status monitoring unit selects grid cells with an urban domain proportion greater than a preset ratio within the grid division area as monitoring points, and collects real-time parameters such as building surface temperature and building burning status at each monitoring point. The parameters of fire status, quantity of flammable materials, and road network traffic status are converted into standardized semantic data using the same semantic coding rules as the forest fire monitoring unit. The cross-domain fire correlation unit performs correlation analysis on the standardized semantic data of monitoring points in the forest and urban areas, identifies the transmission relationship of fire parameters between adjacent forest and urban monitoring points, calculates the time delay and intensity attenuation coefficient of cross-domain fire parameter propagation, and generates a cross-domain fire correlation matrix. The fire data feedback unit integrates the forest fire data, urban fire data, and cross-domain fire correlation matrix according to a preset data format, adds data collection timestamps and grid number information, and transmits them in real time to the boundary fire spread simulation output module.

[0048] Specifically, the dual-domain fire status semantic monitoring module is implemented in four units. First, the forest domain fire status monitoring unit selects grids with a forest domain ratio greater than 70% from the adaptive grid division results as monitoring points. Each monitoring point is equipped with thermocouple temperature sensors (measurement range 0-1200℃, accuracy ±2℃), laser smoke concentration sensors (measurement range 0-500mg / m³, accuracy ±5mg / m³), and heat flow meters (measurement range 0-10kW / m², accuracy ±0.2kW / m²). Real-time acquisition of flame temperature, smoke concentration, thermal radiation intensity, and vegetation burning rate (calculated through vegetation sample weight loss rate, accuracy ±0.05cm / min) is used. The parameters are converted into standardized semantic data using preset semantic coding rules (e.g., temperature 800-1200℃ is coded as "high" and 500-800℃ is coded as "medium"). The data sampling frequency is 1 time / 30 seconds. Secondly, the urban fire monitoring unit selects grids with an urban area ratio greater than 70% as monitoring points, and deploys infrared thermometers (measurement range 0-800℃, accuracy ±1℃), video recognition equipment (resolution 1080P, frame rate 25 frames / second) and weight sensors (measurement range 0-100kg, accuracy ±0.1kg) to collect data on building surface temperature, building combustion status (determined by video recognition as unburned / partially burning / completely burning), quantity of flammable materials (such as the weight of furniture, wood, etc.), and road network traffic status (determined by video recognition as unobstructed / congested / blocked). The data is converted using the same semantic coding rules as the forest domain, and the sampling frequency is also 1 time / 30 seconds. Then, the cross-domain fire correlation unit performs correlation analysis on the dual-domain semantic data, using the Pearson correlation coefficient to calculate the correlation strength of parameters between monitoring points in adjacent forest and urban areas (an absolute correlation coefficient > 0.6 is considered a strong correlation). Time series analysis is used to calculate the time delay (accuracy ± 1 second) and intensity attenuation coefficient of cross-domain fire parameter propagation (attenuation coefficient calculation is based on a distance attenuation model), generating a cross-domain fire correlation matrix including monitoring point number, correlation strength, time delay, and attenuation coefficient. The matrix is ​​updated once per minute. Finally, the fire data feedback unit integrates the data in the format of "timestamp (accurate to milliseconds) - grid number - forest area parameter - urban area parameter - correlation matrix," and transmits it in real time to the boundary fire spread simulation output module via a 5G network, with data transmission latency controlled to < 1 second.

[0049] Preferably, the boundary fire spread simulation output module includes a simulation parameter integration unit, a dynamic spread calculation unit, a simulation result rendering unit, and a simulation data output unit. The simulation parameter integration unit receives feedback data transmitted from the dual-domain fire state semantic monitoring module and the grid division results output by the adaptive grid division processing module. It matches and correlates the fire state parameters in the feedback data with the grid attributes in the grid division results, extracting the real-time fire state parameters, initial environmental parameters, and spatial attribute parameters of each grid cell to construct a simulation calculation parameter library. The dynamic spread calculation unit, based on the multi-source spatiotemporal semantic forest fire town spread assessment model and the adaptive grid fire propagation model, uses data from the simulation calculation parameter library as input and processes the data according to... The simulation results rendering unit calculates the fire spread direction, spread speed, and combustion state changes of each grid cell at each time node according to a preset time step, generating a dynamic spread calculation dataset. Based on the dynamic spread calculation dataset, the simulation results rendering unit uses 3D rendering technology to visualize the fire spread state at each time node, presenting the dynamic effects of flame shape, smoke diffusion, vegetation burning, and building damage, while also annotating the fire state parameter values ​​of each grid cell. The simulation data output unit encapsulates the dynamic spread calculation dataset and the visualization rendering results according to a preset format, generating a simulation results package that includes time series data, spatial distribution data, and visualization files, supporting external devices to read and display the simulation results package.

[0050] Specifically, the four units of the boundary fire spread simulation output module operate as follows: The simulation parameter integration unit receives feedback data (including fire parameters, semantic codes, and correlation matrices) and adaptive grid division results (including grid coordinates, attributes, and quality verification information) from the dual-domain fire state semantic monitoring module. It matches the data by grid number using SQL database association technology, extracts real-time fire state parameters (such as temperature and smoke concentration), initial environmental parameters (such as vegetation type and building density), and spatial attribute parameters (such as slope and elevation) for each grid, and constructs a simulation calculation parameter library. The parameter library adopts a distributed storage architecture, supports read and write operations of more than 1,000 data entries per second, and the data update frequency is consistent with the feedback data transmission frequency (1 time / 30 seconds). The dynamic spread calculation unit is based on a multi-source spatiotemporal semantic forest fire town spread assessment model and an adaptive grid fire propagation model. The simulation time step is set to 5 minutes. Using parameter library data as input, the finite element analysis method is used to calculate the fire spread direction (determined by the gradient of fire state parameters of adjacent grids, with the direction of maximum gradient being the main spread direction), spread speed (calculated based on empirical formulas of fuel load and wind speed, with an accuracy of ±0.01 m / min) and changes in combustion state (determined based on temperature and continuous combustion time, such as temperature > 600℃ and continuous for 5 minutes being determined as complete combustion) of each grid at each time point. The dynamic spread calculation dataset is generated, which includes time point, grid number, spread parameters, and combustion state. The dataset is stored in JSON format, and each data point includes more than 20 key indicators. The simulation rendering unit calls the Unity3D engine to construct a 3D model of the boundary region based on the dynamic spread calculation dataset (the model accuracy is 1:500, including vegetation, buildings, and terrain details). It uses a particle system to simulate the flame morphology (flame height is positively correlated with temperature, accuracy ±0.1 meters) and smoke diffusion (smoke concentration is positively correlated with diffusion range). It uses texture mapping technology to present the dynamic effects of vegetation burning (color gradient from green to yellow to black) and building damage (displaying complete → damaged → collapsed effects according to the burning state). At the same time, the fire state parameter values ​​of each mesh are superimposed on the model (font size is adapted to mesh size, and display accuracy retains 1 decimal place). The rendering frame rate is controlled at ≥30 frames / second, and real-time scaling and rotation operations are supported. The simulation data output unit encapsulates the dynamic propagation calculation dataset (JSON format) and the visualization rendering results (MP4 format, 4K resolution, 30 frames / second) into a simulation result package according to user needs. It supports multiple output methods such as USB 3.0 and cloud storage, and also provides a data query interface that can retrieve specific data by time node, grid number, and parameter type. The query response time is less than 0.5 seconds, meeting the needs of data retrieval and result display in fire prevention and control decision-making.

[0051] The multi-source spatiotemporal semantic forest fire urban spread assessment model of this invention is a core model for integrating multiple types of data in the forest-urban boundary area and quantitatively assessing the fire spread trend. Essentially, it transforms scattered vegetation, topography, meteorological, and building data into quantitative indicators that can support fire spread assessment through semantic parsing and correlation analysis. The implementation of this model requires first acquiring vegetation type distribution (including the proportion of trees, shrubs, and herbs), topographic elevation (accuracy ±0.5 meters), real-time meteorological data (wind speed, temperature, etc., sampling frequency 1 time / minute), urban building density, road network, and historical fire case data through a multi-source spatiotemporal semantic data acquisition module. Then, semantic features are extracted from these data, converting concrete data into semantic feature vectors in the range of 0-1 (such as the ease of vegetation combustion and the influence of topographic slope). Subsequently, the weights of each data type are determined using the analytic hierarchy process (35% for meteorological data, 30% for vegetation data, etc.). The rationality of the weights is verified by combining historical fire data. Finally, the potential fire spread direction, speed range (accuracy ±0.2 meters / minute), and high-risk area distribution are calculated by sliding the calculation according to a time window (every 10 minutes). The purpose of this model is to provide an accurate assessment basis for subsequent fire environment coupled response, avoiding the one-sidedness of traditional single-data assessment; to improve the efficiency of multi-source data comprehensive utilization, making the assessment results more consistent with the actual influencing factors of fire spread in the boundary area, providing reliable data support for subsequent stages of the system, and helping to improve the overall simulation accuracy.

[0052] The semantically coupled fire environment response algorithm of this invention is a key algorithm for achieving deep integration of forest fire urban spread assessment results with real-time fire environment parameters and calculating the influence coefficient of the fire environment on fire spread. Its core is to establish a semantic association between the assessment results and fire environment parameters, quantifying the impact of dynamic changes in the fire environment on the fire. In the algorithm implementation, the algorithm first receives assessment results such as risk level and potential spread speed output from the forest fire urban spread assessment module, while simultaneously collecting real-time fire environment parameters (wind speed 0-25 m / s, accuracy ±0.1 m / s; air density 1.1-1.3 kg / m³, accuracy ±0.01 kg / m³, etc.). The first step matches the semantic features of the assessment results with the fire environment parameters (e.g., high risk level corresponds to wind speed >5 m / s and temperature >25℃). The second step introduces a time decay factor (decaying by 10% every 10 minutes) to calculate the weight of each parameter. The third step generates a fire environment stability index (0-1), a fire spread acceleration probability (accuracy ±3%), and a building impact level through weighted summation and semantic mapping (converting fire behavior and environmental parameters into a unified semantic dimension 0-1 value). The algorithm aims to establish a connection between the assessment results and the real-time fire environment, overcoming the problem of the assessment being disconnected from the fire environment in traditional technologies. It improves the ability to capture dynamic changes in the fire environment, enabling subsequent grid division and simulation calculations to accurately adapt to the real-time fire environment. This ensures that the system responds to fire spread more promptly and realistically, providing on-site fire environment parameters to support dynamic simulation.

[0053] The adaptive grid fire propagation model of this invention is a core model that adjusts the grid division of the forest-urban boundary area according to the dynamic changes of the fire environment and supports fire propagation simulation. Essentially, it enables grid cells to accurately match high-risk fire areas through dynamic adaptation of grid density and shape, thereby improving the targeting of monitoring and simulation. The model implementation requires first determining the initial grid size (20m × 20m, covering the boundary area and extending outward by 1km). Then, it receives parameters such as the fire environment stability index and the probability of spread acceleration from the semantically coupled fire environment response algorithm. When the stability index is <0.3 and the acceleration probability is >60%, the corresponding area grid is densified to 5m × 5m (extending outward by 3 initial grids). When the stability index is >0.7 and the acceleration probability is <30%, it is merged into a 40m × 40m grid (containing at least 4 initial grids). At the same time, the grid shape is adjusted by combining terrain data (45° acute-angled rhomboid grid for slope >25°, with contour lines fitted ≥85%) and building distribution (grid boundaries in areas with building density >30% are aligned with building exterior walls, with a deviation of ±1m). Finally, the grid attributes (number, vegetation type, mean fire environment parameters, etc.) are labeled and the quality is verified (aspect ratio of 1:1 to 1:3, interior angle of 60° to 120°, etc.). The model aims to provide precise spatial units for dual-domain fire monitoring, avoiding blind spots or resource waste caused by fixed grids; improve the adaptability of the grid to the fire environment, enabling subsequent fire monitoring to focus on high-risk areas; and allow simulation calculations to be carried out based on reasonable grid units, thereby significantly improving the system's ability to accurately depict the fire propagation process.

[0054] The dual-domain fire state semantic monitoring platform of this invention is a core platform for monitoring fire state parameters in forest and urban domains separately, establishing cross-domain associations, and providing real-time data feedback. It enables accurate acquisition, semantic encoding, and collaborative feedback of dual-domain fire state data, providing dynamically updated fire state support for fire spread simulation. The platform implementation requires first determining the monitoring priorities based on the adaptive grid division results (high-priority grids are collected once every 30 seconds, medium every 1 minute, and low every 2 minutes). In the forest domain, monitoring points are equipped with temperature sensors (0-1200℃, accuracy ±5℃), smoke sensors, etc., to collect parameters such as flame temperature and smoke concentration and encode them according to rules (e.g., 800-1200℃ is "high"). In the urban domain, monitoring points are equipped with infrared thermometers (0-800℃, accuracy ±3℃), video equipment, etc., to collect building temperature, combustion status, etc., using the same encoding rules. Subsequently, the correlation strength of parameters between adjacent dual-domain monitoring points was calculated using the Pearson correlation coefficient (absolute value > 0.6 indicates strong correlation). The cross-domain propagation time delay (accuracy ± 1 second) and attenuation coefficient were analyzed to generate a cross-domain correlation matrix. Finally, the data was integrated in the format of "timestamp-grid number-dual-domain parameters-correlation matrix" and transmitted in real-time (latency < 1 second) to the simulation output module via a 5G network. The platform's function is to provide the simulation output module with real-time, correlated dual-domain fire status data, overcoming the problem of data ambiguity in traditional monitoring. It improves the real-time performance and correlation of fire status data, allowing simulation calculations to be dynamically adjusted based on the latest fire status, ensuring a high degree of consistency between simulation results and the actual fire spread process, and providing accurate dynamic data support for fire prevention and control decisions.

[0055] like Figure 2As shown, a forest-township fire spread simulation system based on spatiotemporal semantics is implemented. The system operates in the following steps: First, a multi-source spatiotemporal semantic data acquisition module calls upon various types of sensors distributed within the forest-township boundary to collect data on vegetation type distribution, topographic elevation, real-time meteorological monitoring, urban building density, road network distribution, and historical fire case data. The collected multi-source data undergoes format conversion and data cleaning to remove outliers and missing values, generating a standardized multi-source dataset. Second, the standardized multi-source dataset is transmitted to the forest fire-township spread assessment calculation module, where a multi-source spatiotemporal semantic forest fire-township spread assessment model is used to evaluate the standardized dataset. The first step involves extracting semantic features from various data sources in a multi-source dataset, establishing semantic relationships between data, calculating the impact weights of different data types on fire spread, and outputting the assessment results of forest fire urban spread. The second step transmits these assessment results to the fire environment semantic coupling response module. Combining real-time collected wind speed, air density, ambient temperature, and relative humidity parameters, the semantic coupling fire environment response algorithm couples the assessment results with real-time fire environment parameters, analyzes the impact of changes in fire environment parameters on fire spread, and generates fire environment coupling response parameters. The third step transmits these parameters to the adaptive grid partitioning processing module, and based on the adaptive grid fire propagation model, calculates the fire spread based on the fire spread. The spatial distribution differences of environmental coupling response parameters are analyzed. Dynamic grid division is performed on the forest-urban boundary area, adjusting the density and size of grid units in different regions to generate a dynamic grid dataset including grid spatial coordinates, grid attribute information, and fire environment parameter correlation results. The fifth step involves transmitting the dynamic grid dataset to a dual-domain fire state semantic monitoring module. Real-time collection and semantic encoding of fire state parameters corresponding to the forest and urban areas for each grid unit in the dynamic grid dataset are performed. A spatiotemporal correlation mapping between fire state parameters and grid units is established, and the propagation rate and attenuation degree of fire state parameters between different grid units are calculated. This generates a dual-domain fire state monitoring dataset including real-time fire state data, spatiotemporal correlation mapping results, and propagation characteristic parameters. The sixth step involves receiving the dual-domain fire monitoring dataset and the dynamic grid dataset. Combining the calculation logic of the multi-source spatiotemporal semantic forest fire town spread assessment model and the adaptive grid fire propagation model, the module dynamically calculates the spread path, burning intensity, and duration of the fire in different grid units at the forest-town boundary. This generates a simulation result dataset that includes the fire spread status, spatial distribution characteristics, and calibration parameter change trends under the time series. Simultaneously, the simulation result dataset is processed using a preset data encapsulation format to form a standardized simulation result file that can be read and displayed by external devices.

[0056] The forest-town boundary fire spread simulation system, driven by spatiotemporal semantics, comprehensively integrates various types of data, including vegetation, topography, meteorology, and buildings, through a multi-source spatiotemporal semantic data acquisition module. Combined with a forest fire-town spread assessment and calculation module, it delves into the semantic features of the data and establishes correlations, significantly improving the efficiency of comprehensive utilization of multi-source spatiotemporal data. Simultaneously, it enhances the coverage of various fire-influencing factors in the assessment results, ensuring that the results fully reflect the combined effects of vegetation distribution, topographic features, meteorological conditions, and urban building attributes on fire spread. This avoids the biased assessments caused by traditional single-data independent analysis, providing accurate and comprehensive basic data support for subsequent fire spread simulations.

[0057] Another key advantage of the system lies in its deep integration of dynamic simulation and dual-domain collaborative monitoring, which effectively addresses the shortcomings of traditional simulations, such as fixed grids and disconnected dual-domain monitoring. The fire environment semantic coupling response module and the adaptive grid generation processing module work together to dynamically adjust the grid density and size based on real-time changes in the fire environment, significantly improving the adaptability of grid generation to dynamic changes in the fire environment. The dual-domain fire state semantic monitoring module establishes a cross-domain correlation monitoring mechanism for fire state parameters in the forest and urban domains, improving the correlation and completeness of fire state parameter monitoring and accurately capturing changes in the rate and intensity attenuation of fire propagation across domains. Finally, combined with the boundary domain fire spread simulation output module, dynamic simulation is achieved, improving the timeliness and accuracy of fire spread simulation as a whole, making the simulation results more closely match the actual fire spread process, providing more precise technical support for fire prevention and control decisions, and effectively making up for the shortcomings of traditional simulations that are disconnected from actual scenarios.

[0058] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0059] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A forest and town interface fire spread simulation system based on spatio-temporal semantic driving, characterized in that, The application relates to a forest-fire town spread evaluation and simulation system based on multi-source spatio-temporal semantic data. The system comprises a multi-source spatio-temporal semantic data acquisition module, a forest-fire town spread evaluation and calculation module, a fire environment semantic coupling response module, an adaptive grid division processing module, a dual-domain fire state semantic monitoring module and a boundary-domain fire spread simulation output module. The multi-source spatio-temporal semantic data acquisition module collects vegetation type distribution data, terrain elevation data, real-time meteorological monitoring data, town building density data, traffic network distribution data and historical fire case data in a forest-town boundary domain, and transmits the collected data to the forest-fire town spread evaluation and calculation module. The forest-fire town spread evaluation and calculation module extracts and analyzes the semantic features of the received data based on a multi-source spatio-temporal semantic forest-fire town spread evaluation model, and outputs the evaluation results to the fire environment semantic coupling response module. The fire environment semantic coupling response module couples the evaluation results and real-time fire environment parameters by using a semantic coupling fire environment response algorithm, generates fire environment response parameters and transmits the fire environment response parameters to the adaptive grid division processing module. The adaptive grid division processing module dynamically divides the boundary domain space according to the fire environment response parameters by using an adaptive grid fire spread model, and outputs the grid division results to the dual-domain fire state semantic monitoring module. The dual-domain fire state semantic monitoring module performs real-time semantic monitoring and data feedback on the fire state parameters of the forest domain and the town domain in the grid division area, and cooperatively transmits the feedback data and the grid division results to the boundary-domain fire spread simulation output module. The boundary-domain fire spread simulation output module dynamically simulates the forest-town boundary domain fire spread process and outputs the simulation results by combining the feedback data and the grid division results. The multi-source spatio-temporal semantic forest fire town spread evaluation model expression in the forest fire town spread evaluation calculation module is: ; wherein, is a forest fire town spread evaluation value, is a weight coefficient of the first class multi-source data, is a semantic feature extraction value of the first class data, is the first class multi-source spatio-temporal data, is a correlation parameter class number, is a correlation coefficient of the first class data and the first fire spread influence parameter, is the first fire spread influence parameter, is a time attenuation coefficient, is a collection time of the first class data, is a current simulation time; the fire environment semantic coupling response module adopts a semantic coupling fire environment response algorithm to perform coupling operation on the evaluation result and real-time fire environment parameters, and the expression is: ; wherein, is a fire environment coupling response value, Couple( ) is a coupling operation function, is a real-time wind speed parameter, is an air density parameter, is an environment temperature parameter, is a relative humidity parameter, is a fire environment factor class number, is a semantic mapping function, is the first fire behavior parameter, is the first environment parameter. 2.The forest and town interface fire spread simulation system based on spatio-temporal semantic driving according to claim 1, wherein, The semantically coupled fire environment response algorithm expression in the fire environment semantic coupling response module is as follows: ;in, For the semantic coupling coefficient of the fire environment, To evaluate the semantic transformation value of the result, For gradient operators, For time variables, This is the coupling adjustment coefficient. The number of environmental monitoring points, For the first Temperature parameters at each monitoring point For the first The vegetation moisture content parameter at each monitoring point; the expression for the adaptive grid partitioning processing module to dynamically partition the boundary space based on the fire environment response parameters using the adaptive grid fire propagation model is as follows: ;in, To adjust the coefficients for adaptive meshing, For the initial grid coordinate function, They are x and y in a spatial rectangular coordinate system, respectively. Axis coordinates This represents the maximum value of the fire environment coupled response. For grid adjustment coefficients, Let be the partial derivative of the fire environment coupled response value in spatial coordinates. For the number of grid cells, For the first The area and distance weighting function of each grid cell. For the first The area of ​​each grid cell, For the first The distance from each grid cell to the fire point. 3.The forest and town interface fire spread simulation system based on spatio-temporal semantic driving according to claim 2, characterized in that, The adaptive grid fire propagation model expression in the adaptive grid partitioning processing module is as follows: ;in, This represents the probability value of fire propagation. For divergence operators, This is the semantic quantization value of the fire environment coupled response value. For the fire propagation velocity vector, For the propagation regulation coefficient, Divide the grid area into several parts. For the first Fuel load function for each grid region For the first Fuel type parameters for each grid region For the first Wind speed influence function for each grid region For the first The wind speed parameters for each grid region; the expression for semantic feature extraction and correlation analysis of the received data by the forest fire town spread assessment calculation module based on the multi-source spatiotemporal semantic forest fire town spread assessment model is as follows: ； wherein, is a forest fire town spread analysis value, is a semantic feature extraction function, is vegetation type distribution data, is terrain elevation data, is weather real-time monitoring data, is a correlation analysis function, is town building density data, is traffic road network distribution data, is historical fire case data, is a historical data influence coefficient, is a maximum value of historical fire case data. 4.The forest and town interface fire spread simulation system based on spatio-temporal semantic driving according to claim 2, characterized in that, The fire state semantic monitoring model expression used by the double-domain fire state semantic monitoring module in the process of real-time semantic monitoring and data feedback of the forest domain and the town domain fire state parameters in the grid division area is: ; wherein, is a fire state semantic monitoring value, is a semantic monitoring function, is a flame temperature parameter, is a smoke concentration parameter, is a thermal radiation intensity parameter, is a self-adaptive grid adjusted grid area, TotalArea is a total monitoring area of the boundary domain, is a number of monitoring sub-regions, is a domain type weight function of the th sub-region, is a forest domain proportion of the th sub-region, is a town domain proportion of the th sub-region; the expression of the boundary domain fire spread simulation output module combining the feedback data and the grid division result to realize dynamic simulation of the forest and town boundary domain fire spread process is: , wherein, is a fire spread simulation result value, is a dynamic simulation function, is a time step function, is a simulation time step, is a number of simulation parameter types, is a parameter weight function, is the th forest domain simulation parameter, is the th town domain simulation parameter. 5.The forest and town interface fire spread simulation system based on spatio-temporal semantic driving according to claim 2, wherein, When the forest fire town spread assessment calculation module performs semantic feature extraction and association analysis on the received data based on the multi-source spatiotemporal semantic forest fire town spread assessment model, the semantic feature association model expression used is as follows: ,in, For semantic feature relevance, For the first The semantic value of the fire spread affects the parameter. For the first Class data and the first The semantic distance of each parameter. The distance attenuation coefficient is used; when the adaptive mesh generation processing module dynamically meshes the boundary domain space according to the fire environment response parameters using the adaptive mesh fire propagation model, the mesh density adjustment model expression used is: ;in, This represents the grid density value. This represents the minimum value of the fire environment coupled response. Vegetation is the terrain influence function. This is the vegetation influence function. 6.The forest and town interface fire spread simulation system based on spatio-temporal semantic driving according to claim 1, wherein, The adaptive grid division processing module comprises a grid initial generation unit, a fire environment parameter mapping unit, a dynamic grid adjustment unit and a grid quality checking unit. The grid initial generation unit performs initial grid division on the boundary domain space according to the spatial coordinate range of the forest-town boundary domain, terrain elevation data and town building distribution boundaries, generates an initial grid data set comprising a grid number, a grid coordinate, a vegetation type proportion in the grid and a building density in the grid, and divides the boundary domain space into grids with a preset grid precision. The fire environment parameter mapping unit receives the fire environment response parameters transmitted by the fire environment semantic coupling response module, associates and maps the fire environment response parameters according to the grid number and the initial grid data set, calculates the average value, the maximum value and the change rate of the fire environment response parameters corresponding to each initial grid unit, and generates a grid-fire environment parameter mapping table. The dynamic grid adjustment unit adjusts the grid division angle by combining the terrain slope parameter based on the adaptive grid fire spread model, encrypts and divides the grid units with a fire environment response parameter change rate greater than a preset threshold, and combines and processes the grid units with a fire environment response parameter change rate less than the preset threshold, so that the fitting degree of the grid boundary and the terrain contour is ensured. The grid quality checking unit checks the quality of the grid units after dynamic adjustment, calculates the aspect ratio, internal angle and area deviation value of each grid unit, removes the grid units whose aspect ratio exceeds a preset range, internal angle exceeds a reasonable interval or area deviation value is greater than an allowed threshold, re-divides the removed area into grids, and outputs the final grid division result after all grid units meet the quality requirements. 7.The forest and town interface fire spread simulation system based on spatio-temporal semantic driving according to claim 1, wherein, The dual-domain fire state semantic monitoring module includes a forest domain fire state monitoring unit, a town domain fire state monitoring unit, a cross-domain fire state correlation unit, and a fire state data feedback unit. The forest domain fire state monitoring unit selects grid units with a forest domain proportion greater than a preset proportion as monitoring points in the grid division area output by the adaptive grid division processing module, and real-time collects flame temperature, smoke concentration, thermal radiation intensity and vegetation burning rate parameters of each monitoring point, and converts the collected parameters into standardized semantic data through semantic coding. The town domain fire state monitoring unit selects grid units with a town domain proportion greater than a preset proportion as monitoring points in the grid division area, and real-time collects building surface temperature, building burning state, flammable material quantity and road network passing state parameters of each monitoring point, and converts the collected parameters into standardized semantic data using the same semantic coding rules as the forest domain fire state monitoring unit. The cross-domain fire state correlation unit analyzes the standardized semantic data of the forest domain and town domain monitoring points, identifies the fire state parameter transmission relationship between adjacent forest domain monitoring points and town domain monitoring points, calculates the time delay and intensity attenuation coefficient of cross-domain fire state parameter propagation, and generates a cross-domain fire state correlation matrix. The fire state data feedback unit integrates the forest domain fire state data, the town domain fire state data and the cross-domain fire state correlation matrix according to a preset data format, adds data collection time stamps and grid number information, and transmits them in real time to the interface domain fire spread simulation output module. 8.The forest and town interface fire spread simulation system based on spatio-temporal semantic driving according to claim 1, wherein, The interface domain fire spread simulation output module includes a simulation parameter integration unit, a dynamic spread calculation unit, a simulation result rendering unit, and a simulation data output unit. The simulation parameter integration unit receives the feedback data transmitted by the dual-domain fire state semantic monitoring module and the grid division result output by the adaptive grid division processing module, matches and correlates the fire state parameters in the feedback data with the grid attributes in the grid division result, extracts the real-time fire state parameters, initial environmental parameters and spatial attribute parameters of each grid unit, and constructs a simulation calculation parameter library. The dynamic spread calculation unit calculates the fire spread direction, spread speed and burning state change of each grid unit at each time node according to the preset time step based on the multi-source spatio-temporal semantic forest-town spread evaluation model and the adaptive grid fire spread model, and generates a dynamic spread calculation data set. The simulation result rendering unit visualizes and renders the fire spread state at each time node using three-dimensional rendering technology according to the dynamic spread calculation data set, presents the dynamic effects of flame shape, smoke diffusion, vegetation burning and building damage, and labels the fire state parameter values of each grid unit. The simulation data output unit encapsulates the dynamic spread calculation dataset and the visual rendering result according to a preset format, generates a simulation result package including time sequence data, spatial distribution data and a visual file, and supports reading and display of the simulation result package by an external device.

Images